1 ELECTROSPUN CERAMIC FIBERMATS FOR FILTRATION APPLICATIONS IN LUNAR MISSIONS FOR IN HABITAT APPLICATIONS By APRATIM BISWAS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012
2 2012 A pratim Biswas
3 To my parents
4 ACKNOWLEDGMENTS I would like to acknowledge all the support and advice that I have received from my advisor Dr. Wol fgang M. Sigmund. I would also like to acknowledge Florida Space Research Initiative for funding the research. I am also thankful to the Dr. Chang Yu Wu and his research group for collaborating with our research group and the Particle Engineering Research Center, UF and Major Analytical Instrumentation Center, UF for allowing me to use their facilities.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 7 LIST OF FIGURES .......................................................................................................... 8 LIST OF ABBREVIATIONS ........................................................................................... 10 ABSTRACT ................................................................................................................... 12 CHAPTER 1 BACKGROUND ...................................................................................................... 15 1.1 Motivation ......................................................................................................... 15 1.2 Goal, Objectives and the C hallenges General Outline ................................... 16 1.3 Filtration ............................................................................................................ 18 1.3.1 Filtration Technologies Selection ........................................................... 18 1.3.2 Filtration Theories for Fibrous Filters ....................................................... 19 1.4 Fiber Processing ............................................................................................... 23 1.4.1 Overview of V arious Fiber Processing Techniques ................................. 23 1.4.2 Electrospinning ........................................................................................ 24 1.5 Materials Selection ............................................................................................ 26 1.6 Photocatalysis ................................................................................................... 28 1.6.1 Photocatalytic Properties of Titania ......................................................... 28 1.6.2 Effect of Dopants on Photocatalyti c Activity of Titania ............................. 30 2 EXPERIMENTAL METHODS ................................................................................. 42 2.1 Preparation of Aqueous TiO2SiO2 Hybrid Sol ................................................. 42 2.2 Electrospinning of Ceramic Fibers .................................................................... 43 2.2.1 Modification of Fiber D iameter ................................................................. 44 2.2.2 Hea t Treatment ....................................................................................... 46 2.3 Preparation and Electrospinning of Doped Titania Sol ...................................... 47 2.3.1 Preparation of Titania Sol ........................................................................ 47 2.3.2 Addition and Electrospinning of Niobium Doped Titania Sol .................... 47 2.4 Determination of Filtration Properties ................................................................ 49 2.5 Determination of Photocatalytic Properties ....................................................... 50 3 RE SULTS AND DISCUSSIONS ............................................................................. 59 3.1 Effect of Ele ctrospinning Conditions on Fiber Morphology, Crystal Structure and Mechanical Properties of TiO2SiO2 Fibers .................................................. 59 3.2 Filtration Properties of Electrospun Ceramic Fibermat ...................................... 65
6 page 3.3 Influence of Niobium Dopant on the Morphology, Crystal Structure and Photocatalytic Activity of Electrospun Titania Fiber ............................................. 69 4 CONCLUSIO NS AND FUTURE WORK ............................................................... 109 LIST OF REFERENCES ............................................................................................. 112 BIOGRAPHICAL SKETCH .......................................................................................... 119
7 LIST OF TABLES page 2 1 Addition of niobium ethoxide to t itan ia s ol (9 ml) ............................................... 53 2 2 Addition of Nd(NO3)3.6H2O to titania s ol (9 ml) .................................................. 54 3 1 TiO2SiO2 aqueous sol compositions and electrospinning conditions used throughout the research for TiO2SiO2 s ol [E = 1 kV/cm, Q = 0.6 ml/h, collector dis tance = 25 cm] ................................................................................. 77 3 2 Fiber diameter of electrospun TiO2SiO2 fibers after heat treatment in presence of air in a box furnace ......................................................................... 78 3 3 Loss of fiber diameter on heat tr eatment at 600C in air for 3 hours .................. 79 3 4 Influence of addition of ethanol on fiber diameter of heat treated electrospun TiO2SiO2 fibers (2% w/v PVP, MW 1.3x106 used as polymeric aid) ................... 79 3 5 Modifying fiber diameter of heat treated fiber by varying different electrospinning parameters ................................................................................ 80 3 6 Fiber morphology for e lectrospun fibers of pure titania and niobium doped titania after heat treating at 600C for 3 hours. 4% w/v PVP was added. Electric field strength and flow rate were maintained at 1 kV/cm and 0.6 ml/h. .. 81 3 7 Specific surface of niobium doped titania fibers after heat treatment ................. 82
8 LIST OF FIGURES page 1 1 Various filtration mech anisms in action in collecting aerosoli zed particles by a single fiber .......................................................................................................... 34 1 2 Calculated fractional collection efficiency for particulates of different sizes ........ 35 1 3 Schematic diagram of conventional fiber processing using wet/dry/melt spinning technology ........................................................................................ 36 1 4 Schematic diagram of electrospinning set up. .................................................... 37 1 5 Ashby chart for wear rate constant vs hardness for various material classes ........................................................................................................... 38 1 6 Schematics of the photocatalytic reaction process. ............................................ 39 1 7 Effect of doping transition metals on electronic band structure in TiO2 .............. 39 1 8 Influence of atomic number on the effect o f doping transition m etals on the electron band structure of TiO2 .......................................................................... 40 2 1 Processing of TiO2SiO2 aqueous sol ................................................................ 55 2 2 The electrospinning set up used in electrospinning TiO2SiO2. A vertical set up was used throughout the research. ............................................................... 56 2 3 Heat treatment of TiO2SiO2 fibermat ................................................................. 57 2 4 The filtration testing set up used for determining the filtration properties of electrospun TiO2SiO2 fibermats. ....................................................................... 58 2 5 Chamber used to expose sample to ultra vi olet or visible light radiation ............ 58 3 1 Influence of PVP content on the morphology of electrospun TiO2SiO2 fibers ... 83 3 2 TiO2SiO2 fibermats, electrospun from sol containing 0.4% w/v PVP, after heat treatment at 600 C for 3h. .......................................................................... 85 3 3 Flexibility of heat treated TiO2SiO2 fibers .......................................................... 86 3 4 Morphology and c rystallinity of heat treated TiO2SiO2 nanofibers .................... 88 3 5 Chemistry behind synthesis of aqueous hybrid sol of TiO2 -SiO2 ........................ 90 3 6 Electrospun micron sized TiO2SiO2 fibers ......................................................... 91
9 page 3 7 Pore size distribution of electrospun TiO2SiO2 fib ermat obtained using BJH method ............................................................................................................... 92 3 8 Original particle size distribution of aerosol ge nerated in filtration testing set up ....................................................................................................................... 93 3 9 Experimentally determined pressure dr op of electrospun TiO2SiO2 fibermats pl otted vs average fiber diameter ....................................................................... 94 3 10 Plot of mass/ ( square of thickness ) vs average fiber diameter ............................ 96 3 11 Collection efficiency of ceramic fibermats plotted versus average fiber diameter ............................................................................................................. 97 3 12 The effect of using acti vated carbon fiber as collector for electrospun fibers ..... 98 3 13 Dependence of fiber diameter after heat treatm ent on the niobium dopant level .................................................................................................................. 102 3 14 X ray diffraction plots demonstrating the impact of niobium dopant on anatase to rutile phase transformation ............................................................. 103 3 15 Effect of niobium dopant on diameter and specific surface of electrospun TiO2 fibers. ....................................................................................................... 105 3 16 BJH pore size analysis of niobium doped titania fibers. .................................... 106 3 17 I nfluence of niobium dopant on UV/Vis absor ption spectra of titania fibers ...... 107 3 18 Differential reflectance of niobium doped titania ............................................... 108
10 LIST OF ABBREVIATION S Pressure drop across a filter P Penetration factor for a filter Collection efficiency of a filter Q Quality factor of a filter. Also used to denote flow rate in equations des cribing electrospinning process Porosity of a filter ER Collection efficien cy due to interception dp Diameter of particulates in aerosol df Diameter of fiber Ku Kuwabara hydrodynamic factor Et Collection efficiency from entrapment of particles due to inertial impaction p Particle density CC Cunningham factor Viscosity of gas V Face velocity of air stream ED Collection efficiency due to diffusion of small particles D Diffusion coefficient ht Terminal jet diameter Dielectric constant of air Surface tension of a sol or any liquid I Current Radius of cur vature of the jet at the regions of whipping instability in electrospinning process Quantum yield
11 kCT Charge transfer rate kR Recombination rate of holes and electrons A Absorbance. Loss in absorbance Absorptivity C0 Initial dye concentration C Conc entration of dye at any particular time C0 C ; i.e. loss is concentr ation of dye due to degradation Density of any material. For the purpose of this dissertation it stands for the density of the material of which the fiber s are made Eq Equation
12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ELECTROSPUN CERAMIC FIBERMATS FOR FILTRATION APPLICATIONS IN LU NAR MISSIONS FOR INHABITAT APPLICATIONS By Apratim Biswas May 2012 Chair: Wolfgang Michael Sigmund Major: Materials Science and Engineering In the absence of atmosphere and hydrosphere, there are few collision s between dust particles in the lunar envir onment Further particles become charged in presence of cosmic rays and similarly charged particles repel each other. Hence particles retain sharp edges and often have high aspect ratios. When exposed to lunar dust humans show symptoms similar to hay fev er. Such part icles are also damaging to equipment Humans and robots, used in operations can bring such dust particles inside the human habitat making them airborne. High efficiency particulate air (HEPA) filters provide an effective way to trap such part icles. But due to environment conditions, polymer based filters are susceptible to mechanical erosion. The presence of high energy radiation, due to the absence of atmosphere and magnetic fields is also damaging to polymers. Ceramic materials are resistant t o abrasion and radiation and hence were chosen as the preferred class of materials for the filtration media. Among all the ceramics, TiO2 was selected for its photocatalytic activity which may play a key role in energy efficient survival in space or lun ar stations. Such fibers are multi functional with the advantage of self cleaning property in presence of radiation. However ceramic fibers including TiO2 fibers, have a significant disadvantage of their own. They are brittle and were
13 considered too prone to failure to be successfully used as a filtration media when they reach nanometer dimensions. This dissertation describes the advances in fabrication and understanding of fundamentals in overcoming these challenges. In absence of crack initiation sites, amorphous ceramic fibers have near theoretical strength and strain to failure. Amorphous TiO2SiO2 fibermats, with lower flaw populations and exceptional surface quality, have been developed. They can be rolled to a radius of curvature of 3.4 mm exhibiti ng flexibility. The fibermats are also mechanically robust and can withstand the stress associated with general handling and fixture used for holding the filtration media. E lectrospinning was selected as the fabrication method due to superior performance t owards fiber diameter uniformity and the ability to decrease fiber diameters to the nm level. Filtration tests have been carried out on such fibermats concerning a number of key variables such as fiber diameter, particle size, pressure drop and more. Multi functionality as filter material and as photocatalyst allow s the filters to be regenerable. Furthermore, organic vapors (odors) and plant super hormones (ethylene gas) can be oxidized. This is key for a sustainable human base where food needs to be grow n and the level of odors in habitat has to be minimized. Ceramic materials based on TiO2 and titania composites where selected. To enhance the catalytic properties doping with a pentavalent ion, viz niobium with varying concentrations was done. Materia ls were electrospun and characterized. An increase of niobium yields stabilization of the anatase phase at 600C as evident from XRD patterns. Higher treatment temperatures allow a transformation to rutile. This is important since the semiconductor junctio n of anatase to rutile decreases electronhole recombination rate, which enhances the photocatalytic activity Furthermore doping
14 anatase with niobium increases the porosity and with it the catalytically active area. In fact the specific surface area of t itania fibers increases by almost 6 times when doped with only 2.5 at% niobium. However, in this work reduced photocatalytic activity was observed. It is hypothesized that phase separation of the niobium rather than doping in a solid solution occurred whi ch will change the properties of the semiconductor junction in an unfavorable way. The other possible explanation is the decrease in the anatase rutile semiconductor junction in niobium doped titania.
15 CHAPTER 1 BACKGROUND 1.1 Motivation The core motivat ion behind the research was to develop filtration media suitable for application i n lunar environment especially inhabitat There are several factors which make lu nar environment unique. I n the absence of atmosphere there is very little dust particle dus t particle collision The particles are also exposed to plasma and UV radiations and are charged and repel each other [1, 2]. This results in particles having sharp jagged edges and high axial ratios. In habitat, s u ch sharp hard particles can also damage the dermal water vapor barrier  in humans and are also irritating to eyes. The lunar dust particles can also be damaging to the lungs. Hyper reactivity to lunar dust has been observed before by astronauts during the Apollo missions . Such particles are too abrasive for polymer based fibro us filters. The high energy radiations are also damaging to the polymer. On the other hand c eramics are not only hard and abrasion resistant but can also withstand the irradiative environment. In filtration theory of fibrous filters, a reduction in fiber diameter is related to a decrease in pressure drop due to decrease in drag force and an increase in filtration quality factor. When used i n habitat, microorganisms can be trapped in such filters. Photocatalysis can be used as a mechanism to degrade any such deposition and in turn help in repeated use of the filtermat. This served as the motivation behind the search for ceramic fibers with high photocatalytic act ivity. Niobium was added as dopant to titania before fibers were electrospun to study for their impact on the photocatalytic properties of titania fibers. If photocatalytic enough, such fibrous filters would self clean in sunlight, especially considering t he fact that in the absence atmosphere, the lunar environment is highly
16 irradiative. Thus a potentially harmful fact can be utilized in such a way that it yields beneficial result. While the core motivation in the research has been in space applications, such fibermats have enormous potential as high efficiency particulate air (HEPA) filter in terrestrial applications too. The Department of Energy defines HEPA filters as filters which have a minimum efficiency of 99.97% for aerosolized 300 nm particles. Such filters have been proven as an effective protection against contamination of SARS virus through inhalation and preventing aspergillosis in immunocompromised individuals with hematologic malignancies. Development of HEPA filters is also a national security priority as a protection against biological warfare and bioterrorism agents such as anthrax. In presence of ultraviolet radiation, photocatalytic HEPA filters can destroy such virulent species and maybe used multiple times in such applications. 1.2 Goal Objectives and the Challenges General Outline This section of the dissertation aims at clearly identifying the long term goal, in termediate objectives towards it and the challenges which must be surmounted to attain the desired objectives. And along the way, the current state of the art has been explained w ith an emphasis on the gap in knowledgebase that must be filled to attain the final goal. The dissertation is the resul t of the body of work done towards completion of a project where the main goal is t o prepare filtration media for use in lunar mission. With that as the bottom line, we have two prominent choices fibrous filters and membranous filters. The membranous filters have pores in the membrane and the fibrous filters are composed of several layers of fibers deposited on top of each other.
17 There are several f actors which have to be considered before a particular filtration media is selected Briefly these include collection efficiency and pressure drop. Porous fibers have higher collection efficiency and can entrap a larger fraction of particles but the fibro us filters have lower pressure drop and require lesser operating energy. In the absence of abundant energy source, the lesser operating energy requirement makes fibrous filter the winner between the two types. With the type of filtration media thus determ ined, we are faced with a new challenge the nature of lunar dust itself. E rosive process es like atmosphere or hydrosphere are absent [1, 2] and particles are similarly charged due to the presence of cosmic rays. A ll these factors ensure that there is little particle particle collision and hence the particles remain sharp with jagged edges too hard and abrasive for polymeric filter media. A hard abrasion resistant material is required and ceramic materials would be the perfect candidate for that. Heat treated polycrystalline ceramic fibers are notoriously brittle and are impossible to be fitted into the filtration set up. There is a lack of technological evolution towards developing mechanically robust and flexibl e fibermats which can be used in this application. This is an area which has been addressed several times throughout this dissertation with the eventual development of flexible fibermats The filtration properties of the fibermats, thus prepared, have been studied, reported and discussed. In an ideal scenario the fibermats can be recycled and used multiple times. This requires cleaning on a regular basis. Due to lack of abundant energy sources, minimizing the energy utilization is always important in space If a material is developed which exhibits photocatalytic activity in the visible light range then sunlight itself may be
18 used to cl ean and reuse the filters. Thus the final portion of the body of work is dev oted to increasing the utility of the filters and is included in the last part of the dissertation. 1.3 Filtration Keeping in view the fact that the motivation behind the research is developing a solution for filtration application, it is important to briefly explain the various options that are available and the logical reasoning behind choosing fibrous filters. It is also important to briefly explore the filtration theory since one of the objectives i s to try to study how fiber diameter, filter thickness, void ratio etc. influence the filtration pr operties like pressure drop and filtration quality factor. 1.3 .1 Filtration Technologies Selection The types of filters which are most important for sampling and filtering aerosolized particles are fibrous and porous filters. In fiber based filters, a mat of fine fibers is used as filtration media in a direction perpendicular to the flow of airflow and contrary to the popular misconception; such filters can trap aerosolized particle s much smaller than the separation between the fibers. In porous filters the porosity is about 5090% less compared to fibrous filters. The aerosolized particles are trapped when the air stream carrying them flows through the compl ex porous structure. Simila r to fibrous filters, porous filt ers too can trap particles smaller than the pore size. The most important factors to consider while selecting a filter are a) pressure drop, b) collection efficiency and c) the quality factor. In porous filter s the collection efficiency is high but that comes at the expense of high pressure drop. Having a higher pressure drop implies that a higher energ y is required for filtration and hence an increase in the operating expense. Taking all these factors into consideration, fibrous filter was selected to be most appropriate for the application.
19 1.3 .2 Filtration Theories for Fibrous Filters Fibrous filt ers are often used to capture aerosolized particles in air. Before we can understand the details of how a fibrous filter works and why it i s an excellent choice for the intended application a few concepts must be discussed. The first such term is collection efficiency. In the most abstract sense, collection efficiency of a single fiber may be defined as the ratio between the cross sectional area of the air stream from which particles are removed to the projected area in the direction of the flow. This can eventually be formulated as the fraction of entering particles that are retained by the fiber. In filtration theory a term called penetration factor is often used which includes the collection efficiency as a factor. The penetration factor reflects the fraction of entering partic les that exit the filter. The penetration factor is often denoted by P. In terms of the collection efficiency P can be expressed by Eq. 1 3. The second important parameter in the theory of fibrous filters is the pressure drop, The pressure drop is related to the energy required for the air stream to flow through the filter membrane, the higher the energy require d the higher is the pressure drop. (1 1) (1 2) (1 3)
20 An ideal filter should have 100% collection efficiency and a very low pressure drop. Quality factor is the parameter which reconciles both these parameters. Filtration quality factor, Q is expressed by Eq. 14 In filtration theory, instead of considering collection efficiency and pressure drop as separate parameters, the quality factor gives us a single number which can be used to judge a filter. Higher the quality factor is for a fibermat, the better it is as a filter. A large portion of the conducted research is devoted to the understanding of how fiber diameter, filter thickness, porosity etc. influence the filtration propert ies. In light of this it is important to look into the theory of filtration as it relates to a single fiber. Part icles are captured by any or a combination of the three methods depending on their size: a) interception, b) inertial impactio n and c) diffusion. In the theory it is assumed that the particles are trapped onto the fiber d ue to Van der Waals force of attraction. Interception is the process that takes place when the particle reaches within a distance nearly equal to the radius of the particle. The collection efficiency ER, of the interception method is given by Eq. 15 w here, ER p and df are the diameter of particle and fiber respectively. Ku is known as Kuwabara hydrodynamic factor  and is expressed by Eq. 16 (1 4) (1 5) (1 6)
2 1 Sometimes the path of the particle is diverted from the direction of the airflow because of inertia. This is known as the inertial impaction. The efficiency of this type of particle entrapment is ex pressed as Et and given by Eq. 1 7. In the equation p is the particle density and Cc is known as the Cunningham Correction Factor arising from the gas slippage. U is the face velocity of the air and is the viscosity of the air. A different type of particle entrapment dominates when particles are much smaller and dp/df<0.4. In such cases, the particles m ove out of the streamline due to Brownian motion in irregular path and are ultimately captured by diffusion. The collection efficiency of a single fiber due to diffusion ED, is given by Eq. 1 8 where D is the particle diffusion coefficient and U is the face velocity as mentioned previously. T he collection efficiency is proportional to df 2/3, df 3/4 and df 2/3 respectively in the interception, inertial impaction and diffusion modes respectively. Thus from the perspective of all the three entrapment methods, decrease in fiber diameter is associated with an increase in collection efficiency. Another factor to consider is the pressure drop caused by the r esistance to airflow across the fiber. This re sistance is called the drag force and is the main c ontributor to (1 8) (1 7)
22 the pressure drop in the fibrous filters. It is related to fiber diameter by the Stokes law expressed in Eq. 19. w here is the face velocity of the gas as before and df is the fiber diameter. CC is the correction factor related to the contribution of slip flow on the drag force and is called the Cunningham factor [6, 14]. For high Knudsen number (Kn) in the range of 0.5 to 134, CC is expressed by Eq. 110. w here A is an experimentally determined cons tant. Since pressure drop decreases and collection efficiency increases with decrease in fiber diameter, theoretically, a reduction in fiber diameter should lead to an increase in quality factor. A separate section has been devoted in our research to study the relationship between various parameters in filtration. Figure 1 1 illustrates the various filtration mechanisms of capture of aerosolized particles in air as has been explained in this section. There is another concept in the filtration theory which has been used while designing the experimental methods. The col lection efficiency of a fibrous filter is dependent on the size of particulates in the aerosol. D ifferent filtration mechanisms are active for parti culates in different size range and it has been experimentally found that the most penetrating particle size (MPPS) is usually in the range of 100 nm 500 nm[15, 16], with most experiments for filtration quality factor recommended at particle size of 300 nm[7, 16]. Figure 12 illustrates the concept of MPPS. As evident from the diagram, the (1 9) (1 10)
23 MPPS also depen ds on the fiber diameter and generally decreases with decrease in fiber diameter. 1. 4 Fiber Processing While the main application for ceramic fibers in the course of our research has been filtration, synthesis of fibers and especially nanofibers are of great importance in several other areas such as tissue engineering  and catalyst support[20, 21]. In the following sections the diff erent traditi onal ways of fiber processing like dry spinning, wet spinning, melt spinning and newer nontraditional alternatives like electrospinning[22, 23] are discussed F inally electr ospinning has been explored as the preferred option for the specific application of interest 1.4 .1 Overview of V arious Fiber Processing Tech niques The specific traditional technique used to process fibers is based on the solubility of the polymer in solvents and on whether they melt at higher temperatures. For the polymers which are soluble, it is diss olved in a solvent and spun. Aft er this the solvent can be removed either by evaporation or extraction. When the solvent is removed through evaporation it is known as dry spinning. It is known as wet spinning when the solvent is removed by extraction by either utilizing or not utilizing a chemical reaction. Diffusion plays a very important role in both the systems. A different method known as melt spinning can be used f or polymers which ar e not soluble but can be melted As the nomenclature suggests, the polymer is at first melted and the n spun into fibers. Figure 1 3 shows the generic schematic representation of the fiber formation process. As illustrated in the figure, the solution or the melt (depending on the specific process) is fed into the system either by an extruder or by compress ed gas (usually nitrogen). After being passed through a filter, the solution or the melt is fed into a spinneret which is a
24 metallic plate with enough strength to withstand the pressure gradient and having several capillaries. The extruded solution or melt is finally treated depending on the nature of the material and the exact method to form the final fibers. The above generic description of traditional technology has been intentionally over simplified. In real application such technologies are really complicated and involve expens ive equipment having very high precision. At present there are several newer alternative techniques including electros tatic spinning , spray spinning [22, 25] and centrifugal spinning [22, 26, 27] which do not require spinneret and fiber transport rollers Among these processes, electrostatic spinning, more commonly known as electrospinning, has been adopted for our purpos e as it is the most simple, very versatile and provides ample opportunity to vary fiber diameter from few tens of nanometers to several microns [24, 28, 29]. Using suitable polymer as aid, ceramic fibers can be easily electrospun from precursor sol. 1.4 .2 Electrospinning A sign ificant portion of the research has been devoted to fiber processing by electrospinning and it is important to present a concise overview of the electrospinning process. Electrospinning is the process of generating fibers using a high electrostatic field a nd has been known since the 1930s [23, 24, 30, 31]. A very high electrostatic field is applied between a syringe containing polymer or ceramic sol to be electrospun and a metallic collection target. This electric fi eld leads to charge stored in the material and finally fibers are drawn by it. The electrospinning process can be divided into three phases [24, 32]. In the first phase, a fluid jet is ejected from the sy ringe needle and is smoothly accelerated by the electric field. During this process, a cone is formed by the balance of several forces
25 acting on the liquid jet. Gravitational force and electric polarization stress tend to elongate the droplet, forming a cone known as the Taylor Cone. Opposing these two forces there are two other forces surface tension and normal electrical stress which try to form a droplet. The general set up for electrospinning is presented in Figure 14. The figure also illustrates the various forces which are in equilibrium to form the cone. Once the electrostatic force overcomes the surface tension, a fine charged jet is ejected from the Taylor cone. The second phase starts once the jet starts moving towards the collector. In this stage of electrospinning there are three types of instabilities which may occur i) axisymmetric Rayleigh instability due to surface tension leading to the formation of droplets, ii) axisymmetric instability induced by electric field and, iii) nonaxisymmetric instability, leading to bending of fibers[24, 3234]. The third form of instability is the most important from the perspective of reducing fiber diameter. Which instability will dominate during the process depends on the viscosity / conductivity ratio. With increase in the surface charge density and the electric field strength, the effect of classical Rayleigh instability is suppressed while enhancing the dominance of all forms of instabilities induced by electric field. Increase in the nonaxisymmetric instability and whipping mode enhances the ability of the process to reduce the fiber diameter. In the final stage of the electrospinning process, the repulsive force from the surface charge of the jet is balanced by the surface tension. Because these forces compensate each other, the stretching and thinning of the jet due to whipping instability ceases and the jet attains its final diameter. Mathematically the terminal diam eter can be obtained f rom the expression in Eq. 111
26 where ht surface tension, Q is the flow rate of the liquid, I is the current, us of curvature of the jet at regions of whipping instability (R) to the diameter of the jet (h) and is assumed to be approximately 100. 1.5 Materials Selection The materials selection process for this research has been briefly mentioned in the section on motivation. In this section we would first explore the initial material selection in details and then briefly mention the other considerations made to e nhance the photocatalytic activity with the objective to have a potential multifunctional final mat erial Harder a material is, the more resistant it is to abrasion. Since the material needs t o survive the erosive nature of the lunar dust, the primary consideration for the material is hardness. Ashby chart for hardness vs w ear rate constant as illustrated in Figure 1 5 shows the clear superiority of ceramics in this respect. Out of several available options in ceramics, TiO2 is particularly interes ting because of its photocatalytic activity . However similar to most ceramics, TiO2 is brittle While th e fracture toughness for metals are in the range from 10s 100s MPam1/2, for ceramics it ranges from 0.1 4 MPa m1/2 [40, 41]. Among the various forms of ceramics the fracture toughness of single crystals ranges from 0.3 to 2.0 MPam1/2 while polycrystalline ceramics have a higher toughness ranging from 24 MPa m1/2. In comparison amorphous ceramics have a much lower toughness, usually below 1 MPa(1 11)
27 m1/2. This is because while a crack has to propagate along the grain boundaries in polycrystalline ceramics and along a closepacked plane in singlecrystal ceramics, in amorphous ceramics, due to the absence of any preferred plane, crack propagates perpendicular to the surface. But near theoretical fracture toughness have been observed in micron sized SiO2 fibers [40, 41] when fi ber surfaces were coated with a very thin layer of epoxy to prevent crack initiation. Electrospun TiO2 nanofibers crystallize when heat treated at 500C however SiO2 has a much higher crystallization temperature of 1500C and remains amorphous at 500C. For this reason a hybrid oxide of TiO2 and SiO2 was selected as the pr eferred material with the hypothesis that an amorphous SiO2 matrix can improve the mechanical toughness Apart from the consideration of mechanical properties there was another important motivation behind selecting TiO2SiO2 and it has to do with the stability of the precursor sol itself. Ethanol based titania sol is not stable and within a few hours titania starts pr ecipitating in the system. This happens due to the high chemical reactivity of titanium alkoxides, which are commonly used as precursor s to titania, to the moisture present in air. In the aqueous sol prepared in our research, after undergoing hydrolysis and condensation, Si O Ti bonds are formed. This inhibits the crystallization of TiO2 and hence a stable sol is produced. The sol can be stored for several months. Details of the reaction would be provided in a later section. In th e final section of the research an attempt have been made to improve the photocatalyt ic activity of titania by using niobium as a dopant Briefly, the most important component ]/H2O2 and recombination of
28 electrons and holes. Since recombination of electron and holes inhibits the formation of ]/H2O2, reducing the rate of recombination is an effective technique to enhance the photocatalytic activity of a material. P entavalent (Nb5+) ions have been doped in an attempt to create trapping levels for holes to improve the photocatalytic activity. Because of the amount and nature of the background that has to be covered for this part of the research a separate section has been dedicated to it in section 1.6 1.6 Photocatalysis T he intended application for the material is for it to be used as a fibrous filtration media in space within human habitat. So an important criterion for the material is that it should be easy to clean. A pr ocess such as photocatalysis would be ideal for use to clean the fibermats using a light source of appropriate wavelength. If a fiber mat can be developed with a material which requires light having wa velength in the visible range it would be ideal for application even in space stations where it can self clean using sunlight. 1.6 .1 Ph otocatalytic Properties of T itania Titania is a semiconductor with wide bandgap of 3.2 eV for the anatase phase and 3.0 eV for the rutile form [43, 44]. Strictly speaking while all semiconductors have photocatalytic properties oxides and compound semiconductors exhibit better photocatalytic activity [45, 46]. There are two forms of photocatalytic reactions catalyzed photoreaction and sensitized photoreaction. In the catalyzed photoreaction the initial photoexcitation occurs in an absorbate molecule. This excited molecule then interacts wit h the catalyst substrate and transfers electron[46, 47]. On the other hand, in case of sensitized photoreaction, photoexcitation takes place first in the catalyst substrate and the substrate then interacts with absorbate molecule. These processes are compared
29 in the schematics in Figure 16. Generally when mentioned in the context of TiO2, photocatalysis refers to sensitized photoreaction. The photocatalytic reaction may be divided into several parts and each step needs to be individually considered to ascertain where improvements may be made to enhance overall photocatalytic activity. The main reaction steps along with time required for each are summarized below. Flash photolysis [49, 50] was the technique which was used to measure the reaction times. Photoexcitation and generation of charge carriers ; 1015 seconds : Migration of holes to the surface and trapping ; 10x109 seconds: Migration of electrons to the surface of the catalyst and trapping; 100x1012 seconds: Recombination of electron with holes trapped in +; 100x109 seconds: Recombination of holes; 10x109 seconds: Oxidation (103 seconds) or reduction (100 x 109 seconds): (1 12) (1 13) (1 14) (1 15) (1 16) (1 17) (1 18)
30 w here vb and cb stand for valence band and conduction band respectively and ir stands for electron excited by radiation. The photocatalytic efficiency of a mater ial is measured in terms of a quantity called quantum yield. Quantum yield can be defined as the number of events happening for each photon that is absorbed. This quantity combines both the charge transfer rate and recombination rate and considers probabil ities for all pathways for electrons and holes . In an ideal situation, quantum yield ( ) is directly proportional to charge transfer rate (kCT) and inversely proportional to the sum of charge transfer rate and electronhole recombination rate (kR) This is expressed in Eq. 120. A close look at the reaction steps reveals that while oxidation/reduction takes time in the range of milliseconds and 100s of nanoseconds, the recombination of holes and electrons takes 10100s nanoseconds. Thus slowing down the rate of recombination of holes is more effective in increasing the quantum yi eld and in turn improving the photocatalytic activity. In the work described in this dissertation an attempt has been made towards this end by using dopant The choice of the specific dopant has been dis cussed in details in section 1.6 .2 1.6 .2 Effect of D opants on Photocatalytic Activity of Titania Using d opant s with a different oxidation state than titanium is a very effective method to improve the photocatalytic activity . Specifically, t he influence of doping transition metals on the photocatalytic activity of semiconductor photocatalyst has been (1 19) (1 20)
31 of interest for several decades now . Transition metal ions like Fe3+ and Cu2+ improves the photocatalytic activity by generating trapping energy level in the forbidden energy band and inhibits electron hole recombination. This is represented v isually in Figure 17. When doped, the transition metal ion can occupy either interstitial sites or substitutional sites depending on the specific ion. Apart from inhibiting electronhole recombination, there is another effect of addition of dopants which has a great importance in terms of applicability of the material. In undoped TiO2 the energy a photon is required to have to start a photocatalytic reaction is greater than the band gap. This is because to form a charge carrying electronhole pair an electron must be excited to the conduction band. But in the presence of electron trapping energy levels, the energy required by a photon is only the energy difference between the valence band and the lowest trapping level. This has some far reaching c onsequences. Because of the wide band gap of 3.2 eV in undoped TiO2, high energy UV radiation is required for the reaction to take place. However when doped with certain transition metal ions, TiO2 has displayed photocatalytic properties in the presence of radiation having wavelength in the visible light range [44, 52, 54, 55]. Only about 3% 5% of solar energy on earth is in the form of UV and thus, if used on terrestrial surface, only a small fraction can be utilized to initiate the photocatalytic activity in such material [44, 57, 59]. So photocatalysis induced by photons in the visible light range is a highly desired property. Figure 18. A) ex plains how specific tr ansition metals change the density of states in TiO2. The electronic states were obtained by using first principal band calculation based on density function theory. In the presence of transition metals with partially
32 filled d orbitals new sta tes appear from the overlap of oxygen orbitals and orbitals from the dopant. As a general rule, the states exhibit a red shift (i.e. decrease in the energy levels) with addition of dopants and this redshift increases as the atomic number of the dopant added increases . This red shift is explained by the charge transfer between the dorbitals of the dopant atoms and the conduction or valence bands of TiO2 [44, 61]. Photoconductivity both from cathodic process (due to electrons in the conduction band) and anodic process (due to holes in the valence band) has been observed in TiO2 doped with transition metals . This observation can be rationalized by the mechanism illustrated in the schematic in Figure 18. B When doped with v anadium, electron transit ion from valence band to the t2g electron level is made possible by a photon having energy in the visible light range. The holes that are left in the val ence band in the process contribute to anodic photocurrent. The case is quite different in case of TiO2 doped with iron. Since i ron has a higher atomic number compared to v anadium, its t2g electron levels have much lower energy and actually overlap the valence band. Photons having energy in the visible light range can lead to electron transfer from the t2g level to the conduction band. In this case anodic photocurrent from the holes and cathodic photocurrent from electron, both are observed. Chromium has an atomic number in between vanadium and i ron and its t2g energy level lies almost at the middle of the v alence and conduction band. In this case two separate electron transitions occur when irradiated by visible light. There is an acceptor transition in which electron is transferred from the valence band to t2g level and a donor transition where electron is transferred from t2g to conduction band. Similar to
33 the case of iron doping, both anodic and cathodic photocurrents are observed due to presence of both holes and conduction band electrons. Even though several transition metal dopants have shown s ignificant enhancement i n photocatalytic activity, little research work has been publis hed on the influence of doping with niobium In one report titania particles doped with niobium has shown improved photocatalytic activity. For each Ti4+ atom substituted by Nb5+ ion one excess electron exists in the 3d orbital of t itanium atom,  leading to better photoconductivity in ni obium doped titania[68, 69]. Niobium has also been observed to inhibit the phase transformation of titania from anatase to rutile[70, 71]. The probable reason for this is red uction in oxygen vacancies[70, 72]. The electronhole recombination rate in anatase is much lower than rutile and hence anatase shows better photocatalytic activity. These factors motivated the research on studying the effect of niobium doping on the photocatalytic property of electrospun titania fibers.
34 Figure 11. Various filtration m echanisms in a ction in c olle ctin g aerosolized particles by a single f iber. The m ethods are A) Interception, B) Inertial Impaction and C) D iffusion. Reprinted from Hinds, W.C., Aerosol Technology, 1999. John Wiley and Sons Inc.,pp 192 194. (Re printed with permission.)
35 Figure 12. C alculated fractional collection efficiency for p articulates of d ifferent sizes. In conjunction with the information on the Filters, it is evident that the E fficien cy depends on Fiber D iameter[15, 16]. Reprinted from A. Podgrski, A. Balazy, L. Gradon, Chemical Engineering Science, 2006, 61, 6804, with permission from Elsevier.
36 Figure 13. Schematic diagram of conventional fiber processing using wet/dry/melt spinning t echnology. Reprinted from Z. K. Walczak, Processes of Fiber Formation, 1/e ed., 2002 pp. 1, with permission from Elsevier.
37 Figur e 14. Schematic diagram of electrospinning set up. A) Set up used in vertical configuration. B) The various forces in equilibrium forming the Taylor cone. Image A is reprinted from D. Li, Y. Xia, Advanced Materials 2004, 1 6 1151, with permission from Elsevier. Image B is reprinted from W. Sigmund, J. Yuh, H. Park, V. Maneeratana, G. Pyrgiotakis, A. Daga, J. Taylor, J. C. Nino, Journal of the American Ceramic Society 2006 89 395, with permission from John Wiley and Sons. A B
38 Figure 15. Ashby chart for wear rate constant vs hardness for various material classes. Reprinted from H. S. Michael Ashby, David Cebon, Materials: engineering, science, processing and design Butterworth Heinemann, 2/e, 2009, pp 231 with permission from Elsevier.
39 Figure 16. Schematics of the p hotocatalytic reaction process A) Catalyzed p hotoreaction. B) Sensitized photoreaction [46, 47]. Reprinted with permission from A. L. Linsebigler, G. Q. Lu, J. T. Yates, Chemical Reviews 1995 95 735. Copyright American Chemical Society. Figure 17. Effect of doping transition m etals on electronic band s tructure in Ti O2. A ) Band structure of TiO2 b efore doping with transition m etal ion. (B ) Band structure of TiO2 after doping with transition m etal ion. Eg is the band gap . This illustration specifically shows electron traps. Creation of hole traps is also possible [59, 61]. Such energy levels are formed close to the valence band. Reprinted with permission from Titania carbon nanotube composites for enhanced photocatalysis, G. Pyrgiotakis, 2006. A B A B
40 Figure 18. Influence of atomic number on the effect of doping transition m etals on the electron band structure of TiO2. A) Density of States (DOS) diagram for TiO2 doped with t ra nsitional m etal A . The lighter line represents the total DOS and the darker line represents the DOS of the dopant metal. B) The process of photoexcitation of electrons under visible l ight irradiation in TiO2 doped with transitional metals vanadium, iron and c hromium . Reprinted from T. Umebayashi, T. Yamaki, H. Itoh, K. Asai, Journal of Physics and Chemistry of Solids 2002 63 19 09, with permission from Elsevier. A
41 Figure 18. Continued. B
42 CHAPTER 2 EXPERIMENTAL METHODS 2.1 Preparation of A queous TiO2SiO2 Hybrid S ol Since electrospinning has been selected as the techniq ue for processing fibers in the body of work described in this dissertation a sol has been prepared for the purpose. As previously mentioned in the background section, ethanol based TiO2 sols are not chemically stable and titania starts precipitating within hours Apart from this heat treated titani a fibers have coarse grains depending on the heat treatment temperature. Since silica has a crystallization temperature of around 1500C, it remains amorphous when heat treated at 500C, the temperature that is required for crystallization of titania. This can pro duce an amorphous SiO2 matrix. The whole method of preparing the sol can be illustrated by the diagram in Figure 2 1. The synthesis involves preparation of two separate sols sol A and s ol B and then they are mixed homogeneously to form the hybrid sol In the first step 0.5 ml of HNO3 (nitric acid conc entrate, 1N, Fisher Chemical) i s added to 99.5 ml of deionized water to modify the pH level of water to approximately 2.3 (0.005N). This is the common starting ingredient for both the sols. To make sol A, 3 m l of H2O (pH modified to 2.3) i s first added to 18.386 ml of 3glycidoxypropyltrimethoxysilane (99.5% Acros) (GPTMS). The mixture i s stirred for 10 20 minutes with a mechanical stirrer. Finally 28.476 ml of titanium (iv) nbutoxide (99%, Acros) (TB) i s a dded to the mixture and stirred for 1 hour t o obtain sol A. Preparation of s ol B i s started by adding 18.386 ml of 3glycidoxypropyltrimethoxysilane (99.5%, Acros) to 66 ml H2O (pH modified to 2.3) and stirring for 2 hours. 7.379 ml of tetraethyl orthosili cate (98%, Acros) (TEOS) i s added to the mixture and stirred for 2
43 hours to prepare sol B. Finally 49.862 ml of sol A i s added to 91.765 ml of sol B and stirred for 2 hours to obtain the aqueous TiO2SiO2 sol. The preparation of sol involves hydrolysis and condensation f or both sol A and sol B. Detailed discussion of the reactions occurring during the process are given in another section. The final resultant sol has suc h a composition that after the heat treatment, the electrospun fibers have a com position of TiO2:SiO2 molar ratio of 1:3. Unlike ethanol based TiO2 sols, the aqueous TiO2SiO2 hybrid sol can be stored for several months, even in the presence of air. This stability has been attributed to the formation of Ti O Si bonds which inhibit for mation and precipitation of TiO2. This has been explained in further details in a later section. 2.2 Electrospinning of Ceramic F ibers To the TiO2SiO2 sol, prepared according to the way mentioned in the previous section, Polyvinylpyrrolidone (PVP) having a molecular weight of 1.3x106 was added to aid the electrospinning process. Th is sol, with the polymeric aid dissolved in it, was filled in a 5 ml BD syringe. A 15 gauge unbeveled needle (Jensen Global) with a nominal inner diameter of 1.372 mm was attach ed to the syringe. While the needle diameter does not appear in the electrospinning equation it still p lays an important role. D epending on the polymer content and the solvent used the sol tends to clog finer needles Because of this a larger gauge diameter is more appropriate for higher polymer content and more volatile solvent. The syringe containing the aqueous sol was attached to a syringe pump and electrospun in vertical configuration. Vertical configuration was used since in this case the gravity pull s the fiber in the same direction as the electrostatic field. In the horizontal configuration the pull due to gravitational energy is perpendicular to the direction of the electrostatic field. Because of this reason
44 the collector distance from the syringe needle has a significant influence on the position and size of the collection area in the horizontal configuration. A vertical set up is ideal in avoiding any such influence. From a purely technical perspective vertical set up is known to result in a flow rate which is in excess to the desired rate. This sometimes leads to formation of beads or even worse formation of just droplets due to Rayleigh instability. However the consistency of the TiO2SiO2 aqueous sol with dissolved PVP polymeric aid was such t hat no such excess flow was observed. The syringe pump used can control the flow rate to the precision of 0.001 ml/h. The flow rate was varied in the course of the research to modify the fiber diameter. Figure 22 shows the experimental set up used during the electrospinning process. Apart from fiber diameter, the quality of the fibers obtained through the process is also important. Beads may form when Rayleigh instability, as described in details in the background section, dominates over the nonaxisymmetic whipping instability. Such beads are undesirable from the perspective of mechanical strength. They also make the air flow through the fibermat filter media complicated and unpredictable. The theory based on the drag force caused by a single fi ber cannot be applied for such fibers and so fibermats with such fibers cannot be used for testing the hypothesis that fibermats with finer fibers have lower pressure drop and higher quality factor Thus detailed imaging using Scanning and Transmission Electron Microscopy was carried out on the electrospun fibers. The various measures which were taken to modify the fiber diameter and the heat treatment are explained in details in the next few subsections. 2.2.1 Modification of Fiber D iameter As has been di scussed in details in the background section, the drag force and in turn the pressure drop decreases with decrease in fiber diameter. However this is a
45 theory based on a single fiber interacting with air. This makes it very important to study the filtratio n properties of fibermats as a function of fiber diameter. Several parameters were varied to this effect. To better appreciate the work it is worth revisiting the equation o n terminal diameter in electrospinning process. where Q is flow rate, is surfa ce tension of the sol, I is the current flowing through the jet. These are the parameters which can be used to vary the fiber diameter. The first parameter that has been varied in an att empt to reduce fiber diameter i s the electrostatic field strength. Terminal thickness of the electrospun jet is inversely proportional to electric field strength and hence decreases with increase in electric field strength. The electrostatic potential gradients used in the research are 0.5 kV/cm and 1 kV/cm. While a viscous enough ceramic sol can be electrospun as it as, most ceramic sol s need polymer to aid in the electrospinning process. The polymeric chains in the system entangle with each other and form the fiber. Polyvinylpyrrolidone (PVP) having a molecular weight of 1.3x106, was used throughout the research to aid the electrospinning proc ess. Polymer can store charge when the electric field is applied. So increase in the polymer content increases the charge content in the sol apart from an increase in the number of polymer molecules that are entangled with each other to form a fiber. An increase in the polymer content of electrospun fibers also means an increase in the volume proportion of material that is burnt off during the heat treatment process and an increase in percentage reduction of fiber diameter after heat treatment While (2 1)
46 the increase in the charge content and the volume percentage t hat is burnt off during the heat treatment leads to a reduction in fiber diameter, the increase in chain entanglement leads to an increase in fiber diameter Keeping this background in perspective, the amount of PVP added was varied during the research. F inally in course of the research, surface tension was varied by adding ethanol in 1:1 volume ratio to the aqueous hybrid sol. When added to water ethanol is known to reduce surface tension significantly. Surface energy of water decreases by more than 60% when 50 mass% ethanol is dissolved in it. 2 .2.2 Heat T reatment Complete crystallization accompanied by complete burnout of PVP can be achieved at 500C for pure TiO2 fibers . But Ti O Si bonds exist in the hybrid sol and the electrospun fibers. This bond inhibits the crystallization of TiO2  Previous research suggests that the crystallization process starts above 800 C. The advantages of having amorphous f ibers over crystallized fibers have been described in the background section and would be detailed further in a later section. So in order to maintain the amorphous structure but at the same time make sure that all the polymeric aid are burnt off, heat treatment was carried out at 600C for 3 hours in air. A well ventilated box furnace was used for this purpose. During heat treatment the electrospun fibermats warp onto themselves and become unusable as filters and so they were heat treated in between two si licon wafers to prevent this. Figure 23 illustrates the XRD patterns of TiO2SiO2 heat treated at several temperatures as observed in previous research.
47 2.3 Preparation and Electrospinning of Doped Titania S ol 2.3.1 Preparation of Titania S ol As has be en discussed in details in the background section, in an environment where sources of energy are limited, it is of considerable practical interest to have photocatalytically active filter media which can be cleaned either by exposing to ultra violet radiat ion and better yet visible light. The primary focus in this part of the research is to develop and study photocatalytically active fibers. Since titania is a commonly used photocatalytic material this gives a baseline to compare the photocatalytic activit ies of doped materials that have been synthesized in an attempt to enhance the photocatalytic activity. A sol based on ethanol was prepared since titanium alkoxides, which are commonly used as precursors to TiO2, readily hydrolyzes and precipitates withi n a few seconds in presence of water. 3 ml of titanium butoxide was added to a solution of 3 ml of ethanol and 3 ml of glacial acetic acid. The mixture is mechanically stirred for 1 hour to obtain the final titania sol. A cetic acid is added to stabilize the sol and slow down the precipitation process considerably 2.3.2 Addition and Electrospinning of Niobium Doped Titania Sol Sol gel route was selected to dope the titania sol, prepared in the way described above, with niobium. Niobium ethoxide (99.95%, Thermo Fisher) was used for the purpose. Calculated amount of niobium ethoxide was added to produce sols containing 2.5 at%, 5 at% and 10 at % of Nb. Table 21 lays out the exact amount of niobium ethoxide added to achieve desired dopant levels. To ensure tha t the sol is homogeneous it is mechanically stirred for 1 hour at room temperature. Another important factor that has to be considered is that niobium ethoxide is prone to hydrolysis
48 in presence of the moisture present in air. Because of this care has been taken to minimize the exposure time of niobium ethoxide to air. It is worth mentioning that the shelf life of the Nbdoped TiO2 sol i s significantly lower than the aqueous TiO2SiO2 sol. It must be electrospun within 2 days after synthesis, otherwise precipitates starts forming both titania sol and niobium ethoxide being prone to hydrolysis. T he niobium doped titania sol s were electrospun using a vertical set up as described in section 2.2. However there are certain specific modifications made for elec trospinning the doped sols. 4% w/v polyvinylpyrrolidone (PVP) was added to the Nb doped titania sol and stirred for 1 hour before electrospinning. The same quantity of polymer aid was added irrespective of the dopant level The flow rate and electric field strength were kept at 0.6 ml/h and 1 kV/cm respectively, for all the Nbdoped titania sol samples. These measures help eliminate complications in interpretations as much as possible and provide the opportunity to focus on the effect of dopant concentratio n on the photocatalytic activity of titania fibers. The collector used in case of doped titania samples was slightly different from that used for pure titania sol. An activated carbon fiber (ACF) fibermat was placed on top of the aluminum foil connected with the terminal. When fibers come in contact with aluminum foil they transfer stored electrostatic charge. However unlike aluminum foil, ACF fibermat has low electrical conductivity. Thus when collected over ACF fibermats the charge stored in the fibers cannot be dissipated as easily and the fibers deposited on the collector repel each other. This fact has been used to inhibit fusion of fibers while
49 they are still drying, improving the mechanical properties. This would be discussed in more details in a later section. 2.4 Determination of Filtration P roperties The standard used for designing the experiment is based on the Department of Energy (DOE) standard for HEPA filters. The filtration properties were measured for 300 nm par ticles at a face velocity of 4.1 liters per minute. The experimental procedure would be laid out in further details in the next few paragraphs. During the course of the research TiO2SiO2 fibermats in particular, were tested for filtration proper ties. Even though the fibermats are mechanically robust and flexible and are easy to handle, they are not strong enough to be held tightly in the filtration testing fixture. Because of this reason, the fibermats were sandwiched between two ACF fibermats (A CF, American Kynol, Inc., NY). The fibermats were cut into circles with crosssection al diameter of 47 mm. The sandwiched fibermat was fitted snuggly inside the fixture as shown in Figure 24. The filtration testing set up can be broadly divided into three parts aerosolized particle generation unit, the filter and finally the particle counter and analyzer. An attached pressure gauge measures the pressure drop across the filter. As has been previously described in the background section, the most penetrating particle size for fiber based filter media is in the range from 500 nm to 100 nm depending on the fiber diameter and hence for high efficiency particulate air (HEPA) filters it is recommended to perform the tests for 300 nm particles. A suspension of spherical polystyrene latex (PSL) particles having a diameter of 0.29 m (<= 3% CV, Thermo Scientific, 5030A) was prepared in deionized water by adding 2g/liter of the PSL particles. This aqueous suspension was aerosolized in a six jet Collsion neb ulizer
50 (Model CN25, BGI Inc., MA). This aerosol was dried inside a dilution drier as shown in the Figure 24. This step is performed so that the water on the surface of the PSL particles maybe evaporated and the air stream would carry particles which have a very narrow size distribution around 0.29 m. It is i n t he second part of the set up where the filtration takes place. As mentioned previously, the fibermat is sandwiched between two ACF fibermats for support. The face velocity of the air stream with aer osolized particles is 5.3 cm/s corresponding to 4.1 liters per minute (LPM). The pressure drop across the filter is measured by a Magnehelic differential pressure gauge (Model 2010, Dwyer Instruments, IN). The final stage of the experi ment involves a scanning mobile particle counter (SMPS) (Model 3936, TSI Inc., MN) which measures the number of particles in an air stream and also shows the distribution of particles. This information was used to calculate the collection efficiency of the particles. Fibermats with different average fiber diameter and thickness were tested using this set up and details of such particulars including the results will be discussed in a later section. 2.5 Determination of Photocatalytic P roperties The photocatalytic properties of niobium doped fibermats were tested in an aqueous medium. The fibermats were mixed with ethanol (200 proof, Fisher Scientific) and ground up using a mortar pestle. The suspension of ground fibers in ethanol is placed on a heat ing plate and heated at around 70C to evaporate the ethanol and obtain dry ground fibers. To 20 ml of deionized water 1 ml of Brilliant Procion Red MX 5B pigment solution having a concentration of 103 g/ml was added. The pigment has the following chemical structure:
51 Procion Red MX 5B was chosen due to the presence of three benz ene rings and one triazine ring all of which make the dye molecule especially stable compared to dyes which have fewer rings . This is important since it ensures that the pH, temperature and other factors can get stabilized in the system before the degradation of the dye is over. The dried ground fibers were added to the pigment solut ion at a concentration of 0.00048 mg/ml of so lution. A magnetic stirrer was placed in each glass vi al containing the sample and each sample was mechanically stirred continuously during the exposure. As mentioned previously the photocatalytic reaction occur s on the surface and hence new wa ter molecules must be transported to the surface for the creation of oxidizing and reducing species. Stirring the sample during the experiment ensures that newly created species are removed from the surface more quickly than possible by conventional diffus ion. In each batch only two samples were placed inside the chamber. This ensures that the vials would not cast any shadow on each other and all the samples receive equal amount of radiation. All samples were exposed for 8 h to UVB After exposure for the specific time, an UV/Vis Perkin Elmer Lambda 800 spectrometer was used to perform absorption spectroscopy of the samples for wavelengths between 250 nm and 800 nm. The instrument uses a tungsten halogen bulb in the visible range and a deuterium lam p in the ultraviolet range as source for
52 electromagnetic radiation. The presence of particles leads to scattering of l ight and hence very small fraction of light is able to pass through the sample to the detector for the spectroscopy. For this reason, the sample was pulled in a 5 ml BD syringe and then filtered using syringe fil ter having pore diameter of 1.0 m to remove the dispersed fibers P hotocatalytic activity depends on the specific surface area (surface area of a material per unit mass) and materials having larger specific surface area can adsorb more molecules on the surface and shows enhanced photocatalytic activity [79, 80]. Because of this reason, spec ific surface area for each sample was measured according to the theory developed by Brunauer, Emmett and Teller, als o known as the BET technique. To account for the influence of surface area of the material, the specific area determined from the experiment is used to normalize the absorption spectra. The pore size distribution was also obtained using the Barrett Joyner Halenda (BJH) method.
53 Table 21. Addition of niobium ethoxide to titania s ol (9 ml) Composition Volume of niobium ethoxide added 2.5 mol% Nb doped TiO2 0.056 ml 5 mol% Nb doped TiO2 0.114 ml 10 mol% Nb doped TiO2 0.242 ml
54 Table 22. Addition of Nd(NO3)3.6H2O to titania s ol (9 ml) Composition Mass of Nd(NO3)3.6H2O added 0.5 mole% Nd doped TiO2 0.0192g 0.75 mole% Nd doped TiO2 0.0288g 1.0 mole% Nd doped TiO2 0.0385g 1.25 mole% Nd doped TiO2 0.0483g 1.5 mole% Nd doped TiO2 0.0581g 1.75 mole% Nd doped TiO2 0.0680g 2.0 mole% Nd doped TiO2 0.0779g 2.5 mole% Nd doped TiO2 0.0990g 3.0 mole% Nd doped TiO2 0.1180g
55 Figure 21 Processing of TiO2SiO2 aqueous sol. A) Schematic diagram representing the different steps involved in preparation of the aqueous sol. GPTMS: 3glycodoxypropyltrimethoxysilane, TEOS: tetraethylorthosilicate, TB: titanium butoxide. B) TiO2SiO2 sol stored in presence of air after 5 months. No visible precipita tion of t itania particles observed. Figure A is reprinted with permission from H. Park, Electrospinning of Nanofiber for Filtration Media, 2010. A B
56 Figure 2 2. The electrospinning set up used in electrospinning TiO2SiO2. A vertical set u p was used througho ut the research.
57 F igure 23. Heat treatment of TiO2SiO2 fibermat. A) X r ay diffraction p attern for different TiO2SiO2 samples heat t reated at 500C, 800 C and 1100 C for 3 hours from previous research findings. Cr ystallization just starts when heat t reated at 800C for 3 hours. This has been used to plan heat treatment temperature. B) Box furnace used for heat t reating samples. Sample was placed at the same level as the t hermocoupl e to ensure that the effect of thermal gradient of the furnace on heat t reatment is minimal. Figure A is reprinted with permission from H. Park, Electrospinning of Nanofiber for Filtration Media, 2010. A B
58 Figure 24. The filtration testing set up used for determining the filtration properties of e lectrospun TiO2SiO2 f ibermats. Figure 25. Ch amber used to expose sa mple to ultra violet or visible l ight radiation. Duri ng the exposure the sample was mechanically s tir red to reduce the influence of mass t r ansport of dye molecule on the reaction r ate. All samples were exposed to same condition for comparison. Stirrer Fan
59 CHAPTE R 3 RESULTS AND DISCUSSI ONS 3.1 Effect of E lectrospinning C onditions on Fiber M orphology Crystal Structure and Mechanical P roperties of TiO2SiO2 F ibers The main focus of the research has always been to develop a fibremat based filter medium which can be used inhabitat in lunar and Martian environment The ideal characteristics which are desired as far as application in filtration technology is concerned are narrow fiber size distribution, uniform fiber thickness along the length and minimal bead formation. V arious parameters in the electrospinning process were varied to obtain this desired objective. While presence of beads and inhomogeneity in fiber diameter may very well affect the filtration properties but study of how such morphological features influence filtration is beyond the scope of this research. In the case of this research, the influence of process ing parameters on the morphology, crystal structure and mechanical properties of the fibers were of interest. As has been discussed previously th e various parameters that have been modified during the course of the research are a) electric field strength, b) concentration of polymeric aid added and c) flow rate. In addition, the surface tension of the sol was also modified by adding ethanol All these parameters appear in the equation derived for terminal diameter of the jet ( ht) as expressed in Eq. 31. From the equation it appears that fiber diameter is directly related to the surface energy and flow rate of the sol and is inversely rel ated to the electric field strength. Keeping this theory in mind it would be interesting to tabulate all the various attempts that has been (3 1)
60 made to electrospin fibers from the sol with all the various parameters shown next to them. For all sampl es in Table 31, other parameters such as flow rate, electric field strength and collector distance are kept constant. While collector distance does not appear in any equation related to the terminal diameter of jet, there are reasons which make it an impo rtant factor. The terminal diameter is achieved when the repulsive force due to surface charge is compensated by surface tension. Enough separation between needle and the collector must be present to ensure the jet reaches this stage. C onsequently 25 cm has been used as the collector distance throughout the research, which is much larger than conventionally used collector distance. Larger collection distance also helps in deposition of fibers evenly over a larger area. As can be observed from the data the fiber diameter increases with increase in the PVP content used. However no such relationship is observed in the equation for terminal jet diameter in an electrospinning process. The most probable reason behind the observed phenomena is that in presence of higher polymer concentration, the number of chain entanglements in the jet increases. When very low quantity of polymer (between 0.4% w/v and 0.75% w/v) is used, there is a tendency to form beads and the fibers produced exhibit a very large range in terms of dia meter. This may be explained by a qualitative analysis of the various types of instabilities acting on such a system. When the amount of dissolved polymer molecule is lower, the amount of surface charge stored by the system also decreases. Because of this Rayleigh instability becomes dominant over electrically induced nonaxisymmetric whipping instability. Since the electrically induced nonaxisymmetric instability is the reason behind formation of fibers and reduction of
61 their diameters, increase in the c ontribution of Rayleigh instability results in bead formation. Secondly, in the absence of molecular chain entanglements from long chain polymer molecules, multiple jets emerge from the needleend, instead of a steady single jet. This results in huge deviation in terms of fiber diameter. The morphological features are readily visible in Figure 31. The figure also shows a plot representing the average fiber diameter of synthesized TiO2SiO2 fibers as a function of polymer content used to aid electrospinning It must be noted that the fiber diameter that has been discussed until now in this section is the diameter of as processed fibers before heat treatment. The fibers must be heat treated before the polymeric aid is burnt out and all that is left is inorganic ceramic fibermat. Since the initial fiber diameter before heat treatment is reflective of the chain entanglement of the polymer molecules, its not predictive of the final diameter after heat treatment. Since chain entanglement of polymer molec ules is so important, the subject of selection of PVP with molecular weight (MW) of 1.3x106 warrants a little commentary. Attempts were made earlier in the research to electrospin fiber with PVP having MW of 630000. It was observed that even though it is possible to electrospin such a sol but the amount of polymer and flow rate required are higher. Also when such a polymer with low molecular weight is used, there is a greater tendency to form beads. So PVP with a higher molecular weight of 1.3x106 was used for the sake of simplifying the process of interpreting the results The reason for not choosing a molecular weight higher than 1.3x106 is that such an increase would result in increase in fiber diameter Heat treatment was carried out at 600C for 3 hour in atmosphere and fiber diameters measured and are tabulated in Table 32.
62 As mentioned in Table 31, when TiO2SiO2 sol was electrospun with 0.4 w/v percent PVP, no fiber was observed and the corresponding SEM image is predominantly that of droplets, alm os t reminiscent of electrospraying The image before the heat treatment is shown in 31 A Figure 32 represents the same fibermat after heat treatment. Upon heat treatment the beads shrink and fibers become more visible. It is of interest to look at the percentage reduction in fiber diameter after heat treatment and how that varies depending on the polymer content of the electrospinning sol. While fibers electrospun from sol containing high contentration of PVP has more polymer that is burnt off during heat treatment, such increased polymer chain entanglements can trap more TiO2SiO2 precursor. The reduction in cross section area of a fiber varies de pending on which of these two factors is dominant. This is tabulated in Table 33. Ceramic materials, especially nanofibers are extremely brittle. Until this research we found only three studies, and all by the same research group, where flexible TiO2 fibermats were produced . Even TiO2SiO2 fibers, when the fibermats were heat treated at 800C or above, are very brittle and are extremely difficult to handle and almost impossible to use as filtration media. The stress applied by the filtration testing set up fixture and the air flow is too high for us ing such fibermats. However flexible ceramic fibermats were obtained when fibers were collected over an aluminum foil over a heating plate maintained at 85C and then heat treated at 600 C. The fibermat could be easily bent to a radius of curvature of 3.4 mm The flexibility is more at certain areas and such small section s can be bent to an even small er radius of curvature of 1.3 mm. Figure 33. A) C) demonstrates the flexibility of the fibermat. One
63 p ossible reason behind such mechanical properties may be the presence of unburned polymer molecules. However previous research has shown that complete oxidation of PVP takes place when fibers are heat treated at 500C[29, 82]. So the influence of polymer chain entanglements on the flexibility can be ruled out. Such high degree of flexibility is unusual for ceramic fibermats and it is mechanically robust enough for use as filtration media. In the course of our research, equal vol ume of Ti O2SiO2 aqueous sol was mixed with ethanol (200 proof, Fisher Scientific), to reduce surface energy and in turn the fiber diameter. In this particular case, 2% w/v PVP (MW 1.3x106) was added as an electrospinning aid. A comparison o f ele ctrospun fibers containing 2% w/v percent of PVP, with and without ethanol is compared in Table 34. As is evident the average fiber diameter decreased by approximately 133% due to the lowering of surface energy of the sol by ethanol. Scanning and Tr ansmission electron microscopy was used to obtain image of the nanofibers Selected area diffraction of individual fiber heat treated at 600C showed diffused ring pattern, characteristic of amorphous phase. No grain structure was observed in the images ei ther. To understand the mechanism in action behind the enhanced flexibility it is important to revisit the topic of toughness in ceramics. As has been mentioned previously in the background section, polycrystalline ceramics have higher fracture toughness than single crystal ceramics, which in turn have higher toughness compared to amorphous ceramics (glass). This is so because, unlike in single crystal cerami c where crack propagates along a preferred plane, in polycrystalline ceramics once a crack is initi ated it has to traverse through a larger distance along grain boundaries.
64 From this perspective amorphous ceramics should be the most brittle since once a crack is formed in such a material, in the absence of any preferred plane, it propagates perpendicular to the surface throug h the shortest route. However micron sized glass fibers have shown near theoretical strength [40, 41] when the process of crack initiation is inhibited or the number of crack initiation sites are few. Neither any surface defect nor any grain structure was observed i n the SEM and TEM images obtained of the heat treated fibers A diffused ring pattern was observed in selected area diffraction pattern of individual fibers indicating an amorphous structure It is theorized that the reduced number of crack initiation sites in such an amorphous fiber is the reason behind the observed mechanical properties. Figure 34 A) and (B) illustrates the fiber morphology of heat treated fibers electrospun f rom a 1:1 mixture by volume of aqueous TiO2SiO2 sol with ethanol, and using 2% w/v PVP (MW 1.3x106) as electrospinning aid. I t is obvious from previous research findings that 500C is high enough temperature for crystallization of titania and hence presence of crystallized TiO2 is expected in the images and the selected area diffraction pattern. This apparent contradictory observation maybe explained by the fact that Ti O Si bonds are formed during the synthesis of the sol and such bonds inhibit the crystallization of TiO2 [76, 82, 86] D uring the course of preparing the sol hydrolysis of 3gly ci doxypropyltrim ethoxysilane (GPTMS) occurs in the first step In the next stage condensation of the hydrolyzed GPTMS takes place with both titanium butoxide and tetraethylorthosilicate (TEOS). It is during the condensation with titanium butoxide that Ti O Si bonds are formed with the release of butanol. The hydrolysis and condensation reactions are shown in Figure 35 A) and (B) respectively.
65 A final observation may be mentioned about the heat treated TiO2SiO2 fibers The TiO2SiO2 fibers maintain their general fi brous morphology even after vigorously ultrasonicating in deionized water for 20 minutes. In the process they br eak into shorter filaments few tens of micrometer s in length. These short filaments can be used as reinforcements in composite materials. A similar observation was reported in 500 m amorphous silica fiber in a previously published research. 3.2 Filtration P roperties of Electrospun Ceramic F ibermat As has been discussed previously the terminal jet diameter and in turn the fiber diameter in the electrospinning process depends on several factors such as electric field strength between the needle and the collector, flow rate of the sol and the surface energy (i.e. composition). So these parameters were varied to obtain fibermats with different fiber diameter s. The average fiber diameter s after heat treatment o f the various f ibermats that have been t ested for filtration properties are provided in table 35. As expected, t here is a general trend showing increase in fiber diameter with increase in the polymer content and flow rate of the sol. However the fiber diameter decreases notably when PVP content is increased from 6.0% w/v to 6.5% w/v, even though flow rate has been increased by 50% in the later. This can be explained by the fact that the amount of material that can be burnt out increases with increase in the amo unt of polymeric aid used. Also the polymer molecules can store charge in the presence of high electric field. This can explain the reduction in fiber diameter for the sol containing 6.5% w/v PVP As a side note, it is worth mentioning that the fibers of the fibermats used in the filtration tests were s mooth with no observable beads in the SEM images This is important since the hypothesis that pressure drop in fibrous filters decrease with reduction in fiber diameter is based on smooth single fibers. An S EM
66 image of electrospun TiO2SiO2 fibers using 6.5% w/v PVP as polymeric aid is shown in Figure 36 The density of the fibermat was obtained using helium pycnometer and has been found to be 2.26 g/cc. Pore size distribution obtained using BJH technique is presented in Figure 37. It appears that surface pores having diameter approximately 2.5 nm contribute most to pore area. As described in details in the experimental methods section, a dispersion of 0.29 m (<= 3% CV) poly styrene latex (PSL) particles i s aerosolized in the first stage of the filtration test. It is important to understand that even though the PSL particles were largely monodisperse, aerosolized particles had water droplets attached to them. This results in a much larger size r ange for aerosolized particles. The set up has a drying chamber where most of the attached water is evaporated by using a jet of dry air. T he size range for aerosolized particles in the original air stream before it passes through the filter is important To obtain this data the air stream with aerosolized particles is allowed to directly enter the scanning mobile particle counter (SMPS) without any filter in the path. Figure 38 shows the particle size distribution in the original air stream. The particles larger than 0.29 m have water droplets still attached to the PSL particle. The particles much finer than 0.29 m represent the water droplets created by the nebulizer which are not attached to any PSL particles. The high percentage of particles having 0. 29 m diameter is in the range of most penetrative particle size (MPPS). This ensures that the results obtained from the filtration tests performed using this aerosol give the most conservative values. T he most important parameter in determining f iltration properties of a fibermat is its quality factor given by Q[88, 89] in Eq. 3 2
67 Since this parameter is a ratio of collection efficiency and pressure drop, a fibermat with higher quality factor can have higher collection efficiency for same amount of pressure drop thus conserving energy in the process. The relationship between quality factor and fiber diameter is an important one from application perspective. Previous research on the influence of fiber diameter on pressure drop is ambiguous at best. While from the theory a lower pressure drop is expected for finer fibers, increase in pressure drop due to reduction in fiber diameter has also been reported in certain research articles[15, 89]. Sinc e different fibermats tested have different mass and thickness, pressure drop has been normalized with respect to mass and thickness for m eaningful com parison. Figure 39 (A) and (B) show the plot of pressure drop normalized with respect to thickness and mas s respectively vs fiber diameter It may be no ticed that both parameters show a decreasing trend with increase in fiber diameter. This apprent contradiction to theory may be explained by the fact that air has to pass through several layers across the thickness of the fibermat while the theory is based on single fiber consideration. The normalized pressure drop can also be expressed as a function of packing fraction ( ) of fibers. where U, dF are the viscosity of air, face velocity and fiber diameter respectively. P acking fraction ( ) in turn is a function of fiber diameter as expressed in Eq. 34 (3 2) (3 3)
68 w here RF is radius of fiber and 2RC is the distance between the centers of two adjacent fibers. After some mathematical consideration its apparent that packing fraction is proportional to mass/square of filter thickness. S ince mass/( filter thickness )2 decreasses with increase in fiber diameter as shown in Figure 310, it can be inferred that the corresponding packing fraction also decreases. Thus there is a larger fraction of void area in the fibermats with larger fiber dia meter. So less energy is required for air to pass through fibermats with larger fiber diameter Finally the number of stacked layers per unit thickness in a fibermat decreases with increase in fiber diameter. All thes e factors together may explain the observed behvior. Figure 310 C) compares the normalized pressure drop of present electrospun fibermats with those reported in literature and commercially available HEPA and ULPA filters. A particular aspect of the data interpretation must be discussed. T he fibermats were sandwiched between two supporting ACF fib ermats. T he ACF fibermats also contribute to the measured pressure drop and Darcys law has been used to separate out the contribution arising so lely from the ceramic fibermat. According to this law if there are n layers of fibermat then the total pressure drop may be denoted by Eq. 36 w here subscripts t, a and s stand for total, ACF and structures respectively. (3 4) (3 5) (3 6)
69 The second important individual factor is the collection efficiency. A filter may be defined as a high efficiency particulate air (HEPA) filter if the collection efficienc y is equal to or greater than 99.97% for aerosolized 0.3 m particles. Figure 311 shows a plot of collection efficiency vs average fiber diameter. Within t he resolution of our instrument all four fibermats showed 100% collection efficiency for aerosolized 0.29 m particles T he ACF fibermat supports contribute to the collection efficiency similar t o the case of pressure drop. In order to eliminate their contribution and measure the contribution of the ceramic fibemat alone, the penetration factor for each layer may be der ived using Eq. 37 where total penetration factor is given as the product of p enetration factor for all the individual layers[89, 91] : where the subscript t stands for total and p1,p2,p3. stand for the penetration factor for the first, second, third layers and so on respectively. 3.3 Influence of Niobium Dopant on the Morphology, Crystal Structure and Photocatalytic Activity of Electrospun Titania Fiber Unlike on earth, sources of energy are not abundant in the lunar environment. This makes it important to find novel energy efficient ways to clean the filters and photocatalysis fits the bill perfectly. Such a process can use the solar energy on the lunar surface to clean itself without using any external source of energy. Pure titania as well as titania doped with varyi ng levels of niobium were electrospun. The fiber morphologies of all such fibers were studied and XRD were carried out to study their crystal linity. They are compared for their photocatalytic activity in this section As discussed previously in details in the experimental section, TiO2 sol was prepared by mixing titanium butoxide in a 1:1 volume mixture of ethanol and acetic acid. (3 7)
70 Niobium ethoxide was used to dope the sol with niobium. PVP was further added t o aid the electrospinning process. It m ay be recalled that pure titania fibermats are very brittle after crystallization. Electrospun fibers, while still wet, fuse together and this fusion hinder s sliding of the fibers over each other. Part of the brittleness observed in the electrospun ceramic fibermats is because of this fusion, apart from the contribution of the ease of crack initiation in crystallized fibers. To reduce this type of fusion two modifications have been made in the collector. Firstly, the collector is maintained at a temperatur e of around 85C, which is high enough for fast drying of ethanol used as solvent in the sol. Secondly, but more importantly, an ACF fibermat is placed on the aluminum foil which in turn is connected to the ground. The electric field req uired for the elect rospinning i s in between the needle and the grounded aluminum foil. But since ACF fibermat s have very low electrical conductivity, the charged fibers deposited on the ACF surface cannot transfer their charge as easily and efficiently as possible in case of a metallic collector like aluminum foil. Similarly charged fibers deposited on the ACF repel each other and this decreases the probability of contact between fibers. These two factors in combination, result in fibers being deposited over not only a larger area but also having a more uniform high thickness across larger c ross sectional area. Figure 312 compares the thickness across the diameter of the cross section of deposited fibermat. The enlarged pictures show that the modified collector results in lar ger cross section with higher thickness Its also evident that the edges of the fibermat collected directly on aluminum foil tend to delaminate into layers. No such delamination was observed for the fibermat deposited on ACF fibermat. The fiber diameter f or pure titania as well as titania fibers doped with varying levels of niobium dopants are presented in
71 Table 36. All parameters apart from the dopant level were kept constant to simplify comparison of the effect of Nb dopant on the morphology and photocatalytic activity of the fibers Whi le establishing a clear trend of the effect of niobium doping on the fiber diameter requires more work and larger number of data points, its clear that niobium increases the fiber diameter by almost two folds a nd also increases the standard deviation. Figure 313 shows a plot of fiber diameter versus the corresponding amount of niobium added as a dopant. The fiber diameters as well as standard deviation are significantly higher for 2.5 at% Nb compared to 5 at% a nd 10 at% Nb doped titania fibers. In fact the variation of fiber diameter between 5 at% and 10 at% Nb doped titania is insignificant. Addition of niobium ethoxide in the sol is expected to modify its electrical conductivity and surface tension. It is hypothesized that interplay between these two parameters is responsible for the initial sharp increase followed by decrease in fiber diameter with increase in dopant amount A detailed study of the reason behind this behavior is beyond the scope of the present research and has been left as a future work. The cry stal structure of the fibers is of considerable interest from the point of view of photocatalysis for several reasons Firstly, while ther e are scarcely any research article mentioning photocatalysis in amorphous TiO2 [92, 93], almost all available literature is on crystallized titania . Even in the very few ar ticles reporting photocatalytic properties of amorphous TiO2, crystallized TiO2 shows far greater photocatalytic activity compared to the amorphous phase. Secondly, anatase exhibits better photocatalytic activity compared to rutile. This is because even though rutile has a higher bandgap
72 compared to anatase, the electron hole recombination rate is lower in anatase . And finally, semiconductor junction has been proven to improve the photocatalytic activity[96, 97]. In this case if a mixture of anatase and rutile is formed during heat treatment it can possibly enhance the photocatalytic property. This is because even though their valence bands are at similar energy levels, the band gap of anatase is 0.2 eV less compared to rutile and hence the conduction ban d of anatase can act as a trapping level. Figure 3 14 shows the X ray diffraction pattern of titania fibers doped with varying degrees of niobium. The results are as expected from theory. All the fibers were heat treated at 600C for 3 hours. Rutile phase has strong peaks at 27, 36 and slightly below 55 and these peaks are very strong in pure titania fibers However the intensity of these peaks decrease monotonously as the amount of niobium dopant was increased. Some of the peaks in pure titania have contribution from both anatase and rutile but as Nb content increased, the intensity of such peaks decreased. Niobium is a pentavalent element and hence, once present in the lattice, results in reduction in the number of oxygen vacancies needed for anatase to rutile phase transformation. This is the most probable reason behind the strong suppression tendency of niobium on anatase to rutile transformation process. Similar results have been observed previously in Nb doped titania particles [71, 99]. One vital importance of this effect in rea l life applications is that filters made of Nb doped titania would not lose their photocatalytic activity after intentional or unintentional exposure to high temperature. The ramification is this that when used inhabitat in lunar missions these fibers can be cleaned both by heating using microwave as well as by
73 exposing to UV radiation. H igh temperature cleaning in microwave would not decrease its ability to be cleaned by photocatalysis. Before finally moving on to the photocatalytic properties of Nb doped titania there is one last factor which must be considered and that is specific surface area. Firstly, since photocatalysis is a surface phenomenon it is affect ed by the specific area (i.e. surface area per unit mass). For this reason all observed photocatalytic activities must be normalized with respect to specific surface. This is to isolate the contribution coming just from the material irrespective of the amount of exposed surface. Secondly, its also important to know if addition of Nb as dopa nt leads to surface porosity. BET technique (named after Stephen Brunauer, Paul Hugh Emmett and Edward Teller) was used towards this end. The specific surface area of titania fibers doped with varying amounts of niobium a re presented in Figure 3 15. The fiber diameters for the corresponding fibers are also provided since specific surface is also a function of fiber diameter. Neglecting packing density, f or a perfectly smooth fiber of length l and material densit y the surface area and mass can be expressed by Eq. 3 8 and Eq. 39 respectively : Since specific surface is the surface area per unit mass it can expressed mathematically by Eq. 310: (3 9) (3 8) (3 10)
74 This implies that for completely smooth nonporous cylindrical fibers specific surface is inversely proportional to the radius and if a much higher specific area is observed than that expected from this, it can be attributed to porosity. For mathematical convenience, i t is assumed that during the grinding process to produce sample for BET, the average length of broken fibers is equal for all the samples. It is evident from the foregoing analy sis and the plots in Figure 315. A) B) that even though there is a decr ease in 1/radius with increase in the amoun t of doped niobium there is a significant increase in specific surface area. This strongly suggests that the doping result s in tremendous increase in surface porosity. Figure 316 compares the cumulative volume o f open pores on the surface for doped and undoped samples and further verifies this assertion. A particular advantage to the increased surface porosity is that more photocatalytic ally active surface is exposed per unit mass even though it may lead to som e deterioration of mechanical properties. In the final part of the present research, all the titania fiber samples, doped with varying amounts of niobium, were tested for their photocatalytic properties. The initial dye solution has 0.0476 g of Proc ion Red MX 5B for each liter of the solution Doped titania fibers, ground using mortar and pestle, was added to the initial dye solution at a concentration of 4.76x104 g/l. They were exposed to UVB radiation for 8h while stirring continuously Finally sy ringe filters with 1 m pores were used to separate out the particles in order to carry out absorption spectroscopy for wavelength in the range of 250 nm 800 nm The absorbance is proportional to the concentration of undegraded dye in the suspension ac cording to the Beer Lambert law as shown in Eq. 3 11; where A, and l are absorbance, absorptivity and length through which light passes:
75 This in turn implies that a decrease in absorbance ( A) is directly related to the amount of dye that is degraded. If C0 is the initial concentration of dye and C is its concentration after exposure to radiation for a certain amount of time, then the decrease in absorbance can be mathe matically expressed by Eq. 312. The reduction in absorbance can thus be used as a pr oxy for d egradation. Figure 317 show s the absorbance of the residual dye when titania fibers doped with varying amounts of niobium, were used as photocatalyst. As has already been discussed in details in the background section, photocatalysis is a surface phenomenon and so all the data were norm alized with respect to specific surface for comparison. For the purposes of Beer Lamberts law, absorbances for various samples are compared at a particular wavelength. Since Procion Red MX 5B dye has maximum absorbance at 538 nm, the normalized absorbance at this particular wavelength for undegraded dye are also plotted versus the corresponding amount of dopant in titania used as photocatalyst From the observations in Figures 317 it is evident that the photocatalytic activity of titania dec reases with i ncrease in the amount of doped niobium Previously published literature on the photocatalytic properties of niobium doped titania is inadequate. The hypothesis that niobium may enhance the photocatalytic activity is bas ed on the effect of transiti on elements like Fe3+ and Cu2+ on creating trapping energy levels in the forbidden gap of titania which in turn prevent electron hole recombination. It has also been theorized that such dopants reduce the band gap and increase the photonic efficiencies. However instead of the expected redshift, a (3 12) (3 11)
76 blueshift is observed in the differential reflectance with increase in Nbdopant level. This is presented i n Figure 318. A ) and ( B ) This is consistent with the corresponding decrease in photocatalyti c activity observed in Nbdoped titania. However, t here is atleast one article reporting improvement in photocatalytic activity of titania by using niobium as dopant. I n that instance, hydrothermal process was used and the synthesized material w as in the particle form. One possible reason behind the contradictory observations in the present research might be the position of Nb atoms in the lattice. In the theory behind the generation of trapping energy level, it is assumed that Nb5+ substitute Ti4+ in the lattice [102, 103]. However if the atoms of Nb5+ exist in the interstitial spaces then the theory might not be applicable. In the absence of sufficient work on photocatalytic properties of Nbdoped titania it can also be hypothesized that Nb5+ substitution creates certain energy levels which instead of preventing electron hole recombination, actually promote such recombination. Previously published r esearch articles on the effect of chromium dopant on photocatalytic properties of titania have mentioned such promotion of the recombination process at Cr3+ sites. The f act that photocatalytic activity of Nb doped titania decreases steadily with increase in niobium content, mostly point towards this explanation. Finally, the presence of semi conductor junction has been known to significantly improve photocatalytic activity by inhibiting electron hole recombination [96, 97]. From the X Ray diffraction studies it is evident that niobium, when used as dopant, highly inhibits the formation of rutile. Thus there is a decrease in anataserutile junctions. This may also lead to a decrease in o bserved photocatalytic beh avior.
77 Table 3 1. TiO2SiO2 aqueous sol compositions and electrospinning conditions used throughout the research for TiO2SiO2 sol [E = 1 kV/cm, Q = 0.6 ml/h, collector distance = 25 cm. ] Composition of sol Amount of PVP added (MW 1.3x106) Me an f iber diameter TiO2SiO2 sol 0.4% w/v Too many beads and uneven fiber diameter TiO2SiO2 sol 0.5% w/v Too many beads and uneven fiber diameter TiO2SiO2 sol 0.75% w/v Too many beads and uneven fiber diameter TiO2SiO2 sol 1.25% w/v 311 nm TiO2SiO2 sol 1.5% w/v 337 nm TiO2SiO2 sol 2% w/v 444 nm TiO2SiO2 sol 3% w/v 1231 nm TiO2SiO2 sol 4% w/v 1337 nm
78 Table 3 2. Fiber diameter of electrospun TiO2SiO2 fibers after heat treatment in p resence of air in a box furnace Composition of sol Amount of PVP added (MW 1.3x106) Heat treatment condition Mean f iber diameter TiO2SiO2 sol 1.25% w/v 600 C for 3h in air. Furnace cooled. 195 nm TiO2SiO2 sol 1.5% w/v 600 C for 3h in air. Furnace cooled. 136 nm T iO2SiO2 sol 2% w/v 600 C for 3h in air. Furnace cooled. 350 nm TiO2SiO2 sol 4% w/v 600 C for 3h in air. Furnace cooled. 803 nm
79 Table 3 3 Loss of fiber diameter on heat treatment at 600C in air for 3 hours PVP content (% w/v) Percent cross section loss after heat treatment 1.25 60.5 1.5 83.7 2.0 37.8 4.0 63.8 Table 34. Influence of addition of ethanol on fiber diameter of heat treated electrospun TiO2SiO2 fibers (2% w/v PVP, MW 1.3x106 used as polymeric aid) Co mposition Average fiber diameter Standard deviation TiO2SiO2 sol 350 nm 50 nm TiO2SiO2 sol + Ethanol (1:1 volume ratio) 157 nm 42 nm
80 Table 35. Modifying fiber diameter of heat treated fiber by varying different electrospinning parameters Composition Flow rate (ml/h) Electric field Strength (kV/cm) Collector distance (cm) Fiber diameter after heat treatment (nm) 1.5% w/v PVP (MW = 1.3x106) dissolved in TiO2SiO2 sol 0.6 1 25 280 35 4.0% w/v PVP (MW = 1.3x106) d issolved in TiO2SiO2 sol 0.6 1 25 803 116 6.5% w/v PVP (MW = 1.3x106) dissolved in TiO2SiO2 sol 1.5 1 25 1138 299 6.0% w/v PVP (MW = 1.3x106) dissolved in TiO2SiO2 sol 1 1 25 1541 295
81 Table 36. Fiber morph ology for electrospun fibers of pure titania and niobium doped titania after heat treating at 600C for 3 hours 4% w/v PVP was added. Electric field strength and flow rate were maintained at 1 kV/cm and 0.6 ml/h. Composition Fiber diameter (nm) Pure titania 1321 3 39 2.5 at% Nb doped titania 2933 1072 5.0 at% Nb doped titania 2091 687 10.0 at% Nb doped titania 2113 833
82 Table 37 Specific surface of niobium doped titania fibers after heat treatment Composition Fib er diameter (nm) Specific surface (m 2 /g) Pure titania 1321 3 39 4.3 2.5 at% Nb doped titania 2933 1072 16.9 5.0 at% Nb doped titania 2091 687 29.6 10.0 at% Nb doped titania 2113 833 26.1
83 Figure 31. Influence of PVP content on the morphology of electrospun TiO2SiO2 fibers. Images obtained using scanning electron microscopy of fibers electrospun from sol containing A) 0.4% w/v PVP and B) 0.5% w/v PVP. Notice the presence of several beads. When fibers are present, they are highly irregular with a huge variation in diameter. C) Plot showing increase in fiber diameter with increase in PVP content. B A
84 1 2 3 4 0 500 1000 1500 Fiber diameterPVP content Fiber diameter Figure 3 1. Continued. C
85 Figure 3 2. TiO2SiO2 fibermats electrospun from sol contai ning 0.4% w/v PVP, after heat treatment at 600 C for 3h. Figure 31 A) shows the same material before heat treatment. There is a significant reduction in the beads after heat treatment
86 Figure 3 3. Flexibility of heat treated TiO2SiO2 fibers. A) TiO2SiO2 fibermat, heat treated at 600C, exhibiting flexibility and can be bent without fracture. (B) and (C) demonstrate that the fibermat may be curved to a radius of curvature of 3.4 mm and in certain portions up to 1.3 mm respectively. Reprinted f rom A. Biswas, H. Park, W. M. Sigmund, Ceramics International 2012 38 with permission from Elsevier. A
87 Figure 3 3. Continued. B C
88 Figure 34. Morphology and c rystallinity of heat treated TiO2SiO2 nanofibers. A) Scanning electron microscopic image of electrospun fibermat. The TiO2SiO2 sol was mixed with Ethanol to lower Surface E nergy and fiber diameter. (B) No s urface defects were observed in transmission microscope image of individual fibers. (C) Diffused ring pattern, obtained from selected area diffraction of fibers, is characteristic of an amorphous structure. A B
89 Figure 34. Continued. C
90 Figure 35. Chemistry behind synthesis of aqueous hybrid sol of TiO2 -SiO2. A) Hydrolysis and (B) condensation reaction during processing of aque ous TiO2SiO2 sol. Reprinted with permission from H. Park, Electrospinning of Nanofiber for Filtration Media, 2010. A B
91 Figure 36 Electrospun micron sized TiO2SiO2 fibers. Scanning electron microscopic image of TiO2SiO2 fibers electrospun using (A) 6.0% w/v PVP as polymeric aid and 1 ml/h flow rate and (B) 6.5% w/v PVP as polymeric aid and 1.5 ml/h flow rate. A B
92 Figure 37. Pore size distribution of electrospun TiO2SiO2 fibermat obtained using BJH method. A) Pores with diameter of approximately 2.5 nm has the largest contribution towards the pore area. B) Almost ent ire volume of pores are from pores with diameter less than 25 nm. A B
93 Figure 38 Original particle size distribution of aerosol generated in the filtration testing set up. A scanning mobile particle counter was used to obtain the data on the distribution.
94 Figure 39. E xperimentally determined pressure drop of electrospun TiO2SiO2 fibermats plotted vs average fiber diameter Pressure drop normalized with respect to (A) fibermat thickness and (B) mass of fibermat, plotted against average fiber diameter. (C) Comparison of electrospun TiO2SiO2 filtermats with data for commercially available HEPA and ULPA filters. A B
95 0 200 400 600 800 1000 1200 1400 1600 1800 0 4 8 12 Observed pressure drop/thickness K. M. Yun et al (2007) Lydair HEPA Lydair ULPA Millipore HEPAObserved pressure drop/thickness (Pa/ m)Fiber diameter Figure 3 9. Continued. C
96 0 700 1400 0.0000001 0.0000002 0.0000003 0.0000004 Mass/Square of thickness (g/( m)2) Average fiber diameter (nm) Mass/Square of thickness Figure 31 0 Plot of mass/ ( square of thickness ) vs average fiber diameter. Mass/square of thickness is proportional to packing density of fibers.
97 0 500 1000 1500 2000 99.8 100.0 Collection efficiency for 300 nm particles (Percent)Average fiber diameter (nm) Observed Collection Efficiency in Electrospun Fibermats HEPA Standard Figure 31 1 Collection efficiency of ceramic fiberma ts plotted versus average fiber diameter. Within the resolution of our instrument, near perfect efficiency was observed in all the four cases.
98 Figure 312. The effect of using activated carbon fiber as collector for electrospun fibers A) The left semi circle is from a fibermat deposited on an ACF fibermat. The one on the right is from a fibermat deposited directly on aluminum. Notice the difference in size. B) The effect of collector modification on the thickness of the cross section is di splayed by placing the fibermat collected over ACF on top of the one collected over aluminum collector. The fibermat collected on aluminum foil has a tendency to delaminate, and as shown in (C) and (D), this is especially prominent at the edges. An enlarged picture of the cross section of a fibermat, deposited on ACF, is depicted in (E). It shows no such delamination. The titania fibermat is flexible even after heat treating at 600C for 3 hours as shown in figure (F). Figures (A) (E) depict titania fiber mats before heat treatment while figure (F) shows the fibermat after heat treatment. A B
99 Figure 3 12. Continued. C D
100 Figure 3 12. Continued. E
101 Figure 3 12. Continued. F
102 0 5 10 1000 2000 3000 4000 Mean fiber diameter (nm)Niobium dopant level (at%) Mean fiber diameter Figure 31 3 Dependence of fiber diameter after heat treat ment on the niobium dopant level. All other electrospinning parameters were maintained constant for all the four data points.
103 Figure 314. X ray diffraction plots demonstrating the impact of niobium dopant on anatase to rutile phase transformation.
104 Figure 3 14. Continued.
105 0 5 10 0.0004 0.0006 0.0008 1/Average fiber diameter (1/nm)Niobium dopant level (at%) 1/Average fiber diameter 0 5 10 10 20 30 Specific surface area (sq m/g)Niobium dopant level (at%) Specific surface area Figure 31 5 Effect o f niobium dopant on diameter and specific surface of electrospun TiO2 fibers A) 1/fiber diameter plotted vs amount of niobium doped. B) S pecific surface plotted vs amount of niobium doped. B A
106 Figure 31 6 BJH pore size analysis of niobium doped titania fibers. Comparison of distribution of cumulative pore volume for A) pure titania and B) titania doped with 10.0 at% niobium. A B
107 200 400 600 0 6 12 Absorbance/(1/Specific surface)Wavelength (nm) Pure titania 2.5 at% Nb doped titania 5.0 at% Nb doped titania 10.0 at% Nb doped titania 0 5 10 0 4 8 12 Absorbance/(1/Specific surface) at 538 nmAbsorbance/(1/Specific surface)Niobium dopant level (at%) Figure 31 7 Influence of niobium dopant on UV/Vis absorption spectra of titania fibers. A) Spectra of absorbance normalized w.r.t (1/s pecific surface ) for dye solutions after exposure to UVB for 8h. Electrospun titania fibers doped with different amount of niobium were used as photocatalyst. (B) A bsorbance/ (1/s pecific surface ) for 538 nm wavelength. B A
108 300 350 400 450 500 0 3 Differential reflectance (A.U.)Wavelength (nm) Pure titania 2.5 at% Nb doped titania 5.0 at% Nb doped titania 10.0 at% Nb doped titania 0 5 10 3.08 3.12 3.16 3.20 Energy for electron transitionEnergy for electron transition (eV)Niobium dopant level (at%) Figure 318. Differential reflectance of niobium doped titania. A) Differentia l reflectance spectra of doped titania fibers plotted vs amount of niobiu m dopant. B) Photon energy required for electron transition based on differential reflectance spectra. A B
109 CHAPTER 4 CONCLUSIONS AND FUTURE WORK Ceramic filter media was developed fo r in habitat application in the lunar environment. Ceramic s have been chosen as the preferred class of material s d ue to the abrasive natur e of dust parti cles on the Moon. The t wo most common types of filters are porous and fibrous and between these tw o types fibrous filters are preferred as they exhibit lower pressure drop. The lower pr essure drop ensures more conservation of energy But the initial problem with this approach was the general brittleness of ceramic fibers. Most such fibermats are too brittle and cannot withstand the minimum pressure applied during testing and application. However TiO2SiO2 fibermats heat treated at 600C for 3 h ours, were found to be flexible and mechanically robust enough to be used as filtration media. Heat treatment at 600 C for 3 h our is not enough for crystallizing TiO2SiO2 and the material remains amorphous even though the polymeric aid burns off completely. N o surface defect which can act as crack initiation site, was observed on the surface of the glassy fibers under SEM and TEM This reduction in crack initiation sites is proposed to be the reason behind the observed flexibility. Pore size distribution of the TiO2SiO2 fibers obtained using BJH technique, demonstrates that most surface pores have a diameter of approximately 2.5 nm. Tests have been performed to determine the filtration properties of the fibermat. In an apparent contradiction to theoretical expectation, pressure drop normalized with respect to either mass or thickness decreases with increase in fiber diameter. Using mathematical deductions it appears that fibermats with larger fibers have a lower packing density. A lower packing density implies a higher void ratio which in turn means
110 that its easier for air to pass through the filter. This can sufficiently explain the observed discrepancy. The final part of the research is an attempt to increase the utility of the fibermats in such a way that they can be cleaned and reused multiple times. Photocatalysis has been selected for this purpose due to the absence of abundant energy sources in space. Fibers were electrospun from sol s of pure titania and niobium doped titania. The electrospun fibers were characterized for their photocatalytic properties and surface porosity. There seem s to be a steady decline in the photocatalytic activity of titania with the increase in the niobium dopant concentration. In general semi conductor junctions provide trapping levels for electrons and increases photocatalytic activity of a material. Niobium suppresses the phase transformation of anatase to rutile. This decrease in the number and volume of such anatase rutile junctions may be one of the reasons behind the loss i n photocatalytic activity. Also, in the theory behind the hypothesis that doping niobium would increase photocatalytic activity, it is assumed that the niobium atoms substitute titanium atoms from the lattice. Instead, if niobium atoms segregate to the grain boundaries or occupy interstitial spaces the theory is not applicable. There is also a third factor, which if true, can explain the observation. Sometimes doping transition metal in titania creates energy levels which promote electronhole recombinations instead of inhibiting them. This has been observed earlier in case of chromiu m doping of titania. If a similar phenomenon takes place for niobium it can also explain the decrease in photocatalytic activity. And as far as surface area is concerned, it appears that there is a significant increase with increase in the dopant level. B JH analyses of the doped fibers confirm that there is a corresponding increase in the
111 number and volume of surface pores with increase in the amount of niobium added as a dopant Due to the constraints on time and resources several aspects have been left to be studied in the future. Using data obtained from experimental observations in mathematical equations it appears that the packing density of fibers decreases with increase in fiber diameter. But is there a relationship between the applied ele ctric field strength and the packing density? Higher electric field strength ensures a larger electrostatic force which in theory should improve the packing density. This remains an area to be looked into in the future. Finally the location of the depositi on of niobium atoms in the lattice is an important factor. When the sol gel process is used, do the niobium atoms really substitute the titanium atoms in the lattice? Or do they segregate to the grain boundaries? The photocatalytic activity of titania decr eases with increase in the amount of niobium doped into it. But what is the exact mechanism behind this behavior? All t hese are still open questions to be answered in the future.
112 LIST OF REFERENCES  M. Horanyi, Annual Rev iew of Astronomy and Astrophysics 1996, 34 383.  T. J. Stubbs, R. R. Vondrak, W. M. Farrell, Advances in Space Research 2006, 37, 59.  M. Rehders, B. B. Grosshauser, A. Smarandache, A. Sadhukhan, U. Mirastschijski, J. Kempf, M. Dunne, K. Slenzka, K. Brix, Advances in Space Research 2011, 47, 1200.  N. Khan Mayberry, Acta Astronautica 2008 63 1006.  N. Khan Mayberry, J. T. James, R. Tyl, C. W. Lam, International Journal of Toxicology 2011, 30, 3.  W. C. Hinds, Aerosol Technology: Properties, Behavior, and Measurement of Airborne Particles 2nd ed., Wiley Interscience, 1999  DOE STD 302097 Washington, D.C. 20585, 1997.  Understanding Respiratory Protection Against SARS, National Institute for Occupational Safety and Health, 2003  T. Hahn, M. Cummings, A. M. Michalek, B. J. Lipman, B. H. Segal, P. L. McCarthy, Infection Control and Hospital Epidemiology 2002, 23 525.  A. Rengasamy, Z. Zhuang, R. BerryAnn, American Journal of Infection Control 2004, 32 345.  W. C Hinds, Aerosol technology : properties, behavior, and measurement of airborne particles. 2nd ed., Wiley Interscience, 1999 182 205.  C. Y. Chen, Chemical Reviews 1955, 55, 595.  H. C. Yeh, B. Y. H. Liu, Journal of Aerosol Science 1974, 5 191  E. Cunningham, Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character 1910, 83, 357.  A. Podgrski, A. Balazy, L. Gradon, Chemical Engineering Science 2006, 61 6804.  R. S. Barhat e, S. Ramakrishna, Journal of Membrane Science 2007 296 1.  Z. W. Ma, M. Kotaki, R. Inai, S. Ramakrishna, Tissue Engineering 2005, 11, 101.
113  R. Vasita, D. S. Katti, International Journal of Nanomedicine 2006 1 15.  J. A. Matthews, G. E. Wnek, D. G. Simpson, G. L. Bowlin, Biomacromolecules 2002, 3 232.  S. Nair, J. Kim, B. Crawford, S. H. Kim, Biomacromolecules 2007, 8 1266.  E. Ochoa Fernandez, D. Chen, Z. X. Yu, B. Totdal, M. Ronning, A. Holmen, Surface Science 2004 554 L10 7.  Z. K. Walczak, 1/e ed., 2002, pp. 1.  F. K. Ko, in Nanomaterials Handbook 1/e ed. (Ed.: Y. Gogotsi), 2006 pp. 553.  W. Sigmund, J. Yuh, H. Park, V. Maneeratana, G. Pyrgiotakis, A. Daga, J. Taylor, J. C. Nino, Journal of the American Ce ramic Society 2006, 89, 395.  J. Z. Wang, X. B. Huang, J. Xiao, N. Li, W. T. Yu, W. Wang, W. Y. Xie, X. J. Ma, Y. L. Teng, Journal of Materials ScienceMaterials in Medicine 2010 21, 497.  L. J. R. Foster, B. J. Tighe, Polymer Degradation and St ability 2005, 87, 1.  Y. Ono, T. Ichiryu, I. Ohnaka, I. Yamauchi, Journal of Alloys and Compounds 1999, 289 220.  D. Li, Y. Xia, Advanced Materials 2004 16 1151.  D. Li, Y. Xia, Nano Letters 2003, 3 555.  F. Anton, Richard, Schreiber Gastell,Anton, Formhals, United States Patent No. 1975504, 1934  F. Anton, Richard, Schreiber Gastell, United States Patent No. 2160962 1939  Y. M. Shin, M. M. Hohman, M. P. Brenner, G. C. Rutledge, Polymer 2001, 42, 9955.  M. M. Hohman M. Shin, G. Rutledge, M. P. Brenner, Physics of Fluids 2001, 13 2201.  Y. M. Shin, M. M. Hohman, M. P. Brenner, G. C. Rutledge, Applied Physics Letters 2001, 78, 1149.  H. S. Michael Ashby, David Cebon, Materials: engineering, science, process ing and design, ButterworthHeinemann, 2009
114  V. Augugliaro, L. Palmisano, M. Schiavello, A. Sclafani, L. Marchese, G. Martra, F. Miano, Applied Catalysis 1991 69 323.  T. N. Obee, R. T. Brown, Environmental Science & Technology 1995 29 1223.  X. Z. Fu, W. A. Zeltner, M. A. Anderson, Applied Catalysis B Environmental 1995, 6 209.  M. A. Malati, Environmental Technology 1995 16, 1093.  H. Schonhorn, C. R. Kurkjian, R. E. Jaeger, H. N. Vazirani, R. V. Albarino, F. V. Dimarcello, Applied Physics Letters 1976, 29, 712.  P. W. France, M. J. Paradine, M. H. Reeve, G. R. Newns, Journal of Materials Science 1980, 15, 825.  Y. Kanno, Journal of Materials Science Letters 1990, 9 451.  Z. Liu, Q. Zhang, L.C. Qin, Solid Sta te Communications 2007 141 168.  T. Umebayashi, T. Yamaki, H. Itoh, K. Asai, Journal of Physics and Chemistry of Solids 2002 63 1909.  G. Pyrgiotakis, PhD Dissertation thesis, University of Florida (Gainesville), 2006  A. Mills, S. LeHu nte, Journal of Photochemistry and Photobiology aChemistry 1997, 108 1.  A. L. Linsebigler, G. Q. Lu, J. T. Yates, Chemical Reviews 1995, 95 735.  M. R. Hoffmann, S. T. Martin, W. Choi, D. W. Bahnemann, Chemical Reviews 1995, 95 69.  S. T. Martin, H. Herrmann, W. Choi, M. R. Hoffmann, Journal of the Chemical Society, Faraday Transactions 1994, 90, 3315.  S. T. Martin, H. Herrmann, M. R. Hoffmann, Journal of the Chemical Society, Faraday Transactions 1994, 90 3323.  T. J. Kemp, R. A. McIntyre, Polymer Degradation and Stability 2006, 91, 165.  J. M. Herrmann, J. Disdier, P. Pichat, Chemical Physics Letters 1984 108 618.  N. Serpone, D. Lawless, J. Disdier, J. M. Herrmann, Langmuir 1994, 10, 643.
115  Y. Sakata, T. Yam amoto, T. Okazaki, H. Imamura, S. Tsuchiya, Chemistry Letters 1998, 27, 1253.  M. Iwasaki, M. Hara, H. Kawada, H. Tada, S. Ito, Journal of Colloid and Interface Science 2000, 224 202.  H. Kato, A. Kudo, The Journal of Physical Chemistry B 2002, 106 5029.  J. Zhang, L. Xiao, Y. Cong, M. Anpo, Topics in Catalysis 2008 47 122.  E. C. Butler, A. P. Davis, Journal of Photochemistry and Photobiology A: Chemistry 1993, 70 273.  Z. Zhang, C. Shao, L. Zhang, X. Li, Y. Liu, Journal of Colloid and Interface Science 2010, 351 57.  H. Yamashita, Y. Ichihashi, M. Takeuchi, S. Kishiguchi, M. Anpo, Journal of Synchrotron Radiation 1999, 6 451.  W. Choi, A. Termin, M. R. Hoffmann, The Journal of Physical Chemistry 1994, 98, 13669. [ 62] Z. H. Luo, Q. H. Gao, Journal of Photochemistry and Photobiology aChemistry 1992, 63 367.  L. Palmisano, V. Augugliaro, A. Sclafani, M. Schiavello, Journal of Physical Chemistry 1988, 92 6710.  Y. Q. Wang, Y. Z. Hao, H. M. Cheng, J. M. Ma, B. Zu, W. H. Li, S. M. Cai, Journal of Materials Science 1999, 34 2773.  Y. Wang, H. Cheng, Y. Hao, J. Ma, W. Li, S. Cai, Thin Solid Films 1999, 349 120.  Y. Matsumoto, J.i. Kurimoto, T. Shimizu, E.i. Sato, Journal of the Electrochemical Soc iety 1981, 128 1040.  M. Hirano, K. Matsushima, Journal of the American Ceramic Society 2006, 89, 110.  X. J. Lu, X. L. Mou, J. J. Wu, D. W. Zhang, L. L. Zhang, F. Q. Huang, F. F. Xu, S. M. Huang, Advanced Functional Materials 2010, 20, 509. [6 9] P. S. Archana, R. Jose, M. M. Yusoff, S. Ramakrishna, Applied Physics Letters 2011, 98 ,152106.
116  M. Hirano, Y. Ichihashi, Journal of Materials Science 2009, 44 6135.  J. Arbiol, J. Cerda, G. Dezanneau, A. Cirera, F. Peiro, A. Cornet, J. R. Morante, Journal of Applied Physics 2002, 92 853.  R. D. Shannon, J. A. Pask, Journal of the American Ceramic Society 1965, 48, 391.  A. Sclafani, J. M. Herrmann, Journal of Physical Chemistry 1996, 100 13655.  Y. Dai, W. Liu, E. Formo, Y. S un, Y. Xia, Polymers for Advanced Technologies 2010, 22 326.  G. Vazquez, E. Alvarez, J. M. Navaza, Journal of Chemical & Engineering Data 1995, 40 611.  L. P. Xu, Y. X. Zhao, Z. G. Wu, D. S. Liu, Chinese Chemical Letters 2003, 14 1159.  H. Park, PhD Dissertation thesis, University of Florida (Gainesville), 2010.  C. Guillard, H. Lachheb, A. Houas, M. Ksibi, E. Elaloui, J. M. Herrmann, Journal of Photochemistry and Photobiology aChemistry 2003, 158 27.  S. i. Nishimoto, B. Ohta ni, T. Kagiya, Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases 1985 81 2467.  B. Ohtani, Y. Okugawa, S. Nishimoto, T. Kagiya, The Journal of Physical Chemistry 1987, 91 3550.  S. G. L. Sarabjit S ingh, M. Vijayakumar, NanoBiotechnology 2009 5 10.  A. Biswas, H. Park, W. M. Sigmund, Ceramics International 2012 38, 1, 883.  S. J. Park, G. G. Chase, K. U. Jeong, H. Y. Kim, Journal of Sol Gel Science and Technology 2010, 54, 188.  D. A. Reneker, OH, US), Chase, George (Wadsworth, OH, US), Kataphinan, Woraphon (Akron, OH, US), Katta, Prathyusha (Oxnard, CA, US), THE UNIVERSITY OF AKRON (Akron, OH, US), United States Patent Application No. 20080242178 2008  E. T. Bender, P. Katta G. G. Chase, R. D. Ramsier, Surface and Interface Analysis 2006 38 1252.  B. Ding, H. Kim, C. Kim, M. Khil, S. Park, Nanotechnology 2003, 14 532.
117  Y. Liu, S. Sagi, R. Chandrasekar, L. F. Zhang, N. E. Hedin, H. Fong, Journal of Nanoscience and Nanotechnology 2008 8 1528.  J. Wang, S. C. Kim, D. Y. H. Pui, Aerosol Science and Technology 2008, 42 722.  Q. Zhang, J. Welch, H. Park, C. Y. Wu, W. Sigmund, J. C. M. Marijnissen, Journal of Aerosol Science 2010, 41, 230.  K. c. Fan, J W. Gentry, Industrial & Engineering Chemistry Fundamentals 1979, 18, 306.  J. Wang, S. C. Kim, D. Y. H. Pui, Journal of Aerosol Science 2008, 39 323.  K. Tanaka, M. F. V. Capule, T. Hisanaga, Chemical Physics Letters 1991, 187 73.  M. Ka nna, S. Wongnawa, S. Buddee, K. Dilokkhunakul, P. Pinpithak, Journal of Sol Gel Science and Technology 2010, 53, 162.  B. Ohtani, Y. Ogawa, S.i. Nishimoto, The Journal of Physical Chemistry B 1997, 101, 3746.  S. C. Jung, N. Imaishi, Korean Jour nal of Chemical Engineering 2001, 18 867.  L. Wei, C. Shifu, Journal of the Electrochemical Society 2010, 157 H1029.  L. Wei, C. Shifu, Z. Sujuan, Z. Wei, Z. Huaye, Y. Xiaoling, Journal of Nanoparticle Research 2010, 12, 1355.  P. L. Kheam rutai Thamaphat, Boonlaer Ngotawornchai, Kasetsart Journal: Natural Science 2008, 42, 357.  J. T. A. Ahmad, and S. Ismat Shah, Journal of Physics: Conference Series 2007, 61, 11.  S. Brunauer, P. H. Emmett, E. Teller, Journal of the American Chemical Society 1938, 60 309.  D. F. Swinehart, Journal of Chemical Education 1962 39 333.  N. Serpone, The Journal of Physical Chemistry B 2006, 110 24287.  M. Lira Cantu, M. K. Siddiki, D. Munoz Rojas, R. Amade, N. I. Gonzalez Pech, S olar Energy Materials and Solar Cells 2010, 94, 1227.
118  A. Biswas, H. Park, W. M. Sigmund, N. Afshar Mohajer, C. Y. Wu, in 41st International Conference on Environmental Systems American Institute of Aeronautics and Astronautics, Inc., Portland, Oregon, 2011 AIAA 2011 5185
119 BIOGRAPHICAL SKETCH As a graduate student Apratim Biswas has worked on processing and characterizing electrospun fibers as a member of Dr. Wolfgang M. Sigmunds research group in the Department of Materials Science and Engineer ing at the University of Florida. Apart from processing technology, a major part of his work involves studying photocatalysis. He graduated with a Bachelor of Engineering degree with honors in Metallurgy and Materials Engineering from the Bengal Engineering and Science University in India in 2007.