MANANO

Grant Agreement 264710

Seminar : Manufacturing and Applications of Nanostructured Materials Date: 4 September 2013 Location: Kingston University, Roehampton Vale, London SW15 3DW UK Time

Item

Responsible

09:30

Registration and Refreshments (Room RVHW 203)

Kingston University, UK

10:00

Introduction to Seminar (Room RVHW 205)

Chairman, Prof. D.T. Gawne

10:05

Keynote: An overview on silica based gels: an experimental and computational approach

Prof. A. Portugal, Dr. P. Simões, Universidade de Coimbra, PT

10:40

Synthesis of light-weight and mechanically reinforced polymer- silica aerogel for thermal insulation in space applications

Homa Maleki, Universidade de Coimbra, Portugal.

11:10

Vibrational spectroscopy and quantum chemical investigation on polymethylsilsesquioxane aerogel

Mauro Almangano, Active Space Technologies, Portugal

11:40

Coffee (Room RVHW 203)

All

11:50

Self-assembled graphene thin films from graphite intercalation compounds

Kirill Arapov, Technische Universiteit Eindhoven, NL

12:20

Effect of process parameters on splat morphology and coating formation of plasma-sprayed glass

Pedro Bastos, London South Bank University, UK

12:50

Lunch (Room RVHW 203)

13:30

Keynote: Combustion synthesis of intermetallic and ceramic composites (Room RVHW 205)

Prof. T. Zhang, Kingston University, UK

14:00

Influence of the operating conditions of flame spray pyrolysis on the properties of lithium titanate

Vasia Tsikourkitoudi, Centro Tecnológico L'Urederra, Spain

14:30

Electrical conductivity study on CNT films

Karthik Gnanasekaran, Technische Universiteit Eindhoven, NL

15:00

Effect of glass composition on properties of novel nano-structured glass-polymer hybrids

Luciana Serio, London South Bank University, UK

15:30

Coffee (Room RVHW 203)

15:40

Toughening of epoxy matrix fiber reinforced composites with acrylonitrile butadiene and carboxylated acrylonitrile butadiene rubber for Advanced Industrial Applications

Nazli Ozdemir, Kingston University, UK

16:10

Effect of compounding conditions on the morphology and mechanical properties of tin fluoride phosphate glass- polyamide 11 hybrids

Nora Iturraran, Arkema France S.A.

16:40

Summary, Discussion and Networking

All

17:45

Close of Seminar Page 1

Synthesis of light-weight and mechanically reinforced polymer- silica aerogel for thermal insulation in space applications Hajar Maleki, LuisaDurães, António Portugal CIEPQPF, Department of Chemical Engineering, Faculty of Sciences and Technology, University of Coimbra, RuaSílvio Lima, 3030-790 Coimbra, Portugal. E-mail address: [email protected] ABSTRACT Silica aerogels are attractive materials for a number of applications due to their low values of thermal conductivity, very low density and extremely high porosity. However, these materials have very poor mechanical properties, which limits them for their different application. Cross-linking of aerogels with organic polymers is a possible way to enhance mechanical properties of the final material to obtain strong aerogels for practical application. It has also been shown that the type of the silica precursor can have a significant contribution on the properties of aerogels. Here in, the properties of silica aerogels having different aryl-linked and alkyl-linked silanes are examined. In order to have simple, cost and time effective procedures, the sol-gel process and diffusion of monomers to the silica are performed in one single step. In this study, we vary mol % of silicon for alkyl-linked (BTMSH) and aryl-linked (BTESB) and also crosslinker concentration (as a molar ratio of R= [TMSPM] (Silica surface modifier)/[cross-linker]) to improve the strength and thermal conductivity of the resulting aerogels. Therefore, different reinforced silica aerogels with improved properties were obtained. Key words –aerogel, cross-linking, hybrid material, nanoporous materials, tri-methacrylate, solgel

1. INTRODUCTION Due to the a number of specific characteristic such as very low thermal conductivity, very low density and extremely high porosity silica aerogels are attractive materials for a number of building, aeronautical and aerospace applications [1]. However, these types of applications have been severely restricted due to the network fragility of silica aerogels and their weak mechanical properties.[2]. Thus, in order to expand the application range of aerogels, more robust aerogels are needed. Different strategies have been explored to improve the mechanical properties of silica aerogels [3]. In principle, the most effective methods to improve flexibility or elastic recovery in silica aerogels consist in the integration of organic compounds with inorganic part of silica aerogels leading to, organic-inorganic hybrid silica aerogels [4-8]. Until now, several polymeric systems such as

epoxide, polyurea, polyurethane, poly (methyl methacrylate), polyacrylonitrile and polystyrene have been applied to reinforce silica aerogels [4-6, 8, 9]. Normally polymerization is performed via a multi step process. This process involves the introduction of the organic monomers to the silica wet gels through several post gelation washing steps that promote the reaction of the silica functionality with monomers. This multistep liquid-phase based reinforcing approach is quite long and tedious. In order to eliminate these time-consuming post gelation steps, we report, in this work, a one pot synthesis of strong silica aerogels with three different inorganic silica structures. Initially, we modified the surface chemistry of silica aerogels with methacrylate containing silica precursors through co-condensation of Tetramethoxysilane, TMOS, and aryl or alkyl linked bis-silanes with 3-(trimethoxysilyl)

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 2

propyl methacrylate, TMSPM. Due to the dissolution of tri-methacrylate cross-linker, Tris[2-(acryloyloxy)ethyl] isocyanuratecan in methanol (gelation solvent), we introduce the organic monomer to the solution in the initial step of preparation of sol, in order to simplify the process. Then, polymerization occurs by means of a post gelation thermal treatment. The previously reported polymer reinforced silica aerogels through different polymerization techniques, one pot or multi steps process, cause significant increase in density (0.5-0.8 g/cm3) and decrease in the surface area (40600 m2/g). Such strategies also lead to the development of strong silica aerogels with maximum strength ranging from 1.24 to 261MPa. Almost in all reported cases, the main goal was to increase the mechanical robustness without considering the increase in density and most importantly, thermal insulation properties, which are critical properties to consider for space application. In this work, we take into account the main properties of the developed aerogels. We want to synthesize light weight polymer reinforced aerogel with low thermal conductivities. In order to compensate for the increase of the silica backbone thermal conductivity due to the cross-linking with a polymer, we attempt to make an alteration on microstructure patterns of the silica aerogel. This goal can be achieved via co-gelation of the silica backbone with different types of alkyl or aryl-linked bis-silane precursors. For these purpose, we design an experimental structure in which three different critical parameters are varied at three levels. Then, we study the effect of each factor on the measured properties of synthesized aerogels. Such an experimental design is a starting point to understand the relation between the operational factors with the resulting properties of silica aerogels.

2. EXPERIMENTAL DETAILS

Cymit. All reagents were used without further purification. 2.2. Preparation of polymer reinforced silica aerogel For a typical example with 1.3 mol/L of total silicon having 20 mol % TMSPM, 40 mol % BTMSH, and 40 mol % TMOS and R=3, the following procedure was applied: A solution of 1.52 mL (10.56 mmol) of TMOS, 1.74 mL (10.56 mmol) of BTMSH, and 1.23 mL (5.2 mmol) of TMSPM was cooled below 0°C in an ethanol mixed dried-ice bath (Solution A). A second solution containing 12.9 mL of gelation solvent (methanol), 0.73 g of Tris[2(acryloyloxy)ethyl] isocyanurate monomer, 0.07 gr of AIBN (formulated to 10 wt% of the organic monomer), 1.9 mL of H2O ([water]/[silicon]=4) and 0.7 mL of NH4OH was prepared (Solution B). The two solutions were combined and poured into a propylene syringe mold until gel point was reached. Gelation time varied from 1 min to 2 hours depending on the formulation. After aging for 24 hours, the wet gel was demolded and placed in cylindrical tubes containing enough ethanol solvent to cover the gel with the same concentration of the radical initiator used in the gelation step. The samples were refluxed at 60°C for 6 hours to promote polymerization of the organic monomers inside of pores of the wet gel with the silica surface functionalities. After crosslinking, the samples were washed three times with the gelation solvent, with 8 hours intervals between each washing step to remove residual water and ethanol, and finally supercritically dried (Fig.1).

Washing with gelation solvent for 24 h (3*8)

Colloidal sol + cross-linker + initiator

Thermal treatment for polymerization during 6 hrs

Supercritical drying

Cross-linked wet gel

Cross-linked silica aerogels

1.Gelation in 5min-2hrs 2.Aging for 24 hrs

Figure 1. Brief preparation procedure for one pot synthesis of polymer reinforced silica aerogel

2.1. Materials All materials were purchased from Aldrich, except BTMSH, which was purchased from

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 3

3. RESULTS AND DISCUSSION Preparation conditions and resulting properties of polymer reinforced and non-reinforced aerogels are listed in Table 1. Three types of silica precursors were used. The first one, TMOS, is as a core precursor and contributes for the inorganic network formation of silica aerogels. The seconds, Bis(trimethoxysilyl)hexane, BTMSH, and 1,4bis(triethoxysilyl)benzene, BTESB, also contribute to form the inorganic network as co-precursors and are introduced to study the influence of different alkyl and aryl linked bissilane type precursors on the final properties of aerogel. TMSPM is selected to modify the surface chemistry of silica aerogels with methacrylate groups. The selected organic monomer is a multifunctional methacrylate, that in the presence of AIBN radical initiator starts to co-polymerize with silica surface functionality via free radical polymerization. Table 1. Preparation conditions and measured properties for selected polymer reinforced and nonreinforced aerogels. Sample

Density g/cm-3

BTMSH Xb-40 mol %, R=3 X-40 mol%, R=0.5 X-20 mol %, R=3 X-20 mol %,R=0.5

Max. Strength T.C.a (Pa) (W/mk)

X-0 mol %, R=3

0.22 0.39 0.141 0.271 0.173

0.064732 0.0893 0.04429 0.0534 0.0441

213000 270000 25000 399000 22800

X-0 mol %, R=0.5

0.314

0.0938

257000

0.05646

10000

Non-X-40 mol % BTESB

0.184

X-10 mol %, R=3

0.1432

0.03756

10700

X-10 mol %,R=0.5

0.2945

0.04969

279000

X-5 mol %, R=3

0.189

0.0392

87800

X-5 mol %, R=0.5 Non-X-10 mol%

0.314

0.0531

131000

0.145

a.

Thermal conductivity

b.

-X: Cross-linked

c.

Sample is too fragile to test

0.05045

-c

Variables under study in this paper are: i) bissilane type (BTMSH, BTESB) ii) the mole fraction of the total silicon derived from

BTMSH (is varied for levels of 0, 20 and 40 mol%), BTESB (is varied for levels of 5 and 10mol%), iii) the amount of tri-methacrylate cross-linker is given as mole fraction to TMSPM (R) and varied in molar ratios of 0.5, 1.75, 3.Both bis-silane precursors contribute with two silicon atoms in every molecule, and the rest of the silicon is derived from TMOS and TMSPM. CH3 H3C O O CH O Si 3 O H 3C

O O Si O O

+

Si O O

O Si O O

(0-40%)BTMSH OR OEt

OEt Si OEt OEt

EtO Si OEt

(0-10%) BTESB

O O Si O O

+ H3C O Si H3C O O H 3C

O

O O Si O

O O Si O

AIBN, T=60 C

O Si O O

O O Si O

O O

O Si O O

O

O

O O

Tri- methacrylate

O O O Si O

AIBN, T=60 C

O O

Non-crosslinked Si wet gel

O

O

O

O

O O O Si

O O Si O O O Si O

O

O

O

O

N O

O

O

O

O O

O O Si O

TMSPAM

: Silica secondary particles

O O

O N N O N O

O OSi O

O

O

O O Si O

O O

O O Si O

O O N N O N O

Si OO O

Tri- methacrylate

O

O Si O O

O O

O O

O O Si O

Si O O O O Si O

Si OO O O O O Si

Tri- methacrylate

O O O Si

O CH3 Si O CH3 O CH3

O O

O O Si O

O SiOO

AIBN, T=60 C

TMOS

H3C O H3C O Si O H3C

O O

O O Si O

N

O N O

O

O Polymer Cross-linked silica aerogel

Figure 2. Different silica precursors used in this study and proposed cross-linking reaction of prepared silica aerogels with tri-methacrylate. Figure 2 shows the mechanism of polymerization proposed in this study (synthesis and reinforcement of silica aerogel). Randall et al.[3] have shown that hexyl-linked bis-silane (BTMSH) is the most effective among the alkyl-linked bis-silanes studied in their work. Loy et al. [10, 11] have made extensive studies on aryl-linked derived silica aerogels. They concluded that, aryl-linked bissilanes lead to the aerogel with less shrinkage and therefore controlled porosity when compared with aerogels derived from alkyllinked bis-silanes. But there is no systematic report of the mechanical properties of aerogel having aryl –linked bis-silane in the underlying silica structure, namely comparisons with alkyl-linked bis-silane. Therefore, as a complementary study, we attempt to reinforce mechanical properties of aerogels with three different underlying silica structures including alkyl-linked, aryl-linked bis-silane and aerogel without bis-silane precursors. Table 1 presents the properties of the selected aerogels. The compression

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 4

strength of reinforced aerogel is improved over their native counterparts while a slight increase in density and thermal conductivity was obtained by the polymer cross-linking. Representative solid 13C NMR spectra of the selected cross-linked and non cross-linked aerogels with different formulations are shown in Fig. 3. Spectrum b indicates that with R=3, some of the surface methacrylate groups still remain unreacted.

Si-CH2

-CH2 C-C-O C-C=O -C-N-C-O

d)

microstructure pattern of aerogels and extent of silica surface cross-linking, different N2 adsorption values are achieved.

a) Native TMOS: 0.24 g/cm3, 93% porous

O NH-C-NH, Cyanurate C=O, Tri-meth Bz, BTESB C=O, Silica surface methacrylate

C-C=O -C-N-

B, CH2

c) B, CH2

b) X-aerogel, BTMSH 40%, R=0.5: 0.39 g/cm3, 71.1% porous

c) X-aerogel, BTESB10%, R=0.5: 0.249 g/cm3, 86% porous

-C-O C=C

b)

Figure 4. SEM micrographs of selected reinforced and non-reinforced aerogels.

B, CH2

a)

C-C=C CH2-CH2-O-

-C-O

C=C

C=O, Silica surface methacrylate

Figure 3.Solid 13 C NMR spectra of aerogel samples formulated from a) Non-X-TMOS 40% + BTMSH 40% + TMSPMA 20% b) X-TMOS 40% + BTMSH 40% + TMSPMA 20% and R=3, c) XTMOS 40% + BTMSH 40% + TMSPMA 20% and R=0.5, d) X-TMOS 70% + BTESB 10% +

1000 Volume adsorbed (cc/g STP)

Si-CH2

Non-X-BTESB 10% X-BTESB 10%, R=0.5 Non-X-BTMSH 40% X-BTMSH 40%, R=0.5

800 600 400 200 0

0.0

0.2 0.4 0.6 0.8 Relative presure (P/P0)

TMSPMA 20% and R=0.5.

1.2

(a) 3.0 Pore Volume dV/dlog(D) (cc/g)

Comparing the peak areas of cyanurate group at spectra d and c (R=0.5), it can be seen that the extent of polymerization for the aerogel prepared with 10 mol% of BTESB is higher than for the aerogel containing 40 mol% silicon derived from BTMSH. The high extent of cross-linking in this aerogel is due to the low steric hindrance and high accessibility of surface functionality to react with the crosslinker. From SEM images (Fig. 4), the reinforced aerogels contain different microstructure patterns when compared to the native aerogel. In reinforced aerogels the size of secondary silica particles are quiet large and a relative decrease in the surface area values is observed (Fig.5a).This means that the micro and mesoporosity of aerogels collapsed during the cross-linking. However, depending on the

1.0

X-BTESB 10%,R=0.5 Non-X-BTESB 10% X-BTMSH 40%, R=0.5 Non-X-BTMSH 40%

2.5 2.0 1.5 1.0 0.5 0.0 0

20

40 60 80 100 Average Diameter (nm)

120

140

(b) Figure 5. (a) N2 adsorption, and (b) BJH curves from the desorption branches of selected reinforced and non-reinforced aerogels.

As it can be seen in Figure 5b, at R=0.5, despite the high extent of cross-linking in the aerogel with 10 mol% BTESB (confirmed by 13 C NMR), the peak pore diameter is larger when compared to the aerogels with 40 mol% BTMSH. As expected, the larger porous

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 5

diameters result in open structures with improved thermal insulation properties. Therefore, the aryl-linked (BTESB)-derived aerogels are in advantage due to their higher N2 adsorption capability and improved thermal insulation properties (Table 1). 3e+5 a)

3e+5

b) d)

Stress (Pa)

2e+5 2e+5

In this work, more than one order of magnitude improvement in strength (up to 4 MPa) of aerogels is achieved with only doubling the density (up to 0.39 g/cm-3).

1e+5 5e+4

c)

0

e)

0

a) X-BTMSH 40%, R=0.5 b) X-BTESB 10%, R=0.5 c) Non-X-BTMSH, 40% d) X-BTMSH 40%, R=3 e) X-BTESB 10%, R=3

10

20 30 Strain (%)

40

to those previously reported using a multistep process [12]. The methodology is analogous to the one recently used to fabricate epoxyreinforced aerogels [13]. Here in, having in mind the maintenance of the thermal insulation properties of the obtained aerogels and the improvement of the mechanical properties, the density of the obtained aerogels is far below the previously reported epoxy reinforced aerogels.

50

Figure 6. Stress-strain curves for selected trimethacrylate reinforced and non-reinforced aerogels.

Mechanical strength of almost all prepared aerogels in this study is improved by as much as an order of magnitude over their native counterparts. Table 1 and Figure 6 show that the maximum strength at break is increased for increasing concentrations of the cross-linker. This is due to the enlargement of the connection points between the silica secondary particles within the aerogels due to creation of extra connectivity by cross-linking. At R=0.5, due to the high extent of polymerization for aerogels with 10 mol% of BTESB, the mechanical strength is further improved.

4. CONCLUSIONS The mechanical properties of silica aerogel with different silica nanostructures are improved by cross-linking of the surface functionality with tri-methacrylate. In this work, we simplified the manufacturing process by eliminating the diffusion of monomer step. In that way, we produced aerogels of similar density and pore structure

It is also shown that the aerogel made by 10 mol % of silicon derived from BTESB precursor are at the same time stiffer and stronger than those made by incorporation of 0 and 40% of BTMSH, with the comparable density values. According to the SEM micrographs and porosimetry results, it is shown that the microstructure of aerogel is mainly influenced by the underlying silica backbone rather than by the cross-linking effect. Due to the cross-linking, the specific surface area of aerogel made by BTESB is less decreased when compared with other aerogels with the same amount of cross-linker but different underlying silica structure. Also, in this aerogel due to the intrinsic rigidity of the underlying silica and low shrinkage, the mesoporous are quite ordered and larger, therefore, leading to an open structure with improved thermal insulation properties.

ACKNOWLEDGEMENTS The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7-MCITN) under grant agreement No. 264710. The authors would like to thank the DirectorateGeneral for Science, Research and Development of the European Commission for financial support of the research.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 6

REFERENCES [1] A.C. Pierre, G.M. Pajonk, Aerogels and their applications, Chemical Review, Volume 102, Issue 11, 2002, Pages 4243-4265. [2] M.M. Moner-Girona, E.A. Roig, J. Esteve, E. Molins, Sol-gel processing parameters and carbon addition, Journal of Non-Crystalline Solids, Volume 285, Issues 1-3, 2001, Pages 244-250. [3] J.P. Randall, M.A.B. Meador, S.C. Jana, Tailoring Mechanical Properties of Aerogels for Aerospace Applications, ACS Applied Material & Interfaces, Volume 3, Issue 3, 2011, Pages 613-626. [4] N. Leventis, A. Sadekar, N. Chandrasekaran, C. Sotiriou-Leventis, Click Synthesis of Monolithic Silicon Carbide Aerogels from Polyacrylonitrile-Coated 3D Silica Networks, Chemistry of Materials, Volume 22, Issue 9, 2010, Pages 2790–2803. [5] M.A.B. Meador, L.A. Capadona, L. Mccorkle, D.S. Papadopoulos, N. Leventis, Structure-Property Relationships in Porous 3D Nanostructures as a Function of Preparation Conditions: Isocyanate Cross-Linked Silica Aerogels, Chemistry of Materials, Volume 19,Issue 9, 2007, Pages 2247-2260.

[9] H. Zou, S. Wu, J. Shen, Polymer/Silica Nanocomposites: Preparation, Characterization, Properties, andApplications, Chemical Review, Volume 108, Issue 9, 2008, Pages 3893–3957. [10] D.A. Loy, G.M. Jamison, B. M. Baugher, A.S. Myers, R. A. Assink, K.J. Shea Sol-gel synthesis of hybrid organic-inorganic materials. hexylene- and phenylene-bridged polysiloxanes, Chemistry of Material, Volume 8, Issue 3,1996, Pages 656–663. [11] K.J. Shea, D.A. Loy, Bridged Polysilsesquioxanes. Molecular-Engineered Hybrid Organic−Inorganic Materials, Chemistry of Material, Volume 13, Issue 10, 2001, Pages 3306- 3319. [12] B.N. Nguyen, M.A.B. Meador, A. Medoro, V. Arendt, J. Randall, L. Mccorkle, B. Shonkwiler, Elastic Behavior of Methyltrimethoxysilane Based Aerogels Reinforced with Tri-Isocyanate, ACS Applied Material &Interfaces, Volume 2, Issue 5, 2010, Pages 1430–1443. [13] M.A.B. Meador, B.N. Nguyen, D. Quade, S.L. Vivod, Epoxy Reinforced Aerogels Made Using a Streamlined Process, ACS Applied Material &Interfaces, Volume 2, Issue 7,2010, Pages 2162-2168.

[6] D.J. Boday, P.Y. Keng, B. Muriithi, J. Pyun, D.A. Loy, Mechanically reinforced silica aerogel nanocomposites via surface initiated atom transfer radical polymerizations, Journal of Material Chemistry, Volume 20, Issue1,2010, Pages 6863–6865. [7] D.J. Boday, R.J. Stover, B. Muriithi, M.W. Keller, J.T. Wertz, K.A.D. Obrey, D.A. Loy, Strong, Low-Density Nanocomposites by Chemical Vapor Deposition and Polymerization of Cyanoacrylates on Aminated Silica Aerogels, ACS Applied Material & Interfaces, Volume 1, Issue 7,2009, Pages 1364-1369. [8] N. Leventis, Three Dimensional Core-Shell Superstructures: Mechanically Strong Aerogels, Accounts of chemical research, Volume 40, Issue 9, 2007, Pages 874-884.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 7

VIBRATIONAL SPECTROSCOPY AND QUANTUM CHEMICAL INVESTIGATIONS ON POLYMETHYLSEQUIOXESILANES AEROGEL Mauro Almangano1,2, Ricardo Patricio2, António T. G. Portugal1, Pedro N. Simões1 1

2

Department of Chemical Engineering, University of Coimbra, 3030-790 Coimbra, Portugal Active Space Technologies, Rua Coronel Júlio Veiga Simão, 3025-307 Coimbra, Portugal

ABSTRACT Quantum Mechanical (QM) calculations of vibrational data of MethylSilsesQuioxane (MSQ) based aerogel at the B3LYP/6-31+G(d) level of theory are carried out. The Attenuated Total Reflectance Fourier Transform InfraRed (ATR-FTIR) spectroscopy technique is applied to follow the condensation stages of a Sol-gel process and the condensation of colloidal suspension of nanoparticles to the solid phase of bulk material can be monitored. Two characteristic infrared absorption bands are observed at ~1160 cm-1 and ~1060 cm-1 (both stretching mode) and their evolution during the Sol-gel process are analyzed and supported by computer simulation. ATR-FTIR spectra were interpreted by comparison with calculated spectra. A good agreement between the experimental and the computed data is found. Key words – Poly(Methyl Silses Quioxane), aerogel, ATR-FTIR, Quantum Mechanical calculations.

1. INTRODUCTION Aerogels are micro- and meso-porous networks composed by randomly interconnected nano-scale clusters of metal oxides. The aerogel material under investigation was obtained by Sol-gel synthesis. It is well-known that macroscopic behavior, such as elasticity and stiffness, strongly depends on the nanostructural features of the gel network formed in the synthesis. On the other hand, it is difficult following the formation and evolution of the first series of clusters from the very beginning of the Sol-gel reaction. We have used a computational approach to model and simulate the cluster structures that are presumed to be formed during the reaction. Methyltrimethoxysilane (MTMS) was selected as precursor for the production of flexible aerogels by sol-gel synthesis [1][2]. Quantum Mechanical (QM) based calculations of structural and vibrational data of polymethylsesquioxesilanes (PMSQ) based

aerogel were carried out at the B3LYP/631+G(d) level of theory. The most challenging part of such a computational study is the investigation of the polymerization evolution by comparing infrared features obtained theoretically with their experimental counterparts obtained from ATR-FTIR spectroscopy. This provides useful information on the condensation stage of Sol-gel polymerization.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 8

2. EXPERIMENTAL AND COMPUTATIONAL DETAILS Pure (spectroscopic grade) MTMS (CH3Si(OCH3), 98%, Aldrich), methanol (CH3OH, 99.8%, Riedelde Haёn), oxalic acid (C2O2(OH)2, 99%, Fluka) and ammonium hydroxide (NH4OH, 25% in water, Fluka) were used as precursor, solvent, acid and basic catalysts, respectively. The applied Sol-gel route is a two-step acid-base catalysed Sol-gel process followed by ageing and drying stages as described by Durães et al.[3]. ATR-FTIR spectroscopy was applied in studying samples directly in the solid or liquid state without further preparation. All QM calculations were performed at the B3LYP/6-31+G(d) level of theory for isolated molecules Solvent effects were included in the calculations by using the polarized continuum model (PCM). After the geometry optimization of pertinent structures, the nature of the obtained stationary points was verified by vibrational frequency calculations to ensure that the structures were true minima in the potential energy surface. The simulated vibrational spectra were also used in assigning the observed experimental vibrational features.

3. RESULTS AND DISCUSSION The structures/clusters within the plots of Fig. 1 are classified using the notation of Qab where Q stands for the maximum three siloxane bonds for each silicon, a is the actual number of siloxane bonds on each silicon, and b is the number of silicon atoms in the unit. The main FTIR spectral features, detected at different aging times of sols, provide relevant information on the structural evolution of the siloxane skeleton during the process. It is generally accepted that the structures of the substrate out of the sol are not particularly orientated, i.e. the arrangement of the Si-O-Si linear chain and the cyclic, or the ring populations, are made of an average of these population. Nevertheless, it is possible to detect different relative abundance of

these species during the two-step condensation process. We divided the analysis in three different parts: i) condensation by acid catalysis (Fig. 2-a); ii) condensation by basic catalysis (Fig. 2-b); iii) basic condensation after the gel point (Fig. 2-c). In the initial stage of the polymerization, the condensation of conjugated silicic acid molecules leads to a high production of cyclic and linear species consisting of a comparatively ordered structure made by several geometries of the basic Me-Si(OH)3 unit, in caged and linear clusters. This leads to a very broad signal around 1060 cm-1, suggesting different linear species are formed. On the other hand the signal at ~1160 cm-1 indicates that also polycyclic and caged species can be formed. These species are present in different concentrations during the second step. In the initial stage (acid catalysis Fig. 2-a), the signal of caged clusters (1160 cm-1) is weak and broad compared to the signal ascribed to the linear and cyclic structures (1060 cm-1). This indicates that in the first step of condensation in acidic conditions the intramolecular bonding is not extensive. Looking at the evolution of the intensity of the two peaks there is no evidence that the structure is growing. On the other hand, it seems that crosslinking of the primary structure is not favoured. In the second stage, peak related to the caged structures becomes better defined (Fig. 2-b at 1160 cm-1), indicating the idea that the network is growing. This became clear at the gel point (Fig. 2-c), when the two peaks are comparable in intensity.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 9

Fig. 1. Simulated (B3LYP/6-31+G(d)) spectra of PMSQs clusters in the range 950–1400 cm-1.

Fig. 2. ATR-FTIR spectra collected during (a) the acid catalyzed phase of the 2-step Sol-gel reaction ;(b) the base catalyzed phase and (c) after the gel point of the sol The figures are a 2D prospective of a 3D graphs, in which, x axis is the wavenumber, y is the transmittance and Z the time scale. The acid step (fig 2-a) starts at t0 and is followed until t5 at 2640 minutes. Basic step starts at t0 (fig. 2-b), and it is followed until the gel point. Figure 2-c has t0 equal to 240 minutes and is follow until t6 equal to 2580 minutes.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 10

4. CONCLUSIONS Looking towards the characterization of the first condensation products of PMSQs compounds, we have tried in this study to give a qualitative explanation to experimental evidences on how cluster formation is important in the Sol-gel polymerization of monomers, and how such prevailing intramolecular condensation over the intermolecular one can be fundamental to determine the network formation. Therefore, an attempt has been made in assigning the ATR-FTIR (solid and liquid phases) vibrational modes in connection with the molecular cluster formation of the compound MTMS. A relationship between FTIR spectra and microstructural properties is proposed. In particular, the IR peak analysis (experimental and theoretical) has showed to be an important and simple approach to follow the evolution of the structure in time.

Microporous and Mesoporous Materials, 100, 2007, Pages: 350-355. [3] L. Durães, S. Nogueira, A. Santos, C. Preciso, J. Hernandez, A. Portugal. Flexible silica based xerogels and aerogels for spatial applications. Proc. of the 10th International Chemical and Biological Engineering Conference. Coimbra: Universidade de Coimbra, 2008, Pages: 563.

ACKNOWLEDGEMENTS The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7-MC-ITN) under grant agreement No. 264710. The authors would like to thank the Directorate-General for Science, Research and Development of the European Commission for financial support of the research. REFERENCES [1] L. Durães, M. Ochoa, A. Portugal, N. Duarte, J.P. Dias, N. Rocha, J. Hernandez. Tailored silica based xerogels and aerogels for insulation in space environments. Advances in Science and Technology, 63, 2010, Pages: 4146. [2] S. D. Bhagat, C. Oh, Y. Kim. Methyltrimethoxysilane based monolithic silica aerogels via ambient pressure drying.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 11

SELF-ASSEMBLED GRAPHENE THIN FILMS FROM GRAPHITE INTERCALATION COMPOUNDS

Kirill Arapov, Gijsbertus de With, and Heiner Friedrich Eindhoven University of Technology, Den Dolech 2, 5612AZ, Eindhoven, The Netherlands ABSTRACT Solution-processed conductive, flexible and transparent graphene thin films continue drawing attention from science and technology due to their potential for application in organic photovoltaic and flexible displays. Nevertheless, exfoliation and assembly into thin films has not reached much beyond scientific curiosity and industrially up-scalable approaches are still at large. Here we present an up-scalable method for the solution processing of graphite to graphene to large-area, conductive, transparent thinfilms. Contrary to most schemes which use Hummers method for exfoliation, we opt for preserving the conjugation of the π-π electron system throughout the entire process, thus keeping the inherent conductivity of graphene. Therefore we proceed with exfoliation via the graphite intercalation route followed by thermal expansion to obtain expanded graphite. The resulting material can be dispersed in a surfactant-free mixture of 3:2 (vol. parts) isopropanol and propylene glycol to obtain stable colloidal dispersion. The graphene thin film is prepared by self-assembly on an oil/water interface followed by transfer to a substrate of interest. The film thickness is accurately controlled by the amount of the graphene dispersion added on the interface. Thermally annealed thin-films exhibit resistivities in the range between 1 to 3 kΩ/□ at transmittances of up to 75 % at 500 nm wavelength. Keywords – graphene, self-assembly, sheet resistance, thin film, transmittance.

INTRODUCTION Conductive, flexible and transparent graphene thin films continue drawing attention from science and technology due to their potential to replace indium tin oxide (ITO) as transparent electrode. Fabrication techniques include chemical vapor deposition (CVD) [1], epitaxial decomposition of silicon carbide [2], and solution-processing e.g. Langmuir-Blodgett deposition (LB) [3], or film lift-up [4]. The first two techniques require equipment such as vacuum chambers, high temperature furnaces with low throughput, which prevents wide industrial use. In contrast, solution-processing is controllable, inexpensive, up-scalable, and therefore, industrially preferred. Here, the source for graphene-based solution-processing is graphite, a widely available raw material. There are several potentially up-scalable routes to obtain graphene from graphite, for example, by long-time sonication [5] or chemical

modification [6]. The latter includes oxidation of graphite to graphite oxide by Hummer‘s method followed by many research groups around the globe [7]. However, the harsh oxidation leads to the formation of though soluble but nonconductive graphene oxide, reduction of which gives just a partial recovery of conductivity. This constricts the potential of use of graphene oxide to a very few applications that require moderate to low conductivity range. The route of the longtime sonication of graphite is generally inefficient and often requires toxic solvents such as N-methylpyrrolidone [8] and N,Ndimethylformamide [9]. Therefore, still, there is a need in high throughput, cost-effective technique, based on “wet chemistry” approach. Here, we present an easy, efficient and potentially scalable method for manufacturing large area graphene films by self-assembly and transfer by “lift-up” starting from raw graphite. We introduce a technique allowing deposition on glass, polymer or any other substrates of

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 12

different shapes and area of more than 50 cm2 and potentially more than 1 m2.

1. EXPERIMENTAL DETAILS Synthesis of residue graphite bisulfate. The intercalation procedure was conducted according to Rudorff’s technique [10] using sulfuric acid and potassium permanganate as oxidant. A typical synthesis of graphite bisulfate was carried as follows: 2 g of graphite (0.167 mol) was transferred to a flask containing 20 ml of concentrated sulfuric acid a with magnetic stirrer. Under vigorous stirring to create a homogeneous slurry 0.22 g of KMnO4 was added to the flask as one portion. Reaction mass has been left under magnetic stirring for 4 h. The hydrolysis of resulting graphite bisulfate was carried by transferring the reaction mass into 500 ml distilled water. After 4 h of stirring, the reaction mass was left to settle down, and the precipitate was separated by decantation. Further, the solid phase was separated by vacuum filtration and washed by distilled water. The product was dried at room temperature for 24 h. The weight gain as defined by Δm = ((mRGB - mG)/ mG )*100%, where mRGB is the mass of result residue graphite bisulfate, and mG is the mass of starting graphite, was 49 %. Synthesis of the Li-THF-GIC. Preparation was performed according to procedure of Nomine and Bonnetain [11]. 1.28 g of naphthalene C8H10 (0.01 mol) was dissolved in 100 ml of freshly distilled THF by vigorous stirring followed by addition of freshly cut and rinsed in hexane, 0.12 g of metallic lithium (0.017 mol). Then 0.5 g of graphite (0.042 mol) was added to reaction mixture as one shot. The reaction flask was sealed and left for 72 h under magnetic stirring. Li-THF-GIC was separated from side products by decantation followed by rinsing with freshly distilled THF. The residue was filtered and dried in ambient conditions for 20 minutes. Weight gain is 29 %. Thermal expansion of GICs. Typically, the microwave assisted thermal expansion was conducted for 1 minute in a 600 W home appliance microwave oven. The GIC sample was placed at the bottom of crucible in a layer of 3-5 mm. The product was transferred to a glass container and used further without further treatment. The material was analyzed BraggBrentano X-ray powder diffractometer Rigaku Geigerflex using Lindemann capillaries of internal diameter 2 mm

Liquid phase exfoliation of expanded graphite. Typically, 1 mg of expanded graphite was added to 50 ml round bottom flask containing 10 ml of mixture of 3:2 (vol.) isopropanol and propylene glycol. The dispersion was sonicated (Branson 1510 70 W sonication bath) for 2 hours in ice water to obtain deep dark colloidally stable dispersion.

2. RESULTS AND DISCUSSION Our solution-processing approach starts (Fig. 1) from raw graphite, which is both an abundant and cost-effective source of graphene. Contrary to most schemes, which use Hummer’s method for exfoliation [6] we opt for another approach preserving π-π electron conjugation and thus keeping the inherent conductivity of graphene throughout the entire process. In this work we start from donor and acceptor-type graphite intercalation compounds (GIC). While a number of researchers use either a donor-type or acceptor-type GIC as a starting point to graphene, very few of them make a comparison of graphene obtained by each route. To fill this gap, we will evaluate the performance of graphene films by means of conductivity and transparency obtained from donor and acceptor-type GICs. Our approach includes the synthesis of both donortype LiC6-18(THF)3,4-1,4 [12] (Li-THF-GIC) and acceptor-type residue graphite bisulfate (RGB) where the distance increases from 0.335 nm for the starting graphite to 1.24 nm for Li-THF-GIC [15] and to 0.35 nm for RGB [13, 14].

Fig.1. Scheme of graphene dispersion preparation starting from raw graphite.

To increase interplanar spacing more significantly, we subjected the GICs to thermal expansion using a home appliance microwave oven. High-speed thermal treatment of the GIC results in abrupt conversion of intercalated species to the gas phase. The rapidly expanding

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 13

Intensity / a.u.

gas causes a significant increase of distance between the graphene layers along the c-axis of graphite. Utilization of microwave oven speeds up the thermal shock by direct and rapid interaction with polar intercalants (e.g. H2O, SO42-, Li-THF) due to high relative permittivity of the latter and hardly any interaction with the graphite matrix due to its low relative permittivity. Acceptor-type expanded graphite (EG) is highly porous sponge-like structure with clearly distinguishable layers while donor-type EG exhibits a concertina – like structure with layers collapsed into balloons. X-ray diffraction for both donor and acceptor-type derived EGs showed a halo in 2θ = 10 - 400 indicating fully disordered structure of stacked layers (Fig. 2).

been used for film preparation without further purification steps. Self-assembly of graphene thin films has been performed by adding graphene dispersion to an oil/water interface as described elsewhere. The controllable addition of the graphene dispersion allows reproducibly and accurately vary thickness and, thus, conductivity and transparency of the final thin films. An example of deposited film on a glass substrate are presented in figure 3. As prepared, films were dried at 110 0C in ambient conditions.

Fig.3. Self-assembled graphene thin film deposited on a cover glass

10

30 50 2θ / degrees

70

Fig.2. X-ray diffractograms of expanded graphites obtained from donor-type GIC (blue line), acceptor-type GIC (red line).

Next, EG is exfoliated in the liquid phase [16]. Our thin-film preparation procedure does not require solubility of graphene. Unlike graphene oxide, a non-conductive product of harsh graphite oxidation, graphene is not soluble neither in polar nor in non-polar solvent. However, it is dispersible in solvents and mixtures up to certain concentration. Here, dispersion is a key to efficient material exfoliation whereas insolubility is a driving force for self-assembly. It has been found experimentally that an optimal medium for exfoliation is a mixture of 3:2 (vol. ratio) isopropanol and propylene glycol (Figure. 2e) allowing dispersion up to 0.1 mg/ml of graphene content stable for several weeks. All dispersions are colloidally stable for at least one week if stored at room temperature. Storage at 4 0 C and lower significantly prolongs the stability of dispersion for up to 1 month. More importantly, such remarkable properties along with non-toxicity and water-miscibility of components allows us to use prepared dispersions to formulate conductive paints, inks or, as reported here, use for large area thin film self-assembly. The obtained dispersions have

As prepared, the films have high and largely deviant sheet resistance making them irreproducible and, therefore, practically inapplicable. However, conductivity of the films can be improved by thermal annealing [17] at 400 0C in a flow of dry nitrogen to remove water and residue solvents. It is remarkable, that the conductivity increases for both thermally annealed acceptor and donortype derived films by about 60 times and leads to resistances up to 1000 Ω/□ with moderate transmittance. More importantly, the thermal treatment leads to the leveling of conductivities with minimum of any adsorbates within the film. To continue the investigation of thin films’ properties we analyzed sheet resistance (Rs) and light transmittance at 500 nm (T500) (Fig. 4). In order to compare two sets of Rs vs T500 data we apply t-test [18]. According to our calculations, the performance of both types of films is similar resulting in the difference less than the sampling error. This leads us to the conclusion that different routes lead to identical materials after the thermal expansion step, which is theoretically reasonable as thermal treatment is leading to removal of any material between graphene membranes, and thus, to a reset of intrinsic properties. Thus, we found that the properties of our undoped films are very comparable to the best results of other groups for the films obtained by LB film deposition [17].

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 14

ACKNOWLEDGEMENTS The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7-MC-ITN) under grant agreement No. 264710. The authors would like to thank the Directorate-General for Science, Research and Development of the European Commission for financial support of the research. REFERENCES Fig. 4. Sheet resistance as a function of transmittance at 500 nm for thin film prepared from donor-type (red circles), and acceptor-type (black triangles) GICs. Solid lines correspond to linear fits of the data points.

3. CONCLUSIONS In conclusion, we have demonstrated an easy, fast, dispersion-based approach to self-assemble and transfer graphene film onto a substrate of interest. We describe a route to use conductive intermediates through the entire process to end up with highly conductive graphene thin films with better performance in comparison with previously published graphene oxide based approaches. We found the optimal conditions to uniformly disperse graphene as solid particles. At the same time, we considered graphene insolubility as a benefit, which, at certain conditions, triggers self-assembly and allows controlling it with high precision. Further, we have prepared films of both donor and acceptortype intermediates with widely varied conductivities and transparencies to evaluate their performance. Our evaluation showed similar performance for both types of films indicating that at certain steps, supposedly, thermal expansion, two intercalation compounds were unified to the same structure. Annealed and undoped films exhibit surface resistances in a range of 3-1 kΩ/□ with transparencies up to 75% at 500 nm. In addition, from one hand, we believe that the approach demonstrated here is industrially viable as it allows to up-scale manufacturing at every stage with minimal costs. From another hand, controllable doping of such films to achieve desired properties opens large field for any possible application including transparent electrodes, gas sensors and electronics.

[1] X. Li, Y. Zhu, W. Cai, M. Borysiak, B. Han, D. Chen, R. D. Piner, L. Colombo and R. S. Ruoff, Transfer of large-area graphene films for highperformance transparent conductive electrodes, Nano Lett, 9, 2009, 4359-4363. [2] P. Sutter, Epitaxial graphene: How silicon leaves the scene, Nat Mater, 8, 2009, 171-172. [3] J. Kim, L. J. Cote, F. Kim, W. Yuan, K. R. Shull and J. Huang, Graphene oxide sheets at interfaces, J Am Chem Soc, 132, 2010, 8180-8186. [4] Y. Kim, N. Minami, W. Zhu, S. Kazaoui, R. Azumi and M. Matsumoto, Langmuir–Blodgett Films of Single-Wall Carbon Nanotubes: Layer-bylayer Deposition and In-plane Orientation of Tubes, Japanese Journal of Applied Physics, 42, 2003, 7629-7634. [5] Y. Hernandez, V. Nicolosi, M. Lotya, F. M. Blighe, Z. Sun, S. De, I. T. McGovern, B. Holland, M. Byrne, Y. K. Gun'Ko, J. J. Boland, P. Niraj, G. Duesberg, S. Krishnamurthy, R. Goodhue, J. Hutchison, V. Scardaci, A. C. Ferrari and J. N. Coleman, High-yield production of graphene by liquid-phase exfoliation of graphite, Nat Nanotechnol, 3, 2008, 563-568. [6] W. S. Hummers and R. E. Offeman, Preparation of Graphitic Oxide, J Am Chem Soc, 80, 1958, 1339-1339. [7] D. C. Marcano, D. V. Kosynkin, J. M. Berlin, A. Sinitskii, Z. Sun, A. Slesarev, L. B. Alemany, W. Lu and J. M. Tour, Improved synthesis of graphene oxide, ACS Nano, 4, 2010, 4806-4814. [8] X. Zhang, A. C. Coleman, N. Katsonis, W. R. Browne, B. J. van Wees and B. L. Feringa, Dispersion of graphene in ethanol using a simple solvent exchange method, Chem Commun (Camb), 46, 2010, 7539-7541. [9] T. T. Dang, V. H. Pham, S. H. Hur, E. J. Kim, B. S. Kong and J. S. Chung, Superior dispersion of highly reduced graphene oxide in N,Ndimethylformamide, J Colloid Interface Sci, 376, 2012, 91-96. [10] W. Rudorff, Hofmann, V., Uber Graphitsalze, Zeit. Anorg. Allg. Chem., 238, 1938, 1-50. [11] B. L. Nomine M, J Chim Phys 66, 1969, 1731–1741. [12] F. Beguin, H. Estrade-Szwarckopf, J. Conard, P. Lauginie, P. Marceau, D. Guerard and L.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 15

Facchini, Composes ternaires graphite-lithiumtetrahydrofuranne: Synthese et etude par rayons X et resonance magnetique, Synthetic Metals, 7, 1983, 77-84. [13] M. L. P. Yu. L Sementsov, and L G. Chernysh, Structural transformations during preparation of fine forms of exfoliated graphite, Powder Metallurgy and Metal Ceramics, 37, 1998, 545-551. [14] G. Hennig, The Properties of the Interstitial Compounds of Graphite. II. The Structure and Stability of Graphite Residue Compounds, The Journal of Chemical Physics, 20, 1952, 1438. [15] M. Inagaki and O. Tanaike, Determining factors for the intercalation into carbon materials from organic solutions, Carbon, 39, 2001, 10831090. [16] A. O’Neill, U. Khan, P. N. Nirmalraj, J. Boland and J. N. Coleman, Graphene Dispersion and Exfoliation in Low Boiling Point Solvents, The Journal of Physical Chemistry C, 115, 2011, 54225428. [17] K. H. Park, B. H. Kim, S. H. Song, J. Kwon, B. S. Kong, K. Kang and S. Jeon, Exfoliation of non-oxidized graphene flakes for scalable conductive film, Nano Lett, 12, 2012, 2871-2876. [18] G. W. C. Snedecor, William G., Statistical Methods, Blackwell Publishing Professional., Iowa State University Press, 1989

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 16

EFFECT OF PROCESS PARAMETERS ON SPLAT MORPHOLOGY AND COATING FORMATION OF PLASMA-SPRAYED GLASSES Pedro Bastos, Yuqing Bao and David Gawne Department of Engineering and Design, London South Bank University, London SE1 0AA, UK ABSTRACT This paper describes the effect of plasma spray parameters, particle size and chemical composition on splat formation and flattening ratio of plasma-sprayed glass coatings. Borosilicate and soda-lime glasses of specified particle size ranges were investigated. The borosilicate splats exhibited optimum flattening ratios of 1.10 across the entire ranges of power and spray distance investigated. Conversely, the fine soda-lime glass showed high sensitivity to both plasma spray power and spray distance and gave flattening ratios of 1.5 for optimized plasma spray parameters. On the other hand, coarse soda-lime glass resulted in lower flattening ratios than the fine glass of the same composition. The research has shown that optimization of plasma spray parameters, particle size and composition can provide dense, high-quality glass coatings. Keywords – Glass Coating, Plasma Spraying, Splat Flattening.

1. INTRODUCTION Atmospheric plasma spraying is a process used for producing thick coatings. It can be used in many engineering applications, including resistance to corrosion, wear and heat [1]. The process consists of introducing a coating powder into a plasma jet (with temperature and speed up to 15000K and 300ms-1, respectively), in which they are melted, accelerated and projected onto a substrate. One of the main advantages of the technique is that virtually any material can be sprayed as long as its melting point is below the decomposition temperature [2]. When particles impact on the surface of the substrate, they flatten and solidify to form splats. At the time of impact with the substrate, the kinetic energy of the particle is transformed into viscous, thermal and surface energy [1]. The extent of flattening of the feedstock powder particles at impact has a major effect on how the splats knit together and form a dense coating. In particular, it makes a large contribution to properties such as adhesion, porosity and surface topography [1-3]. Furthermore, these parameters are often

interrelated, which makes it difficult to understand the formation mechanisms [4]. This paper studies the formation of plasmasprayed glass coatings, on which little research has previously been undertaken. Glass offers the possibility of forming dense, smooth coatings, which have potential as interlayers in electronic applications, such as solar cells. The research is aimed at investigating the conditions and mechanisms leading to the formation of high-quality coatings. 2. EXPERIMENTAL DETAILS 2.1. Materials A borosilicate glass (Glass A) and a soda-lime glass (Glass B) were produced at Glass Technology Services Ltd (Sheffield, UK). The chemical composition and thermal properties of the glasses are presented in Table 1. The glasses were milled into powders and sieved to selected size ranges. Glass A has an average diameter of 56 μm, while glass B was separated into two different size ranges: B1 (average diameter of 57 μm) and B2 (average diameter of

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 17

76 μm). Diameters were obtained using imaging of SEM pictures. Table 1 - Chemical composition and thermal properties of the glasses used

Spray distance (mm)

100

Torch scan speed (mm/s)

1250

Hydrogen flowrate and arc current (parameters that affect power) and spray distance were varied. Four different power values (37, 40, 48 and 52 kW) and five different spray distances (70, 100, 125, 150 and 200 mm) were used.

A

B

%

%

SiO2

79.28

74.26

Na2O

6.02

13.05

CaO

0.05

10.54

Al2O3

2.21

1.30

MgO

0.40

0.20

K2O

‐ 

0.60

TiO2

-

0.01

Fe2O3

-

0.04

B2O3

12.04

-

Thermal expansion coefficient (x106/K)

3.95

9.13

Splats morphologies were analyzed by SEM and measured using imaging software. Because not all the splats obtained were perfectly round, the area measured for each splat was transformed into its equivalent diameter, this is, the diameter of a circle with the same area. Flattening ratio was calculated using

Softening point (°C)

797

737



Glass transition temperature (°C)

536

573

Chemical composition

2.2. Plasma spray deposition Processing was carried out with a Sulzer Metco F4MB-XL plasma gun under ambient pressure. Glass slides and stainless steel, used as substrates, were fixed on a static sample holder. Tests, unless stated otherwise, were performed using the parameters listed in Table 2. Splats (on glass slides) and coatings (on glass slides) were obtained. Splats were obtained by passing the gun once onto the substrate, using a high gun traverse speed, in order to allow formation of individual splats. Coatings were obtained by scanning the substrate, with steps of 5 mm, five times. Table 2 - Plasma spray parameters Plasma spray parameters

Screening of parameters

Working gas composition

Hydrogen and Argon

Hydrogen flowrate (SLPM)

14

Argon flowrate (SLPM)

50

Carrier gas composition

Argon

Carrier gas flowrate (SLPM)

20

Arc current (A)

600

Power (kW)

48

2.3. Image analysis

.1

where D represents the equivalent diameter of the splat and d the equivalent diameter of the original powder particles. The glass slides with the coatings on top were cut and the cross section observed by SEM.

3. RESULTS AND DISCUSSION 3.1. Morphology study A number of studies have been made on splats flattening. One of the most popular models is the one developed by Madejsk [5], which was obtained for alumina splats: 1 3 1 .2 1.2941

.3



.4

This means flattening ratio of the splats depends on droplet diameter (d), impact velocity (u), liquid density ( ), liquid viscosity ( ) and liquidgas surface tension ( ). Besides these, liquidsolid contact angle also is an important

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 18

Model

C

α

Material sprayed

Watanabe [6]

0.82

0.20

n-eicosane

1.16 0.125

Aluminium

Jones [7]

According to equation 6, flattening ratio depends on Reynolds number (equation 3). This way, a higher flattening ratio can be obtained by increasing particle density, velocity or diameter, or by decreasing its viscosity. Particle density is intrinsic to the material, and cannot be changed. Particle velocity can be changed by controlling plasma jet velocity (controlled by plasma spray parameters, like nozzle diameter, plasma gases flowrate and power). The diameter can also be easily changed. Although viscosity cannot be changed directly, it can be controlled by temperature, according to equation 7. .7 where is the viscosity, a characteristic constant, Q the activation energy for viscous flow, R the gas constant and T the temperature. According to equation 7, if temperature of the particles is increased, their viscosity decreases. This way, Reynolds number increases (according to Eq. 3), leading to higher flattening ratios All these models consider solidification only starts after flattening is complete. 3.2. Effect of deposition conditions

Plasma power

High power values lead to high thermal and kinetic energies at the plasma jet, increasing plasma jet temperature which, consequently, increases particles temperature. Glass A was selected because it is the most common type of glass used for traditional enameling. The splats flattening ratio for different plasma power are shown in Figure 1 (spray distance set at 100 mm). 1.30 Flattening ratio

Table 3 - Flattening ratio parameters used by different authors

3.2.1.

1.20 1.10 1.00 0.90 0.80 30

35

40 45 Power (kW)

50

55

Figure 1 - Evolution of the flattening ratio of glass A versus plasma power

All splats flattening ratios are very similar and low (maximum of 1.08 at 52 kW), due to poor particle flow. Higher power values could not be tested due to equipment limitations. 3.2.2.

Spray distance

Spray distance allows the control of the particles residence time in the plasma jet. The higher the residence time is, the more melted the particles will be. However, very high spray distances can lead to particles cooling down before hitting the substrate. Spray distance was studied and the results are presented in Figure 2 (power set at 48 kW).

Flattening ratio

parameter. Models developed consider the initial kinetic energy of the droplets is dissipated as viscous energy (expressed by the Reynolds number) and surface tension energy (expressed by the Weber number). The high velocities used in plasma spray processes make the Weber number very high. For this reason, Madejsk model can be simplified by: . .5 1.2941 Other models have been developed expressed in terms of Re, in the form of: .6 Table 3 presents the parameters C and α for the different models developed.

1.30 1.20 1.10 1.00 0.90 0.80 0.70 50

75 100 125 150 175 200 Spray distance (mm)

Figure 2 - Evolution of the flattening ratio of glass A versus spray distance

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 19

The flattening ratios are again low and with low variability, with a maximum of 1.05 at a spray distance of 70 mm, due to poor particle flow. Figure 3 is an SEM picture of a coating sprayed with glass A, using the conditions that led to the higher flattening ratio (1.08), 52 kW of power and a spray distance of 100 mm.

increases with increasing power, reaching a maximum of 1.48 at 48 kW. If the power is increased more, the flattening ratio starts decreasing due to decomposition and partial vaporization of the particles. Maximum flattening ratio for glass A was obtained at a higher power (52 kW). For the same deposition conditions, glass B1 always exhibits higher flattening ratio than glass A due to its higher flowability when impacting the substrate. Figure 5 shows flattening ratios of the particles sprayed at different spray distances (power set at 48 kW).

Figure 3 - SEM picture of plasma-sprayed glass A with 52 kW and a spray distance of 100 mm

Flattening ratio

1.80 1.60 1.40 1.20 1.00 0.80

Figure 3 proves the glass did not flow after impact, causing accumulation of unflattened particles, which led to a coating with very low density. 3.3. Effect of glass compositions According to equation 8, viscosity of the droplets can be controlled, not only by temperature, but also by the materials characteristic constant ( ). Glass B1 was then investigated. The flattening ratios of the particles are presented in Figure 4. Flattening ratio

1.80 1.60 1.40 1.20 1.00 0.80 35

40

45 Power (kW)

50

55

50

75

100 125 150 175 200 Spray distance (mm)

Figure 5 - Evolution of the flattening ratio of glass B1 versus spray distance

Flattening ratio reaches a maximum at a spray distance of 100 mm (1.48). Smaller spray distances decrease the residence time of the particles in the plasma jet, reducing its melting degree and leading to poor flowability when impacting against the substrate. For higher spray distances, particles start cooling down before they impact on the surface, leading to poor spreading on the substrate. Again, unlike glass A, glass B1 is sensitive to spray distance. Besides, for the same deposition conditions, glass B1 always exhibits higher flattening ratio than glass A. Figure 6a represents the coating with the higher flattening ratio (48 kW and spray distance of 100 mm), while Figure 6b represents the coating obtained with the same 48 kW, but with a spray distance of 70 mm (flattening ratio of 1.29).

Figure 4 - Evolution of the flattening ratio of glass B1 versus plasma power

Figure 4 shows that, unlike glass A (Figure 1), glass B1 is sensitive to power. Flattening ratio

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 20

a

Table 4 - Glass B flattening ratios for different plasma spray parameters, using different particle sizes

Glass B1  St.  Power (kW) ξ  Dev.  37 1,06 0,16  40 1,24 0,19  48 1,48 0,27  52 1,27 0,20  St.  SD (mm)  ξ  Dev.  70 1,29 0,24  100 1,48 0,27  125 1,30 0,21  150 1,12 0,18  200 1,04 0,20   

b

Figure 6 - SEM pictures of glass sprayed with 100 mm (a) and 70 mm (b) spray distance

Figure 6a proves the coating obtained is better quality and denser than the coating obtained using glass A (Figure 3). As particles flow is better, less pores are obtained, leading to a higher quality coating. However, contrarily to what was expected, coating represented in Figure 6b has a denser structure, with less porosity than Figure 6b. This means that flattening ratio is not the only parameter influencing coating formation. Because the spray distance is low (70 mm), the plasma jet acts as a strong heating source during the spray process, continuously heating the substrate and splats, which melts completely the particles. Because there are no pores left during the particles accumulation, a very dense coating is obtained. 3.4. Effect of particle size To study the influence of particle size glasses B1 and B2 were used. Splats flattening ratios are presented in Table 4.

Glass B2 St.  ξ  Dev. 0,48  0,19 0,89  0,15 0,99  0,16 1,07  0,16 St.  ξ  Dev. 1,08  0,20 0,99  0,16 1,01  0,23 0,98  0,15 0,91  0,20

For low power (37 kW), the flattening ratio of glass B2 (bigger particle size) is only 0.48 because the power is very low and the particle size too big. For these reasons, the plasma is not capable of properly melt the big particles, so when they impact on the substrate bounce back and do not stick. Only the smaller particles are adequately melted and can form a splat, leading to a flattening ratio lower than 1. On the other hand, glass B1 (with a lower particle size) does not have this problem. Even for low power, the particles melt, allowing splat formation. As power increases, so does the flattening ratios of both glasses. This is because the increase of power allows a higher melting degree, even for big particles. Regarding power study, maximum flattening ratio for glass B1 is obtained with 48 kW (1.48), while maximum flattening ratio for glass B2 is obtained with a higher power (52 kW, ratio of 1.07). Higher power leads to higher plasma jet temperature, which provides more heat to the particles. If, like in the case of glass B2, particle size is higher, more heat is required to melt the inner part of the particles. This is the reason why glass B2 needs higher power to achieve maximum flattening ratio than glass B1. However, the flattening ratio is never very high for glass B2 (maximum of 1.07) because the

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 21

power is not enough to properly melt the entire particle. Regarding spray distance study, glass B1 presents a maximum flattening ratio of 1.48 for a spray distance of 100 mm, while glass B2 exhibits a maximum for a spray distance of 70 mm (1.08). Glass B1 flattening ratio decreases to 1.29 for smaller spray distances (70 mm) due to small residence time of the particles in the plasma jet. For spray distances higher than 100 mm, flattening ratio decreses, reaching a minimum of 1.04 for 200 mm. This happens because particles start cooling down before splating on the substrate. Glass B2 is not sensitive to plasma spray distances. Although in general it decreases with the increase of spray distance, the differences are not very significant. SEM pictures of splats obtained using the same parameters are presented in Figure 7.

Figure 8 - SEM picture of the coating obtained with the same plasma spray parameters, for glass B2

Glass B2 (Figure 8) originated a dense coating, with almost no porosity. However, when compared to the coating sprayed with glass B1 using the same conditions (Figure 6b), glass B2 is not as dense and good quality. In Figure 8 the shape of the glass particles is obvious, due to poor glass melting.

4. CONCLUSIONS

Figure 7 - SEM pictures of splats obtained with the same plasma spray parameters, for glasses B2 and B1

Both splats presented in Figure 7 were obtained using the same parameters (48 kW and spray distance of 100mm). The effect of particle size is quite evident by the analysis of the splats. Glass B2, due to its high diameter, was not properly melted, which led to poor flattening ratio. On the other hand, glass B1 was properly melted and good flattening ratio was obtained. SEM picture of the coating plasma-sprayed with glass B2 is presented in Figure 8, using 48 kW and 70 mm spray distance.

1. Dense glass coatings were successfully plasma-sprayed on glass slide substrates. 2. The borosilicate glass showed little sensitivity to plasma spray power and spray distance, exhibiting a maximum flattening ratio of 1.1 at a spray distance of 100 mm and plasma power of 52 kW. 3. The fine soda-lime glass showed high sensitivity to both plasma spray power and spray distance, exhibiting a maximum flattening ratio of 1.5 at a spray distance of 100 mm and 48 kW. This glass provided substantially higher flattening ratios under all deposition parameters than the borosilicate glass. 4. The coarse soda-lime glass gave lower flattening ratios than the fine glass of the same composition. ACKNOWLEDGEMENTS The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7-MC-ITN)

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 22

under grant agreement No. 264710. The authors would like to thank the Directorate-General for Science, Research and Development of the European Commission for financial support of the research. REFERENCES 1.

2.

3.

4.

5.

6.

7.

P. Fauchais, M. Fukumoto, A. Vardelle, M. Vardelle, Knowledge Concerning Splat Formation: An Invited Review, Journal of Thermal Spray Technology, Volume 13, Issue 3, 2004, Pages 337-360. T. Zhang, Y. Bao, D.T. Gawne, Process model of plasma enameling, Journal of the European Ceramic Society, Volume 23, Issue 7, 2003, Pages 1019-1026. J. Mostaghimi, M. Pasandideh-Fard, S. Chandra, Dynamics of Splat Formation in Plasma Spray Coating Process, Plasma Chemistry and Plasma Processing, Volume 22, Issue 1, 2002, Pages 59-84. G. Montavon, S. Sampath, C.C. Berndt, H. Herman, C. Coddet, Effects of Vacuum Plasma Spray Processing Parameters on Splat Morphology, Journal of Thermal Spray Technology, Volume 4, Issue 1, 1995, Pages 67-74. J. Madejski, Solidification of droplets on a cold surface, Int. J. Heat Mass Transfer, Volume 19, 1976, Pages 1009-1013. T. Watanabe, I. Kuribayashi, T. Honda, A. Kanzawa, Deformation and Solidification of a droplet on a cold substrate, Chemical Engineering Science, Volume 47, Issue 12, 1992, Pages 3059-3065. H. Jones, Cooling, freezing and substrate impact of droplets formed by rotary atomization, J. Phys. D: Appl. Phys., Volume 4, 1971, Pages 1657-1660.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 23

INFLUENCE OF THE OPERATING CONDITIONS OF FLAME SPRAY PYROLYSIS ON THE PROPERTIES OF LITHIUM TITANATE (Li4Ti5O12)

V.P. Tsikourkitoudi1,2, C. Fernández1, P.J.S. Foot2, T. Zhang2 1

LUREDERRA Technological Centre, Área Industrial "Perguita", Calle A, nº 1, 31210 Los Arcos (Navarra - Spain) 2

Kingston University London, Faculty of Science, Engineering and Computing, Penrhyn Road, Kingston upon Thames, Surrey KT1 2EE (UK) ABSTRACT The main purpose of the present study is the investigation of the influence of the operating conditions of flame spray pyrolysis (FSP) technique on the properties of lithium titanate (Li4Ti5O12, LTO) powder. In the present study, LTO nanoparticles have been fabricated by FSP, a scalable, one-step, dry process. Lithium titanate is a promising material for lithium-ion battery anodes, but the reduction of its particle size to the nanosize is necessary for the enhancement of its electronic conductivity. However, the decrease of the particle size under a critical value could cause the deterioration of the electrochemical properties of LTO. By varying the operating conditions of FSP, nanoparticles of different sizes are obtained. More precisely, by increasing the precursor feed rate or the precursor concentration, larger nanoparticles are synthesized, whereas by increasing the dispersion gas flow rate, the diameter of the nanoparticles is decreased. The phase composition of the powder is examined by X-ray diffraction and its morphology is investigated by Transmission Electron Microscopy revealing inhomogeneous material with large and small particles, a fact that can be attributed to the presence of isopropanol to the precursor solution. Key words – flame spray pyrolysis, lithium titanate, nanoparticles

1. INTRODUCTION The investigation for the replacement of the present battery components with new materials having higher performance and better safety has led to titanium oxides. More precisely, lithium titanate (LTO) [1] has attracted great interest as anode material for advanced lithium ion batteries, as its lithium insertion potential is between 1,2 and 2,0 V vs. Li, i.e. within the stability window of most common organic electrolytes [2]. Spinel LTO is characterized by a two phase electrochemical transition process with a flat voltage profile. Although, LTO has low theoretical capacity and high potential vs Li, it has some specific properties that make it a promising anode material for the next generation of lithium ion batteries. In the first place, it does

not undergo any significant volume changes during lithiation/delithiation. There is minimal volume change between LTO and its lithiated phase (Li7Ti5O12). For this reason, it is characterized as a “zero-strain” material. Moreover, it intercalates lithium at high potential vs Li metal, a fact that minimizes the risk for lithium plating on its surface. Additionally, it is thermally stable in both the charged and discharged state and it does not provoke the decomposition of the electrolyte, and thus no solid electrolyte interface is formed during cycling. Finally, it presents excellent reversibility during a charge/discharge cycle. Nevertheless, LTO is a poor electronic conductor (~10-13 S/cm) due to the existence of free octahedral sites in its lattice and the oxidation state of Ti4+ and has a moderate Li+ diffusion coefficient (~10-8 cm2/s) [3].

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 24

In order to enhance the electronic conductivity of LTO, several solutions have been proposed, among which the reduction of the particle diameter to nanosize. Electronic conductivity is enhanced due to the improvement of the kinetic properties of the electrons because of the decrease of their diffusion paths. In order to achieve an optimum particle size, the synthesis method should be carefully selected and the synthesis conditions should be optimized. Flame spray pyrolysis (FSP) is a one-step synthesis method of nanoparticles from a precursor solution (liquid feed). The precursor is converted into nanoparticles after the combustion of the spray produced into the flame zone. The size and morphology of the nanoparticles depend mainly on the precursors and the operating conditions of FSP [4]. The most representative parameters that mainly affect the particles size are the precursor feed rate, the dispersion gas flow rate and the precursor concentration. In the present study, LTO nanoparticles have been synthesized by FSP and characterized accordingly. The motivation for the current investigation is the importance of the particle size to the electrochemical performance. Although small particle size facilitates diffusion and improves charge transfer properties, if it becomes too small, the charge capacity of the electrochemical energy storage is decreased due to decrease of lithium vacancy sites in the crystalline bulk [5]. More quantity of lithium is stored mainly on the surface of the particles, leading to reconstruction of the surface by the creation of a layer of inactive material [6]. In this way, as the particle size is decreased, capacity loss is becoming more evident. By changing the operating conditions of FSP, nanoparticles of different sizes can be obtained.

2. EXPERIMENTAL DETAILS 2.1. Precursor solution preparation Nanosized Li4Ti5O12 powder was synthesized by titanium tetraisopropoxide (TTIP) and lithium acetylacetonate (Li-acac) with a stoichiometric ratio 4:5. For the precursor solution, TTIP and Li-acac have been dissolved in different solvents with varied concentrations. More precisely, different solvents have been used, i.e. 2-ethyl hexanoic acid, toluene, xylene and isopropanol. At first place, the solubility of Li-acac in the aforementioned solvents has been investigated. Solutions of concentration of 0,5 M have been prepared. It was found that Li-acac is fully dissolved only in 2-ethyl hexanoic and a transparent solution is obtained after stirring. Attempts to dissolve Li-acac in mixtures of the aforementioned solvents have been performed. It was observed that for the mixture of 2-ethyl hexanoic acid and toluene (or xylene), the volume ratio for which a transparent solution was obtained (i.e. fully dissolved Li-acac after stirring) was 1:4, whereas for higher quantity of toluene (or xylene), there was gelification of the solution provoked by the continuous stirring. For the mixture of 2-ethyl hexanoic and isopropanol, the volume ratio for which a transparent solution was obtained was 2:3. No precipitation occurred during the next day, so the precursor of titanium, i.e. TTIP, has been added to the stable transparent solutions of Li-acac in order to investigate the compatibility and stability of the LTO precursor mixture. It was found that the precursors are compatible and there was no precipitation after the addition of TTIP and the yellow solution was transparent and stable. The fact that there is no precipitation indicates that all the metal components remain dissolved and, hence, they are intimately mixed. This property is essential for FSP because it minimizes the risk for phase separation [7].

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 25

The precursor mixture for the initial experiments was chosen to be that of 2-ethyl hexanoic acid and isopropanol (volume mixture 2:3). 2.2. Flame spray pyrolysis apparatus The flame spray pyrolysis apparatus consists of a gas-assisted atomizer for the generation of the aerosol precursor. The precursor solution flows through the capillary tube in the middle of the atomizer while the dispersion gas (O2) (which assists the atomization of the liquid feed) through its annular gap. A pump is used for the feeding of the precursor solution, which is continuously stirred by a magnetic stirrer during the experiment. The nozzle of the atomizer is surrounded by two annuli that form a diffusion flame by flowing CH4 through the inner and O2 through the outer annulus (which are used for the ignition of the precursor solution). A flame of high temperature is formed within which the precursors decompose and the organic compounds undergo complete combustion. The metallic compounds of the precursor are nucleated and primary particles of the oxides are formed and aggregated in the high temperature flame region. Additional sheath O2 is supplied through a sintered metal plate ring surrounding the outer annulus in order to assure the complete conversion of the reactants. The flow of the gases is controlled by mass flow controllers. Cooling of the nanopowder begins immediately away from the combustion zone by the natural flow of gases. The particles are collected in a filter in a stainless steel housing that is connected with the cone shaped burner unit with the aid of a pump. 2.3. Experimental conditions During FSP, the precursor solution was fed at constant flow rate of 5-10 ml/min through the capillary gap of the nozzle and was instantaneously dispersed by oxygen flow of 514 ml/min at standard conditions. Oxygen and not air is used as a dispersant in the nozzle as by using oxygen good dispersion and short droplet

lifetime are achieved [8]. The flow rate of the CH4 was 1-2 l/min and the flow rate of the carrier gas (O2) was set at 2-4 l/min. The oxygen sheath flow rate was 2-7 l/min. 2.4. Characterization of nanoparticles The specific surface area (SSA) of the as prepared nanoparticles was determined by nitrogen adsorption at 77 K using the BET technique in a Micrometrics Tristar II instrument. The phase composition of the nanoparticles was examined by X-ray diffraction (XRD) on a Bruker D8 Advance instrument (Cu Ka source, 20 kV, 5 mA). The diffraction was measured between 2θ angles of 10 and 65o with a step size of 0,1o. The morphology of the LTO powder was evaluated by Transmission Electron Microscopy (TEM) in a JEOL JEM1010 instrument. Images were taken at 90 KV. To prepare the sample, the powder was dispersed in a small quantity of ethanol, sonicated for 10 min. Then, one drop was placed on a formvar on copper support grid.

3. RESULTS AND DISCUSSION In the present study, the effect of the operating conditions of FSP on the nanoparticles’ size is investigated. As flame spray pyrolysis is a versatile technique, by changing the operating conditions, i.e. the liquid feed rate, the dispersion gas flow rate and the precursor concentration to name a few, the particle size and morphology is affected [9]. More precisely, the parameters that were changed were the precursor feed rate, the dispersion gas flow rate and the precursor concentration. Their effect on the size of the LTO nanoparticles is studied. Initially, the precursor feed rate has been varied between 5 and 10 ml/min while the other operating parameters have been kept constant. According to BET measurements, the size of

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 26

Furthermore, the dispersion gas flow rate has been varied between 5 and 14 ml/min leading to a decrease of the nanoparticles’ size from 29 nm to 18 nm. By increasing the dispersion gas flow rate, the droplet concentration of the spray flame is decreased along with the particle concentration and the particle residence time in the high temperature flame. Thus, the particle size is decreased. Finally, the effect of the precursor concentration on the particles size has been investigated. It has been concluded that by varying the precursor solution concentration between 0,5 and 0,9 M, the size of the nanoparticles has been varied between 27 and 29 nm. An increase in the metal concentration of the precursor solution provokes an increase to the particle concentration and to the residence time at high temperature. However, the variation of the precursor solution concentration does not affect as much the size of the nanoparticles as the variation of the precursor feed rate and the variation of the dispersion gas flow rate. A typical XRD spectrum of the LTO nanoparticles is presented in Figure 1 revealing that the main crystal structures of spinel Li4Ti5O12 (JCPDS Card No. 49-0207), rutile TiO2 (JCPDS Card No. 89-4920) and anatase TiO2 (JCPDS Card No. 86-1156) with peaks of Li0,57Ti0,86O2 (JCPDS Card No. 70-2365). A typical image of the LTO nanoparticles obtained by TEM is presented in Figure 2.

As it can be seen from the TEM image, the synthesized powder is inhomogeneous, i.e. it consists of large and small nanoparticles.

This fact can be attributed to the presence of isopropanol in the precursor solution. Li4Ti5O12 Li0,57Ti0,86O2 TiO2 (anatase) TiO2 (rutile) Intensity (a. u.)

nanoparticles varied between 20 and 29 nm. This fact can be attributed to the increase of the enthalpy content of the flame due to the increase of the precursor feed rate. Thus, the particle residence time at high temperature is increased along with the particles’ concentration in the flame. For this reason, by increasing the liquid feed rate, larger particles are synthesized as the particle sintering rate is enhanced due to an increased rate of particles’ collisions.

10

20

30

40

50

60

o

2θ ( )

Fig. 1. X-ray diffraction spectrum of LTO nanoparticles.

Fig. 2. TEM image of LTO nanoparticles.

4. CONCLUSIONS An investigation of the influence of the operating conditions of the FSP synthesis of LTO was carried out in a FSP apparatus consisting of a gas-assisted atomizer for the generation of the aerosol precursor. Lithium titanate nanoparticles have been produced and characterized. The particle size of LTO nanoparticles is critical for the application of the material as an anode for lithium ion batteries. By changing the operating conditions of FSP, nanoparticles of different sizes were obtained, revealing that the precursor solution concentration does not affect as much the size of the nanoparticles as the precursor feed rate and the dispersion gas flow rate. The phase

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 27

composition of LTO nanoparticles was investigated by XRD, revealing that the main crystal phases that appear are Li4Ti5O12, Li0.57Ti0.86O2, anatase TiO2 and rutile TiO2. The nanoparticles appear to be spherical when evaluating their morphology by TEM. However, their size distribution is not uniform, a fact that can be attributed to isopropanol used as a solvent in the precursor solution. ACKNOWLEDGEMENTS The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7-MC-ITN) under grant agreement No. 264710. The authors would like to thank the Directorate-General for Science, Research and Development of the European Commission for financial support of the research.

effects in the Li4+xTi5O12 spinel, J. Am. Chem. Soc., Volume 131, 2009, Pages 17786-17792. [7] A.C. Sutorik, S.S. Neo, D.R. Treadwell, R.M. Laine, Synthesis of ultrafine b(-alumina powders via flame spray pyrolysis of polymeric precursors, J. Am. Ceram. Soc., Volume 81, 1998, Pages 1477-1486. [8] G.L. Chiarello, I. Rossetti, L. Forni, Flamespray pyrolysis preparation of perovskites for methane catalytic combustion, Journal of Catalysis, Volume 236, 2005, Pages 251-261. [9] R. Mueller, L. Mädler, S.E. Pratsinis, Nanoparticle synthesis at high production rates by flame spray pyrolysis, Chemical engineering Science, Volume 58, 2003, Pages 1969-1976.

REFERENCES [1] J.H. Kim, Y.C. Kang, Electrochemical Properties of Nano-sized Li4Ti5O12 powders prepared by flame spray pyrolysis, Int. J. Electrochem. Sci., Volume 8, 2013, Pages 3379-3389. [2] B. Scrosati, J. Garche, Lithium batteries: Status, prospects and future, J. Power Sources, Volume 195, 2010, Pages 2419-2430. [3] C.H. Chen, J.T. Vaughey, A.N. Jansen, D.W. Dees, A.J. Kahaian, T. Goacher, M.M. Thackeray, Studies of Mg-Substituted Li4−xMgxTi5O12 Spinel Electrodes (0≤x≤1) for Lithium Batteries, J. Electrochem. Soc., Volume 148, 2001, Pages A102-A104. [4] S.E. Pratsinis, Flame aerosol synthesis of ceramic powders, Prog. Energy Combust., Volume 24, 1998, Pages 197-219. [5] T.J. Patey, R. Büchel, M. Nakayama, P. Novák, Electrochemistry of LiMn2O4 nanoparticles made by flame spray pyrolysis, Phys. Chem. Chem. Phys., Volume 11, 2009, Pages 3756-3761. [6] W.J.H. Borghols, M. Wagemaker, U. Lafont, E.M. Kelder, F.M. Mulder, Size

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 28

ELECTRICAL CONDUCTIVITY STUDY ON CNT PAPER FILMS Karthikeyan Gnanasekaran, Gijsbertus de With, and Heiner Friedrich Eindhoven University of Technology, PO Box 513, 5600 MB Eindhoven, The Netherlands ABSTRACT The conductive properties of carbon nanotubes (CNTs) depend on the arrangement of the individual CNTs forming the 3D network. Here we study the self-organization of CNTs of different size and size distribution forming dense 3D structures. Our experimental findings show a non-linear electrical conductivity behavior illustrating the influence of size and size distribution which alters the self-organizing behavior and results in difference in performance. In addition, the density variation by changing the size and size distribution behaves to be independent of the conductivity measured. The overall study illustrates the significance of the particle size distribution, thus the resulting conductivity of the system. Key words – Carbon nanotubes, network topology, packing density, size distribution.

1. INTRODUCTION Since the discovery of carbon nanotubes (CNT), they have been considered as a potential material in almost every application such as molecular electronics and electrical applications, catalyst supports, antistatic paints and many more [1, 2]. Applications such as electrochemical devices (anodes/cathode), catalytic support films require the use of CNT in the form of paper films. The exceptional physical properties of CNT such as electrical conductivity, stiffness, thermal conductivity are a result of its sp2- hybridized carbon lattice, rolled into a cylinder. In general, these physical properties are utilized in functional materials by homogenously dispersing the CNT in a polymer matrix. This increases the electrical conductivity of a conventional polymer, which are commonly insulating, by six orders of magnitude, incorporating only a very small quantity of carbon nanotubes [3]. In such conductive composites not only the CNT’s intrinsic properties influence

conductivity but more importantly the effectiveness of the CNT network formation. Theoretical and experimental studies show the possibility of the formation of the controlled self-organized network of rod- like nanofillers [4-7]. Although these reports were based on low volume fraction dilute system of polymer nanocomposites, here we study the selfassembly of dense CNT networks in the form of CNT paper films. This study is focuses on the influence of the CNT size and size distribution on the network topology and, thus, the resulting electrical conductivity. Here we try to optimize the macroscopic conductivity of CNT paper film by adjusting rod-like particle packing. 2. EXPERIMENTAL DETAILS Two different carbon nanotube systems namely M1 and M2 having the same diameter but different average lengths were chosen as model systems. Although both M1 and M2 are polydisperse, the polydispersity

MANANO Seminar, 4th September 2013, Kingston University, London, UK

Page 29

of M1 is negligible compared to the polydispersity of M2.

topology which results in a change in overall electrical conductivity [6, 8, 9].

Carbon nanotube dispersions were prepared by adding 1 mg of model material in 10 ml CHCl3 and sonicated for 1 hour. The paper films were prepared by vacuum filtration of the dispersion, or mixtures thereof, over a porous cellulose ester membrane. Any residual CHCl3 were removed by drying at room temperature for 24 hours before measuring electrical conductivity.

Many studies proved that increase in aspect ratio increases the electrical conductivity because of the formation of highly entangled networks with the optimum balance between the mean free path and the contact resistance [9-11]. Despite this anticipated linear behavior, our paper films showed a nonlinear effect by changing the composition as shown in Figure 1.

The DC conductivity was measured by standard four-point setup with parallel probes separated by 5 mm. The current was applied through the outer probes by Keithley 237 source measure unit and potential difference was measured between the inner probes by Keithley 6517A electrometer. Packing density variations on the paper films as a function of weight fraction was calculated from geometric variations on the volume occupied by the following equation:  = VCNT / Vocc where  is volume fraction, VCNT is volume of the CNT calculated from known mass and density (2.1 g/cm3), Vocc is the geometrical volume occupied by the paper films. 3. RESULTS AND DISCUSSION By mixing the dispersions of the model materials at different ratios, the CNT average aspect ratio and the CNT size distribution in the final dispersion are altered (either increased or decreased). This change in size and size distribution affects the overall packing density and the network

Fig. 1. Conductivity plotted as a function of weight fraction of M2.

In general, for random packing of rod-like particles, the volume fraction decreases exponentially with increasing aspect ratio. At low aspect ratios, the particles are more densely packed which results in a higher volume fraction. Increase in aspect ratio, the packing density decreases which in turn relates to the decrease in volume fraction [12]. The parameter that is controlled by mixing the model material dispersions at different volume fractions is the polydispersity of the CNT population (size distribution). Since the CNT diameter distribution in both the model materials is same, it is assumed that there is no influence

MANANO Seminar, 4th September 2013, Kingston University, London, UK

Page 30

with respect to diameter on polydispersity. Whereas, in the length distribution, large quantity of relatively monodisperse short CNT (M1) were replaced by lesser quantity of polydisperse long CNT when the weight fraction of M2 is increased from 0 to 1. Figure 2 shows that the porosity of CNT increases linearly with increase in average aspect ratio and polydispersity. This linearity in the volume fraction suggests that the influence of the weight average aspect ratio surpasses the polydisperse distribution on particle packing on our model materials because it is expected that polydisperse rods pack more densely than monodisperse rods with same average length. The linearity of porosity and the non-linearity of the conductivity as a function of weight fraction imply that the conductivity measured is independent to the density variations.

porosity measurement suggests that nonlinearity in conductivity measured is independent of density. This leads us that the change in size distribution or use of more than one sized model materials (e.g.: binary system, tertiary system ...) results in change on network topology and thereby results in the possibility of forming optimized conductive network. This study can be further validated by the electron microscopy analysis to understand the influence of size distribution on network formation. Despite the practical implications for conductive applications, the current findings intrigues the fundamental understanding of packing of rod-like nanofillers of different aspect ratios and modeling of macroscopic conductivity using nano/mesoscopic properties such as network topology. ACKNOWLEDGEMENS The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7-MCITN) under grant agreement No. 264710. The authors would like to thank the DirectorateGeneral for Science, Research and Development of the European Commission for financial support of the research. REFERENCES

Fig. 2. Porosity plotted as the function of CNT length distribution.

[1].

4. CONCLUSIONS

The non-linearity of conductivity plot as a function of weight fraction of M2 strongly indicates that the change in structural morphology can drastically change the electrical conductivity. The linearity of the

[2].

A. Hirsch, The era of carbon allotropes, Nature Materials, Volume 9, Issue 11, 2010, Pages 868-871. D. Iannazzo, et al., CoumarinConjugated Multiwalled Carbon Nanotubes for Potential Biological Applications: Development and Characterization. Journal of

MANANO Seminar, 4th September 2013, Kingston University, London, UK

Page 31

[3].

[4].

[5].

[6].

[7].

[8].

[9].

[10].

Nanoscience and Nanotechnology, Volume 12, Issue 6, 2012, Pages 5030-5038. M. Pasquali, Polymer composites: Swell properties and swift processing, Nature Materials, Volume 3, Issue 8, 2004, Pages 509510. R. H. J Otten and P. van der Schoot, Connectivity percolation of polydisperse anisotropic nanofillers, The Journal of Chemical Physics, Volume 134, Issue 9, 2011, Pages 094902-15. R. H. J Otten and P. van der Schoot, Continuum Percolation of Polydisperse Nanofillers, Physical Review Letters, Volume 103, Issue 22 2009, Pages 2257041-4 A. V. Kyrylyuk, et al., Controlling electrical percolation in multicomponent carbon nanotube dispersions, Nature nanotechnology, Volume 6, Issue 6, 2011, Pages 364369. M. C. Hermant, et al., Lowering the percolation threshold of singlewalled carbon nanotubes using polystyrene/poly(3,4ethylenedioxythiophene): poly(styrene sulfonate) blends, Soft Matter, Volume 5, Isuue 4, 2009, Pages 878-885. X. Shui and D.D.L. Chung, Electrical resistivity of submicrondiameter carbon-filament compacts, Carbon, Volume 39, Issue 11, 2001, Pages 1717-1722. C. Berger, et al., Multiwalled carbon nanotubes are ballistic conductors at room temperature, Applied Physics A, Volume 74, Issue 3, 2002, Pages 363-365. M. O. Lisunova, et al., Percolation behaviour of ultrahigh molecular weight polyethylene/multi-walled

[11].

[12].

carbon nanotubes composites, European Polymer Journal, Volume 43, Issue 3, 2007 Pages 949-958. K. Shehzad, et al., Effects of carbon nanotubes aspect ratio on the qualitative and quantitative aspects of frequency response of electrical conductivity and dielectric permittivity in the carbon nanotube/polymer composites, Carbon, Volume 54, 2013 Pages 105-112. A. Wouterse, S. Luding, and A.P. Philipse, On contact numbers in random rod packings, Granular Matter, Volume 11, Issue 3, 2009, Pages 169-177.

MANANO Seminar, 4th September 2013, Kingston University, London, UK

Page 32

EFFECT OF GLASS COMPOSITION ON PROPERTIES OF NOVEL NANO-STRUCTURED GLASS-POLYMER HYBRIDS Luciana Serio, Yuqing Bao and David Thomas Gawne London Southbank University, 103 Borough Road London SE10AA, UK ABSTRACT Phosphate glass-polymer hybrids are an emerging class of nanomaterial with unique characteristics derived from nano-micro interactions helped by both components being fluid during processing. This condition allows the control of hybrid morphologies and lowesr the high viscosity inherent in solid fillers. This research investigates the effect of glass composition and glass content on the nano/microstructure and thermo-chemical properties of these new materials. Four new glasses were developed and selected for the preparation of glass-polyamide 11 (PA11) hybrids by melt-blending. The melt-blending trials were carried out using a Micro-DSM extruder. The experimental results show that the melting point of the hybrids decreased compared to that of pure PA11 in all glass-PA11 hybrids. In particular, the interaction parameter for GlassD-PA11 hybrids, measured according to the Flory-Huggins solution theory, was equal to -0.0111. This result suggested that the glass particle size decreased with increasing glass content and revealed an interaction between GlassD and polyamide 11. GlassD showed the most homogeneous particle dispersion in PA11 matrix compared with the other phosphate glasses. Keywords – hybrids, melting point depression, morphology, PA11, poly-phosphate glasses.

1. INTRODUCTION Despite their popularity, glass polymer composites have a few shortcomings due to the present of glass as a solid phase and limited chemical interaction among the components: the high viscosity which becomes intractable with increasing glass amount in the composite, the poor surface finish, the difficulty in orientating glass fibers, the presence of micro-defects at the glass-polymer interface [1]. To overcome these problems a new class of materials, inorganic/organic hybrids, has been developed in the last 20 years [2-8] by blending new low-Tg phosphate glasses with polymeric materials. Tick (1984) presents a specific glass that has a molar composition of 50% SnF2+ 20% SnO + 30% P2O5, a Tg of approximately 125 °C and a density of 3.75 g/c3 [9]. The peculiar characteristics of this glass allow it to be fluid in the range of the processing temperatures of the majority of the thermoplastic matrices. This condition results in the possibility of controlling the nano-molecular scale interactions taking place between the hybrid components and

of modeling hybrid morphologies, obtaining significant improvements in properties, compared to those achieved from classical polymer blends and composites [2-9]. Because of the low softening point and low Tg of the glass in question, these hybrids can be processed conventionally with glass loadings of up to 60% by volume, managing to avoid the intractable viscosity problem inherent to a too high filler concentration (up to 30% by volume) [2-9]. In the majority of the recent studies the chosen matrixes were polyamide (PA) [2, 4, 7, 8], "higly interacting commodity resins which are shown [have been shown] to have a high degree of physiochemical interaction with the low Tg phosphate glass. In fact Hersh, Onyiriuka and Hertl (1995) have shown that ammonia will adsorb to the surface of phosphate glasses [9], condition that could improve the miscibility in the liquid state and the adhesion in the solid state of the hybrid components. Even though a few studies are available in literature, in none of them Rilsan PA11 has been used. Rilsan PA11 is a new and innovative type of PA produced by Arkema (France) for more

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 33

than fifty years. This PA is manufactured from a renewable source (castor oil) and is used in a large number of applications (automotive, transport, textile, oil and gas, wire & cables, electronics) thanks to its outstanding properties: high dimensional stability, low density, excellent resistance to hydrocarbons, ease of processing, a wide range of working temperatures (-40°C / +130°C). In addition, the glass composition has never been modified in the previous studies and the effect of glass composition on this type of hybrids has not been well investigated yet. The goal of this project is to develop new phosphate glass/PA11 hybrids by melt-blending processes and characterized by high miscibility among the components and with enhanced properties. In particular, the major goal of this study is the investigation of the effect of low Tg phosphate glass composition on the structure and properties of hybrid materials.

2. EXPERIMENTAL DETAILS 2.1.Materials 4 different glasses were prepared in small (10-20 g) batches of the NH4H2PO4 reagent-grade oxide powder, SnO and SnF2 provided by SigmaAldrich Co. Compositions of P2O5/Sn0 mol% =1.5 and SnF2 mol% in a range 30-60% were used. The glass compositions are listed in Table 1. All chemicals were mixed and heated together at 450°C for 25 minutes, annealed for 30 minutes and then cooled to room temperature. Table 1: Formulation for Tin-Flouride-Phopshate (TFP) glasses Glass name

Molar composition (mol%)

Density (g/cm3)

Tg (°C)

P2O5

SnO

SnF2

GlassA

42

28

30

3.45 ± 0.03

155 ± 4.1

GlassB

36

24

40

3.50 ± 0.07

149 ± 2.9

GlassC

30

20

50

3.72 ± 0.11

127.5 ± 3.3

GlassD

24

16

60

3.92 ± 0.04

106 ± 2.5

The DSC results showed that Tg decreases in a range 155-106 ℃ with increasing SnF2 mol%. Furthermore the Tg of GlassC, having composition of 30P2O5 + 20SnO + 50SnF2, appeared to be similar to the results available in literature [2-9]. The glass degradation occurred in the temperature range 350-500 °C. Polyamide 11 RILSAN (PA11), provided by Arkema Company (France) was used in combination as matrix thermoplastic for phosphate glass. The PA 11 and the glasses were used to produce a series of hybrids with nominal total glass contents in the range of 0-20% vol. Compounding: The glasses were first ground manually and sieved at <200µm. After mixing the selected glass with PA 11 powder, the two components were mechanically blended with a roller and compounded with a Micro-DSM extruder at an average screw speed of 100 rpm to produce a homogeneous dispersion of glass powder throughout the matrix. The extruder temperature profile was 210/230/230 ℃. Volume of the sample was kept constant and equal to 10 cm3. The residence time was 50 s. The Force/time data generating by the extruder were extrapolated for each experiment. Each hybrid composition was tested three times.

2.2.Hybrid Testing Equilibrium Melting Point and Melting Point Depression Technique: the samples, which meltcrystalized at different isotherms (from 160 to 170 °C), were studied by a Perkin-Elmer DSC-7 differential scanning calorimeter over a wide temperature range at a heating rate of 10 K/min. The weight of the samples was typically 8-10 mg and the samples were encapsulated in hermetically sealed aluminum pans. Morphological Microstructures Analysis: The cross-sections of the PA11 nano-composites were investigated by SEM to examine the dispersion of low Tg glass within the PA11 polymer matrix.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 34

3. RESULTS AND DISCUSSION 3.1.Equilibrium Melting Point and Melting Point Depression Technique The equilibrium melting point (Tm0), defined as a melting temperature (Tm) of an ideal crystal having infinite crystal size, is a fundamental physical property of polymers. A drop in Tm0 is often attributable to thermodynamically favorable interactions between the polymers. Furthermore Tm0 is important in study of crystallization phenomena, such as lateral growth rate and nucleation rate. The Hoffman–Weeks approach [2-8] was used to determinate the equilibrium melting points of the pure PA11 and hybrids. In this method, the observed melting point (Tm) is plotted versus the crystallization temperature (Tc), resulting in a linear plot. The equilibrium melting point of the material Tm0 is the intersection point of this plot with the line Tc=Tm. Typical Hoffman– Weeks plots were obtained for PA11 and the hybrids are shown in Fig 1 and Fig 2 and reported in Table 2. The equilibrium melting point Tm0 of neat PA11was equal to 200.5 °C and in close agreement to the value reported in literature [1011]. The addition of 20 vol% glass to the organic matrix leads to a decrease in the Tm0 of the resulting hybrid compared to PA11, for all types of glasses. It can be hypothesized that some sort of interaction between the low Tg glasses and PA11 occurred. There was not a clear trend of Tm0 decreasing among different types of glass/PA11 hybrids. However, the highest and the lowest Tm0 were observed in GlassB/PA11 hybrid and GlassD/PA11 hybrid respectively. In particular, Tm0 decreases significantly from 200.5°C (PA11) to 186.5°C in GlassD/PA11, indicating that interaction between Glass D (the glass having the lowest Tg and highest ?? please check SnF2 content) and PA11 could be improved more than the interactions between PA11 and the other glasses. To further study the peculiar behaviour of GlassD, different GlassD/PA11 hybrids were prepared by changing the glass content in the

hybrids. The equilibrium melting points Tm0s calculated with Hoffman-Weeks approach are shown in Figure number ? and summarized in Table 2. Tm0 decreased with increasing glass content in the hybrids (Fig 2), strengthening the possibility of a thermodynamically favorable interaction between GlassD and PA11. The Flory-Huggins solution theory was used to calculate the interaction parameter between the organic matrix and the GlassD. Flory-Huggins solution theory is a mathematical model of the thermodynamics of polymer solutions successfully used [2, 8, 12] to describe the interaction between polymers under thermodynamic equilibrium, as shown in Eq. 1 [2, 8, 12], equation that can be reduced to Eq. 2:

∆

Χ 1

(1)

∆

Χ

(2)

where:  Tm0 and Tmb0 are the equilibrium melting points of the crystalline polymer and the blend or hybrid, respectively;  Va and Vc are the molar volumes of the repeat unit of the amorphous and the crystalline polymers, respectively;  is the volume fraction of the specified component;  R is the universal gas constant;  is the heat of fusion of the crystalline component;  is the Flory-Huggins interaction In the case of Pglass/polyamide hybrids, the Pglass was taken as the amorphous polymer as it has been reported elsewhere [2, 8]. The repeat unit of the Pglass first proposed by Tick[9] was used to calculate the molar volume (Va) of the TFPglasses. The heat of fusion value of the PA11 used in the present calculations was 230 kJ/mol.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 35

nylon GlassC/PA11

GlassA/PA11 GlassD/PA11

0.0111, satisfying the condition for possible miscibility between GlassD and PA11. The same procedure was used in [2, 8] for GlassC/PA6 hybrids. In this study, known as represents the first reported evidence of miscibility of inorganic glass and organic polymer, the interaction parameter (χ = –0.067) resulted to be more negative than the parameter in GlassD/PA11 hybrids system. Thus the interactions between PA6 and GlassC could be higher than interactions between PA11 and GlassD system.

GlassB/PA11

205

200

Tm [°C]

195

190

185

180

Table 2: Equilibrium melting point of PA11 and Glass/PA11 hybrids determined by HoffmanWeeks approach.

175 155

165

175

185

195

205

Tc [°C]

 

Fig 1: Equilibrium melting point of different glass/PA11 hybrids with 20 vol% glass.

0%

2.50%

5%

10%

20%

205

200

Blend

PA11 (vol%)

Glass (vol%)

Tm0 [°C]

St. Dev.

PA11

100

0

200.5

± 2.5

GlassA/PA11

80

20

187.5

± 1.6

GlassB/PA11

80

20

193

± 1.3

GlassC/PA11

80

20

188.8

± 1.2

GlassD/PA11

80

20

186.5

± 0.7

10GlassD/PA11

90

10

188.3

± 0.2

5GlassD/PA11

95

5

189

± 0.6

2.5GlassD/PA11

97.5

2.5

189.2

± 1.3

190

3.85E‐04

185

3.65E‐04

180

175 145

155

165

175

185

195

205

Tc [°C] Fig 2: Effect of glass content on the Equilibrium melting point of GlassD/PA11 hybrid.

Plotting the melting-point depression (left-handside of Eq. 2 versus the volume fraction of the amorphous component squared should result in a straight line that passes through the origin. The slope of this line is 0.0019 and the fit of the linear line gives an R2 value equal to 0.99. The interaction parameter χ measured according to the Flory-Huggins solution theory was equal to -

1/Tmb0 ‐ 1/Tm0 (°C‐1)

Tm [°C]

195

3.45E‐04

y = 0.0019x + 0.0003 R² = 0.99

3.25E‐04

3.05E‐04

2.85E‐04

2.65E‐04 0

0.01

0.02

0.03

0.04

0.05

Φ2 Fig 3: Melting point depression of the TPF glass/PA11 hybrid as a function of glass content.

3.2.Hybrid Micro-sructure The uniformity of the dispersion of the glass particles in the PA11 matrix of GlassA/PA11,

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 36

GlassB/PA11 GlassC/PA11 and GlassD/PA11 hybrids with 20 vol% glass was examined through SEM. The corresponding micrographs are as shown in Fig 4 (a, b, c, d). All hybrids showed the PA11 matrix with a secondary phase of glass particles with a concentration of 17–20 vol%, proving the immiscibility between PA11 and glasses. However the hybrids exhibit very dissimilar morphology. GlassB/PA11 is the hybrid characterized by the most heterogeneous glass particle distribution. Some defects (like circular cracks around the biggest particles) were observed. This demonstrates the lack of interaction between GlassB and PA11, confirming the results from the rheology measurements and thermal analysis. This hybrid exhibited the highest equilibrium melting point, in line with a more heterogeneous morphology and bigger glass particle size, compared to the other hybrids. Except for GlassB/PA11, the particle dispersion and particle size homogeneity seem to improve as the SnF2 is added to the glass. The hybrid containing GlassA (being composed of 30 mol% SnF2), exhibits a better particle dispersion and particle size homogeneity than GlassB/PA11 but a worse one compared to hybrids reinforced with GlassC (composed of 50% SnF2) and GlassD (composed of 60% SnF2). Furthermore the particle size decreases with increasing mol% SnF2 in the glass. This phenomenon is due to the decrease in the glass Tg and glass softening point with increasing mol% SnF2 added to the glass compositions. As their Tg and softening point drop, TPF glasses could reach similar value to the PA11melting point (≈190°C), allowing both hybrid components being fluid during processing. This condition permits to control and tailor the interaction and morphologies among the components and to achieve a better dispersion of the filler phase in the matrix. PA11 hybrid composed of GlassA and GlassC showed big glass particles of 50-150 µm in diameter. The dimension of this glass spots is lower in GlassC/PA11. Both hybrids show rounded glass nanoparticles of 50-100 nm in diameter.

(a)

(b)

(c)

(d) Fig 4: SEM micrographs of PA11 filled with 20% of: GlassA (a); GlassB (b); GlassC (c); GlassD (d).

Pores, of 2-40 µm in diameters are visible in both GlassA/PA11 and GlassC/PA11 hybrid, demonstrating the lack of glass/matrix interaction along a few places of the crosssection. The pores are smaller and less common in GlassC/PA11. GlassD/PA11 hybrid shows the best morphology among the hybrids. The micro-glass spots present in the others are no more visible. The

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 37

cross section area exhibits nano-micro particles separated from each other and with uniform distribution. Furthermore the glass particles are not perfectly rounded but seem to be slightly deformed, which demonstrates that GlassD has a similar softening point to the PA melting point. Pores along the cross-section are no longer present confirming the better GlassD-PA11 interaction compared to the other hybrids. These results are in line with the depression of the hybrid melting point. In fact GlassD/PA11 hybrid was the material exhibiting the lowest equilibrium melting point, result explained by the lowest particle size compared to the other hybrids. 4. CONCLUSIONS Effect of glass composition, on miscibility and morphology of PA11 hybrid reinforced with different tin-fluoride-phosphate glasses were studied. The observations made are described below. The equilibrium melting point Tm0 of PA11 was decreased by the ?? of TFP glasses in all hybrids. The lowest Tm0 were shown in PA11 hybrid containing the 20 vol% of GlassD, the glass having the lowest Tg and the ??SnF2 content: Tm0 decreased significantly from 200.5°C (PA11) to 186.5°C (GlassD/PA11). It can be inferred that the ?? of SnF2 could improve the interaction between the glass and PA11. The thermodynamically favorable interaction between GlassD and PA11 was confirmed by the Tm0 decrease with increasing glass content in the GlassD/PA11 hybrids. The interaction parameter between GlassD and PA11, calculated by the Flory-Huggins solution theory, was equal to 0.0111 suggesting the possible tendency for miscibility between the two hybrid components. Except for GlassB/PA11, the hybrid showing the worst morphology, the glass particle dispersion improved and particle size decreased with increasing mol% SnF2 in the glass. GlassD/PA11 hybrid showed the best morphology among the hybrids: micro-nano particles and no pores were present. The bonding between glass and matrix seemed to be improved. The presence of slightly deformed glass particles confirmed that GlassD had a similar softening point to the PA11 melting point.

The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7-MC-ITN) under grant agreement No. 264710. The authors would like to thank the Directorate-General for Science, Research and Development of the European Commission for financial support of the research. REFERENCES [1] S.L. Gao, E., R. Acta Materialia, Nanostructured coatings of glass fibers: Improvement of alkali resistance and mechanical properties, Volume 55, Issue 3, February 2007, Pages 1043–1052. [2] Urman K, Otaigbe JU. Novel phosphate glass/polyamide 6 hybrids: miscibility, crystallization kinetics, and mechanical properties. J Polym Sci B 2006;44:441–50. [3] Adalja SB, Otaigbe JU. Creep and recovery behavior of novel organic–inorganic polymer hybrids. Polym Compos 2002;23:171–81. [4] Urman K, Iverson D, Otaigbe JU. A study of the effects of processing conditions on the structure and properties of phosphate glass/polyamide 12 hybrid materials. J Appl Polym Sci 2006;105:1297–308. [5] Urman K, Schweizer T, Otaigbe JU. Uniaxial elongation flow effects and morphology development in LDPE/phosphate glass hybrids. Rheol Acta 2007;46:989–1001. [6] Otaigbe JU, Quinn CJ, Beall GH. Processability and properties of novel glass-polymer melt blends. Polymer Composites 1998;19:18–22. [7] Urman K, Otaigbe JU. Novel phosphate glass/polyamide 6 hybrids: miscibility, crystallization kinetics, and mechanical properties. J Polym Sci B 2006;44:441–50. [8] K. Urman J. U. Otaigbe, New phosphate glass/polymer hybrids-Current status and future prospects. Prog. Polym. Sci. 32 (2007) 1462-1498 [9] Tick, P. A. Phys Chem Glas. 1984, 25, 149-154. Hersh LS, Onyiriuka EC, Hertl W. Amine-reactive surface chemistry of zinc phosphate glasses. J Mater Res 1995;10:2120–7. [] Liu, S., Yu, Y., Zhang, H. and Mo, Z. (1998), Isothermal and noisothermal crystallization kinetics of nylon 11. J. Appl. Polym. Sci., 70: 2371-2380. [] Wang, B., Ding, Z and Hu, G. (2008), Melting behaviour and isothermal crystallization kinetics of nylon11/EVOH/dicumyl peroxide blends. Polym Eng Sci, 48:2354-2361. [Deimede VA, Fragou KV, Koulori EG, Kallitsis JK, Voyiatzis GA. Miscibility behaviour of polyamide11/sulfonated polysulfone blends using thermal and spectroscopic techniques. Polymer 2000; 41:9095-101.

ACKNOWLEDGEMENTS

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 38

Toughening of epoxy matrix fiber reinforced composites with acrylonitrile butadiene and carboxylated acrylonitrile butadiene rubber for Advanced Industrial Applications Nazli Ozdemir, Tao Zhang Kingston University London, SW15 3DW, United Kingdom ABSTRACT Toughening of diglycidyl ether of Bisphenol A, dicyandiamide matrices with carboxylated acrylonitrile butadiene rubber and acrylonitrile butadiene rubber has been accomplished using different processing techniques such as triple roll mill, high shear mixer and ultrasonicator and the rheological, cure, mechanical and morphological characteristics of both systems have been analyzed. Mechanical properties of both toughened matrices and the carbon fiber composites with the toughened matrices have been studied. SEM images show that CNBR nanorubber(CNBR-NP) has been dispersed evenly within the matrix whereas high amount of agglomeration within NBR nanorubber(NBR-NP) system caused plasticization of the matrix ending up in lower glass transition values and deterioration of some of the properties such as interlaminar shear strength, lap shear strength and activation energy. Due to achievement of even and nano level dispersion within the CNBR-NP system, Tg stayed the same and both fracture toughness and peel strength values have been improved considerably. Difference in dispersion of nanorubber has been related to the difference of Van der Vaals forces between single particles and the polarity of the systems. Key words – acrylonitrile butadiene rubber, carboxylated acrylonitrile butadiene rubber, nanoparticle, toughening, epoxy resin

1. INTRODUCTION Epoxy resins are widely used in many different applications ranging from coatings, structural adhesives and as a matrix material for advanced composite materials in automotive and aerospace industries due to their outstanding thermal and mechanical properties, their high resistance to creep and their chemical stability. However, they have their drawbacks such as low fracture toughness, their notch sensitivity and poor resistance to crack propagation due to their brittle nature. DGEBA based epoxy resins cured with dicyandiamide in the presence of an uron accelerator have been widely used in industry, especially in the preparation of prepregs and when toughened with different particulates as curable structural adhesives between metallic plates [1,2,3,4,5].

The knowledge of rheological and curing behaviour of the resin systems is necessary to have a correct control of the process mechanism and a subsequently engineering design. In regards to the brittle nature of highly crosslinked epoxy resins, scientists have been toughening the formulations with nano and micro sized particulates for many years. It is shown by many studies that toughness and impact resistance can be improved by achieving an even dispersion of elastomeric particles within the matrix [6,7,8]. However, in order to achieve this, plasticization of the matrix due to soluble rubber particles with the epoxy matrix should be prevented. This is called flexibilisation of the matrix in which case the mechanical damage is reduced through lowering modulus, in other words through plasticization. This way stress is relieved through distortion of the material. Toughening with carboxyl-terminated butadiene

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 39

acrylonitrile and hydroxyl terminated butadiene acrylonitrile have been very common for 30 years since today [9,10]. Although there are many studies on toughening of DGEBA/DICY formulations in purpose of improving the mechanical properties; there is no study on nano-acrylonitrile butadiene rubber toughening of them. This is due to the non-stoichiometric and complex nature of the DICY curing agent, which makes the final properties uncontrollable. Solid acrylonitrile butadiene rubbers (NBR), with high content of acrylonitrile are suitable tougheners. This is due to the high content of acrylonitrile imparting better compatibility between NBR and the epoxy resin [11]. This study constitutes a complete work on rheological, cure, mechanical and morphological characterization of two different nanorubber toughened systems. Main aim of the work is as listed below: 1. Optimize the processing steps within both systems in order to achieve a good dispersion of nanorubber. 2. Improve the mechanical properties such as fracture toughness and impact resistance of epoxy resin when keeping other properties constant and later commercialize the toughened matrix.

3. Study the toughening mechanisms of the matrix.

1. EXPERIMENTAL DETAILS 1.1. Materials Nanorubber, acrylonitrile butadiene rubber (NBR-NP) and carboxylated acrylonitrile butadiene rubber (CNBR-NP) with trade name VP-501 were received in powder form, mainly in agglomerated cluster form at diamaters of 100 µms. Fumed silica received from Aerosil with 1 µm diameter was used in some of the formulations to modify the rheological

behavior but not necessarily in all of the formulations. Fumed silica when dispersed in resin matrix increases viscosity considerably preventing resin leakage during curing of laminates in autoclave under pressure. Due to this reason, in some of the formulations fumed silica has been used. The epoxy resin (Araldite LY1556, diglycidyl ether of bisphenol-A with epoxide equivalent weight=188, DICY curing agent (Dyhard D50EP, a masterbatch of micronized DICY with 49-51% content, particle size<10 micron in a liquid epoxy resin with content of epoxy groups=5.1-5.5), and the accelerator 2:1 blend of LY1556-3, 3’-(4Methyl-1, 3-phenylene) bis (1, 1dimethylurea), Dyhard UR500 were used. Both the curing agent and the accelerator are blends of epoxy resin in order to enable an effective mixing. 200 GSM bidirectional carbon plies have been used for processing of composite laminates with the nanorubber toughened matrix. 1.1. Sample Preparation Even distribution of nanorubber within the matrix is crucial in order to get the most out of the mechanical properties. For this, each step of processing was optimized carefully. Processing methods producing controlled particle size distribution, dispersion, and interfacial interactions would be advisable to obtain customized properties in new composites. In this research methods to improve the dispersion of nanoparticles into the matrix include mechanical mixing, ultrasonic dispersion, high shear mixing and triple milling but no surface modification. The processing steps are listed as below and given in Figure 1: 1. Leave the nanorubber at 70°C for 16 hours; 90°C is the softening temperature of the nanorubber. 2. Disperse dried nanorubber in DGEBA matrix and speed mix the sample. If fumed silica is to be used, disperse both tougheners together into the matrix.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 40

3. Triple mill the formulation for 7 times at RT, fix the gap size by trial and error technique. After 3rd times milling speed mix the nanofluid for 6 minutes at 2000 rpm. (b)

4. Other processing technique (which is more suitable for micron sized particles): High speed mixing + Ultrasonication for 1 day; at this point the sample should be watched carefully in order to prevent excessive heating which softens the nanoparticles inside. 5. Put a magnetic bar inside the nanofluid and degas the formulation at 70°C when the magnetic bar is rotating at a speed of 320rpm. Leave it overnight.

(c)

(d)

(e)

6. Add the hardener and the accelerator and speed mix the formulation for 6 minutes. (f)

7. For processing of the laminates, hand layup technique is used and the stacked plies have been vacuum bagged in an autoclave at 30 psi with the cure cycle of 0.5ºC/min till 120 ºC, 1hr@ 120 ºC, 0.5ºC/min till RT. For processing of the resin casts, toughened and degassed resin has been poured into PTFE moulds and have been cured in an oven at the same cure cycle. Specimens for mechanical testing have been cut using a wet sour.

Figure 1(a) Triple Milling, (b) 3 times triple milled sample almost translucent, (c) Vacuum degassing with rotation of a magnetic bar inside the sample, (d) Degassed sample with bubbles united at the surface, (e) High shear mixing and ultrasonication of the blend, (f) High shear mixer probe embedded inside the sample, the pot inside the ultrasonicator cell.

2. RESULTS AND DISCUSSION The experimental data collected and the comparison between the two nanoparticle systems are as listed below in Table 1.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 41

Table 1. Comparison of Nanorubber Systems NP

VP401

VP501

Particle Size Analyzer

~100µm in powder form

~100µm in powder form

SEM, inside resin

~1 µm agglomerates

<100nm nanoparticles

Spot Viscosity

~30 Pas when 10 phr rubber is dispersed

~44 Pas when 10phr rubber is dispersed

Activation Energy (Kj/mole)

Increasing with nanorubber addition

Decreasing with nanorubber addition

Tpeak (°C)

138°C for 10RHI, @10°C/min

143°C for 10R14HI, @10°C/min

Tg (°C)

Decreases ~7°C with 20phr nanorubber addition

No change

ILSS (MPa)

Decreases from 71 to 46 MPa with 10phr addition

-

G1c (J/m²), laminate

X2 the value with 10phr rubber addition

X2 the value with 10phr rubber addition

-

Considerable increase in peel force with rubber addition, FW

Climbing Drum Peel Test

Tpeak: Cure Peak Temperature Tg: Glass Transition Temperature (ºC) ILSS: Interlaminar Shear Strength (MPa) SENB: Single Edge Notch Beam Test Data G1c: Double Cantilever Beam, Mode I Fracture Toughness LSS: Lap Shear Strength

In powder form both NBR-NP and CNBR-NP have almost the same size, 100 µms.

The particle diameters were found using Mastersizer 3000 Particle Size Analyzer. However when dispersed in the resin, using different dispersion techniques such as triple mill, high shear mixing, ultrasonication and degassing; the rubber agglomerates and clusters can be broken down to even smaller diamaters, nano dimension for CNBR-NP (~100nms) and micro dimension for NBR-NP (1 µm). Fracture surfaces of toughened and

untoughened specimens have been analyzed using scanning electron microscopy at secondary electron mode. The samples were vacuum coated with gold using a sputter coater. All images were taken using an accelerating voltage of 20-25 KeV with a magnification between 90 times and 2000 times. Achievement in even dispersion within VP501 system may be related to the difference in the polarity of the two nanorubbers, their chemical formulations and the Van Der Waals forces between the molecules. In a way the carboxyl group of CNBR-NP may enable even better dispersion within DGEBA matrix. The rheological behaviour of both nanofluids with nanoparticles dispersed at an increment of 5 phr till 20 phr within the matrix was studied at fixed temperatures in order to have an understanding of the effect of NBR-NP and CNBR-NP on the rheological characteristics of DGEBA resin. The rheological behavior of the CNBR-NP system eased the processing of laminates. Due to the higher viscosity of the nanoresin system, leakage was observed less compared to NBR-NP system. CNBR-NP system did not necessarily require usage of fumed silica in order to prevent resin leakage ending up in dry spots, which have a considerable effect on mechanical properties. During curing when the system reaches higher temperatures around 80°C, the viscosity of the system decreases considerably and the resin flows leaving dry areas within the laminates. Due to this issue within NBR-NP batch

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 42

formulations the composites had dry spots and this problem has been solved by usage of fumed silica. Fumed silica when dispersed within the matrix has an incredible thickening effect even with 1 phr addition and prevents resin leakage. Activation energies of the samples were calculated with Flynn Wall Ozawa and Kissinger technique. For this, formulations with 5, 10, 15 and 20 phr NBR-NP and CNBR-NP addition were prepared and heated at heating rates of 5, 10, 15 and 20ºC/min from RT to 300°C and the peak temperatures were recorded. Activation energy, in other words the energy required to initiate the crosslinking reaction decreased for about 4kJ/mole with CNBR-NP dispersion within the matrix whereas it increased for about 5kJ/mole with NBR-NP dispersion. This difference may be related to the dispersion difference between the systems. As can be seen from SEM images given in Figures 2&3; nano dispersion can be achieved evenly within CNBR-NP system whereas only micro dispersion is achieved within NBR-NP system. The agglomerates with 50µms diameter within NBR-NP formulation cause uneven crosslinking and deteriorate the cure properties. Peak temperature of the CNBR-NP system is almost 5°C higher than NBR-NP system which can be attributed to the creation of higher amounts of surface area within this system holding more crosslinking ingredients and in a way delaying the reaction. Glass transition measurement has been conducted with the facilities DSC, TMA and DMA. There is a 7°C decrease in the glass transition temperature with 20 phr NBR-NP addition to the matrix whereas the glass transition temperature does not change with CNBR-NP addition. This interesting result may be attributed to the even structure formation within CNBR-NP batch samples whereas agglomeration is observed within

NBR-NP samples causing a weaker structure. Agglomerates distort the stoichiometric balance resulting in an incomplete cure. The ILSS test was conducted in accordance with ASTM D2344 at a crosshead speed of 1.3m/min. All specimens were supported in a fixture and loaded at mid span. The ILSS was calculated as follows: ILSS = 0.75F/b*d, where F represents the breaking load, b the width of specimen and d the thickness of specimen. Interlaminar shear strength of NBRNP batch decreased for about 25 MPa with 10phr rubber addition.

Figure 2. SEM images of high shear mixed 10phr NBRNP containing resin system, secondary electron mode

(a)

(b) (b)

(c)

Figure 3. SEM images of 10VP501/ R14HI Triple Milled, secondary electron mode

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 43

DCB testing was performed in the perpendicular direction to the length under a constant speed of 10mm/min using a Zwick Z250 testing machine. In Mode I delamination, the delamination surfaces separate perpendicularly to the surface of plane of delamination. Table 2. Shows that there is a huge increase in G1c values with nanorubber addition. VP401 and VP501 have almost the same toughening effect on the matrix. Table 2. G1c Test data

compared to agglomerated NBR-NP rubber. 2. G1c no matter what type of nanorubber has been used or how evenly they have been distributed has been improved the same with same amount of different rubber addition. 3. Peel force increased considerably with nanorubber load level, which enhances the idea that CNBR-NP systems could be used as an adhesive in commercial applications. 4. CNBR-NP, due to nano size dispersion has not deteriorated any material properties such as Tg and the activation energy and greatly improved the fracture toughness and peel strength.

Sample

G1c J/m²

Processing

Untoughened

392

Speed mix

10phr NBR/NP

814

High Shear Mixing & Ultrasonication

10phr CNBR/NP

836

Triple Mill

ACKNOWLEDGEMENTS

20phr CNBR/NP

1360

Triple Mill

The authors would like to thank Cytec Engineering Materials for supplying the chemicals and 7th Framework Program for financial support.

Climbing drum peel test is used to compare adhesion between flexible and rigid adherends, or between flexible facing of a sandwich structure and its core. Specimens are placed in the climbing drum peel apparatus. The grips of a Universal Test machine are initiated at a specified grip separation significant enough to roll the drum upward. The standard speed is 25.40 ± 2.54 mm/min [1± 0.10 in/min]. The test data gave an idea that there was a considerable increase in the peel strength values with nanorubber addition.

3. CONCLUSIONS For two different rubber modified systems, rheological, cure, mechanical and morphological characteristics have been examined. The following conclusions were obtained. 1. CNBR-NP due to lower Van Der Waals forces among the molecules has been distributed inside DGEBA system more evenly and in nano dimensions

REFERENCES

[1] Alan G. McKown, Particulate Adhesive Containing Polyepoxides, Carboxylated ButadieneAcrylonitrile Copolymer and a Urea Derivative as a Curing Agent, United States Patent Office 3, (1972), Pages 655-818 [2] Shinn-Gwo Hong, Journal of Polymer Research 12, (2005), Pages 295-303 [3] Qian Wang, Journal of Applied Polymer Science, Vol 87, (2003), Pages 2295-2305 [4] Shinn-Gwo Hong, Thermochimica Acta 417 (2004), Pages 99-106 [5] Dalip Kohli, US 2011/0048637(2011), A1 [6] Guicun Qi, Polymer Chem., (2011), 2, 1271 [7 ] Huang Fan, Science in China Ser. B Chemistry (2005) Vol. 48 No.2 148-155 [8] Garima Tripathi, Materials Science and Engineering A 443 (2007), Pages 262-269 [9] L.Calabrese, European Polymer Jounal 39 (2003), Pages1355-1363 [10] Reza Akari, Iran Polymer Journal (2013) [11] Johannes Karl Fink, Reactive Polymers Fundamentals and Applications: A Concise Guide to Industrial Polymers (2005)

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 44

EFFECT OF COMPOUNDING CONDITIONS ON THE MORPHOLOGY AND MECHANICAL PROPERTIES OF TIN FLUORIDE PHOSPHATE GLASS- POLYAMIDE 11 HYBRIDS Nora ITURRARAN1, Karine HURAUX1, Yuqing BAO2 and David Thomas GAWNE2 1

2

ARKEMA CERDATO S.A., Route de Launay, Serquigny 27470, France

London South Bank University, 103 Borough Road, London, SE1-0AA United Kingdom

ABSTRACT Novel hybrids of tin fluoride phosphate (TFP) glass (composition of 50% SnF2+20% SnO+30% P2O5) were synthesized with polyamide 11 and their morphology and mechanical properties investigated. Hybridization was achieved by melt blending up to 34 vol. % of glass using different compounding conditions (temperature, screw speed and residence time). Scanning electron microscopy (SEM) showed that the morphology was greatly influenced by the extrusion processing temperature and the glass content. Transmission electron microscopy (TEM) studies revealed nanoparticles of 40nm in size and suggested good compatibility. In order to determine the existence of miscibility between hybrid components, measurement of the loss tangent using a Dynamic Mechanical Analysis (DMA), was carried out. The presence of two transition peaks in the hybrid containing 34 vol. % tin fluoride phosphate glass implied an immiscible system showing heterogeneously distributed regions of very different molecular mobilities. Contrary to the plasticizer effect reported in literature for some polyamide 6 -TFP glass hybrids, the measurements of mechanical properties by DMA showed a reinforcement effect of glass in the polymer reflected by the increase of storage modulus (E’) at low and high temperatures in hybrids containing 18, 25 and 34 vol. % tin fluoride phosphate glasses, achieving the highest modulus at 25 vol. %. Tensile testing revealed a transition of material behaviour from ductile to brittle for high glass contents. Key words – Hybrid polymers, mechanical properties, morphology, phosphate glass

1. INTRODUCTION Polymer hybrids are potentially nanostructured materials with high performance properties that could make them important industrial material in the future. The increased attention received by polymer nanocomposites in recent years, has opened the way to research on nanostructured glass polymer hybrids for many application. Using polymer matrices and adding glass fillers to obtain nanoscale morphology can lead to interesting modifications and give rise new structures and properties. An additional benefit could be synergetic effects that help creating an improved material comparing with the neat polymer [1-5]. The high control degree over the composition and tuneable properties of the hybrid guide this research to investigate the effect of compounding conditions on the hybrid morphology together with miscibility between components and mechanical

properties. One way to achieve tuneable properties and interaction/ compatibilization between components is to add glass filler with the Tg low enough, where both hybrid components are fluid during the polymer processing temperature. This could allow the molecular level mixing of the hybrid components. The miscibility degree will surely influence the final properties and the morphology of the hybrid [6]. For this reason a new type of inorganic glass fillers like TFP glass are emerging. TFP glass is a low Tg phosphate glass that are both water-resistance and chemically durable. Its low Tg permit the melt mixing with several polymers by conventional processing methods like extrusion avoiding some processing problems inherent in classical composites [7]. The studies using polyamide 6 and polyamide 12 matrices have reported a good interaction between both components describing it as exceptional physiochemical interaction. The

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 45

interaction will result from binding amine group to the glass surface [6]. This could facilitate the miscibility in fluid state and adhesion in the solid state. Afterwards it found interesting to investigate about the processing parameters effect on the morphology, the miscibility and mechanical properties of novel hybrid composed of TFP glass and polyamide 11. 2. EXPERIMENTAL DETAILS

2.1. Materials The chemical components of glass were supplied by Sigma Aldrich Company. They are Tin (II) fluoride, Tin (II) oxide and Ammonium phosphate monobasic. The polyamide 11 used was supplied by ARKEMA as a trade name of Rilsan ®.

2.2. Preparation of glass The TFP glass has a molar composition of 50% SnF2+ 20% SnO+ 30% P2O5 and was synthesized in London South Bank University laboratory. The glass synthesis procedure followed was reported on Tick patent [8] and results in a glass with a density of 3.75 g/ cm 3 and a Tg around 135°C.

2.3.Preparation of hybrids Prior to melt mixing, generally the glass was ground using a mortar and sieved at 100 µm. The TFP glass- polyamide 11 hybrids were extruded using a micro twin screw extruder (µDSM) equipped with a micro injection machine. Most of the melt-mixed hybrid materials were collected in extruded form. Furthermore, some characterization tests require dog bone injected samples. In the table 1 are summarized the conditions: Table 1: Testing samples composition and compounding conditions Condition 1 2 3

T° Screw Residence Glass (C°) Speed Time Vol.% (rpm) (sg) 8 200 50 5 8 200 100 5 8 200 200 5

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

8 8 8 8 8 8 8 8 4 4 8 8 13 13 18 18 25 25 34 34 0

4 8 13 18 25 34

220 240 260 300 250 250 250 250 200 250 200 250 200 250 200 250 200 250 200 250 250 250 250 250 250 250 250

100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100

5 5 5 5

2 5 7 10 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5

E E E E

E E E E E E E E E E E E E E E E EI

EI EI EI EI EI EI

2.4.Characterization methods All the hybrid morphological evaluations were performed by FEI XL 30 SEM, excluding the nanoparticles detection of the hybrid, which was analysed by JEOL JEM 1230 TEM. To study dynamic mechanical properties, a DMA 242 Netzsch dynamic mechanical analyzer was used. The tensile test was carried out in Z1455 machine at 25 mm/min of speed. The mechanical properties were measured in samples prepared according to ISO 527-1BA.

3. RESULTS AND DISCUSSION 3.1.Morphology investigation Glass morphology was analysed and angular and break facet observed. The particle size range measured was 0-190µm, Fig. 1.

Process Extrusion(E) Injection (I) E E E

Fig. 1: Glass morphology

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 46

In order to study the effect of compounding conditions, morphology and particle size were determined using SEM. Screw speed Fig. 2 gives SEM micrographs showing fraction surface of condition 1 and 3 samples.

a)

b)

Fig. 2: The effect of screw speed on condition: a) 1; b) 3

Although there are coarse particles present in the two conditions, microstructural changes have been produced. The small and round dispersed phase particles means that the glass leaves its initial break faces, sharp and flat morphology. Residence time The micrographs of condition 8 and 11 are shown in the fig. 3. The micrographs are showing larger particle size for longer residence time that could suggest coalescence effect. The coalescence phenomenon occurs when two droplets moving in a flow field collide and maintain the contact until they merge to form a bigger one. The increase of residence time can favour positively the completion of coalescence process.

a)

b)

Fig. 3: The effect of residence time on the morphology of condition: a) 8; b) 11.

Temperature The condition 5 and 7 were observed to evaluate the influence of temperature, fig.4.

a)

b)

Fig. 4: Effect of processing temperature on the morphology on condition: a) 5; b) 7

The generation of the hybrid morphology is largely influenced by the processing temperature. The condition 7 sample shows homogeneous dispersion and very small particle size that can be considered as it is a nanostructured hybrid. The size of the glass particles in the matrix at smaller scale to ensure the nanoscale structure of the hybrid was observed by TEM. Condition 7 sample revealed nanoparticles with a 40 nm size fig. 5.

Fig. 5: Nanometer particles in TEM micrograph condition 7

Temperature versus glass content The influence of the temperature with glass content on the morphology of the hybrids was investigated at condition from 12 to 23, fig. 6. At 200°C comparative tests showed rough dispersion for low rates glasses. The dispersion at condition 20 became very fine with a few large particles. The presence of holes at low and high rates of glass those were not longer present at condition 20. The comparative observations at different rates of glass using a processing temperature of 250°C showed relatively coarse dispersions for rates of low glass, which then were refined. These dispersions are significantly thinner than those observed previously at processing temperature of 200°C. For rates from condition 13 and 17, we noticed that particle sizes greater than a few µm were not particularly spherical while those below the µm were perfectly spherical.

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 47

a)

viscosity, the shear rate and interfacial tension have great influence on it. Reactive compatibilization to achieve nanoscale particles took place. Our hypothesis is based on the interaction between the polyamide 11 and the phosphate as the chemical affinity between them is well known.

b)

3.2.Miscibility c)

d)

e)

f)

Fig. 6: Effect of processing temperature versus glass content on the morphology of condition: a) 12; b) 13; c) 16; d) 17; e) 20; f) 21.

At both processing temperatures 200°C and 250°C, they were obtained better dispersion at higher temperature, obtaining the best at condition 20 and 21. Clear evidence of great influence of processing temperature and glass content on the morphology related to compatibility of the hybrid is observed. The particle size ranges measured for condition from 12 to 23 are listed in table 2. Table 2: Glass particle size (PS) of different glass content hybrids at 200°C and 250°C Glass Vol. % 4 8 13 18 25 34

PS (µm) 200°C 1-150 1-125 1-100 1-160 1-140 1-210

PS (µm) 250°C

The samples made in condition from 24 to 30 were evaluated by DMA. The results of loss tangent (tan δ) versus temperature obtained from dynamic mechanical analysis provide additional information related to the miscibility mechanism between components. In the thermogram the Tg is identified as the highest value of tan δ peak. The height and the width of the peak describe the structural homogeneity and linkages, and the shift of peak the interaction between components. The results are summarized in table 3 and plotted in fig. 7. Table 3: DMA results of hybrids Condition 24 25 26 27 28 29 30

Tg (°C) 63 61 59 58 58 63 60

Polymer Glass PW PW Tan δ Tg Tan δ Δ°C Δ°C max (°C) max 40 0.126 39 0.127 38 0.128 37 Glass presence 0.13 Glass presence 0.13 38 41 Glass presence 0.13 35 150 0.215 0.127

1-40 1-70 1-50 1-20 1-20 1-20

The morphology generation from first angular glass particles to smaller and round dispersed phase in the polymer is governed by the viscosity ratio and capillary number [9]. Microstructural changes result from the breakup of particles under the influence of compounding condition. The droplets break up occurs under favourable condition converting the dispersed phase particles smaller. The

Fig. 7: DMA curves from condition 24 to 30

There is a peak displacement and slight decrease of width with the increase of glass. The height is slightly higher too. At condition

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 48

27, the curve starts to show other energy absorption. It is not possible to measure the Tg due to the absence of peak maximum. But we can relate to glass as it occurs close to 150°C. This is nearby of the glass Tg. At low glass contents it seems that glass movement follows the polyamide 11 chains movements. There is also a presence of energy dissipation around Tg in the condition 28 and 29. It is evident that as TFP glass is added to the polyamide 11 matrix, the Tg of the polyamide 11 doesn’t suffer a significant variation. The Tg of hybrids decrease slightly, maximum 5°C, or was either unchanged like in the case of hybrid of condition 29. The effect of the filler on the Tg of nanostructured composite materials is strongly dependant on filler- matrix interaction. There are many nanocomposites systems, where fillers only impose slight changes on neat polymer Tg [10]. The slight change was attributed to excellent nanometerscale dispersion. The detection of another chain mobility around 150°C starts after condition 27 hybrid. The chain mobility becomes more pronounced increasing vol. % TFP glass, obtaining a clear second transition peak in the hybrid of condition 30. This could be assigned to the restricted motions of polyamide 11 chains in the interface region due to strong interaction between both components. The presence of two transitions is the results of the heterogeneously distributed region of very different molecular mobilities. As the two peaks could be associated with the two hybrid components, the DMA results showed an immiscible hybrid system at high TFP glass filler loading. Dynamic mechanical properties The storage modulus (E’) of the hybrids is summarized in the table 4. In general, the incorporating of rigid filler particles to the neat polymer matrix should result in reinforcement effect. An increase of E’ in a wide range of test temperatures describes the existence of reinforcement effect in the composite. The conditions 24 to 27 have essentially the same or very similar storage modulus value. It can

distinguish the increase of E’ at several temperatures after condition 28 sample recording the maximum value at condition 29. In the table 4 are summarized E’ at 23°C and 150°C. Table 4: Storage modulus of hybrids from condition 24 to 30 Condition 24 25 26 27 28 29 30

Glass Vol. % 0 4 8 13 18 25 34

E’ (MPa) 23°C 1886 1824 1688 1881 2710 3080 2871

E’ (MPa) 150°C 261 252 223 250 314 388 297

The results at both temperatures, low and high, show the same evolution of E’ values against the glass content. The interaction responsible of the increase of E’ hybrid could be chemical or physical. The chemical interaction is a hybrid components bonding by chemical reaction. The physical interaction is related to particle ordering in the polymer matrix called agglomeration or percolation. The glass forms like a network in the matrix increasing the resistance of the material. It is common that high filler loading composites processed by melt mixing method using conventional processing techniques, suffer a particle agglomeration or percolation effect. Static mechanical properties To study the influence of the glass in the hybrids static mechanical properties, samples of 24, 26 and 29 were measured by tensile strength test. A ductile behaviour was identified with high deformation capacity before fracture for condition 24 and 26. The hybrid is fragile at condition 29 and it is deformed but almost there is not plastic deformation. The ductility of the polyamide 11 and the fragility of glass are well known in the material world. It is usual to obtain a reinforcement effect with mixing both components. This effect is characterized by the increase of tensile stress value and elastic

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 49

modulus, and the decrease of strain percentage. The condition 29 shows a reduction of deformation and a slight increase of tensile stress and elastic modulus suggesting slight reinforcement effect. These parameters are quantified in table 5: Table 5: Static mechanical properties of conditions 24, 26 and 29. Condition Glass vol.% 24 26 29

0 8 25

Elastic Modulus (MPa) 1007±37 1056±21 1086±208

Tensile Stress (MPa) 34±1 34.8±0.1 36.9±0.8

Deformation at break (%) 319±31 290±41 53±28

As a result, the glass content is one of the parameter that the mechanical properties depend on. At low glass content, the ductile behaviour of the matrix is capable to absorb the tensile strength applied with slight variation. At high glass content the brittle behaviour of the dispersed phase begins to have more importance due to significant reduction of the polymeric matrix. At condition 29, the material suffers a change in behaviour from ductile to fragile. Similar elastic modulus values were obtained.

4. CONCLUSIONS TFP glass with the molar composition of 50% SnF2+ 20 % SnO + 30 P2O5 was synthesized with polyamide 11 matrix by melt blending up to 25 vol. % of glass by changing the glass content and compounding conditions. Completely different morphologies were produced as a result of the compatibilization among the components. The morphology SEM analysis of hybrids revealed a major influence of extrusion processing temperature and glass content on the morphology of hybrids. Evidence of good compatibility and adherence between the hybrid components was found resulting from the reduction of the phase dispersed due to compatibiliziser located at the interface. Better homogeneity and dispersion of particles in the hybrid were detected at high temperatures and high glass loadings. The

optimum conditions were identified at 250°C and 25 vol. % tin fluoride phosphate glass. A nanostructure hybrid from macroscopic fillers was found to contain particles of 40 nm by transmission electron microscopy for the sample processed at a temperature of 300°C. The glass Tg characterization by DMA showed an immiscible hybrid system in the hybrid loaded with 34 vol. % TFP glass. The appearance of two transition peaks is the result of heterogeneously distributed regions of very different molecular mobilities. A clear phase separation of the hybrid components was found at this glass content. An increase in the storage modulus (E’) measured by DMA at low and high temperature was found at high loadings hybrid and revealed a reinforcement effect related to physical or chemical interactions between the hybrids components. A percolation phenomenon also played a role in this behaviour. Tensile testing showed a transition of material behaviour from ductile to brittle. The reinforcement effect found in the DMA testing was not evident in the tensile test results due the variation of elastic modulus in the samples. ACKNOWLEDGEMENTS The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7-MC-ITN) under grant agreement No. 264710. The authors would like to thank the Directorate-General for Science, Research and Development of the European Commission for financial support of the research. We thank Cerdato Research Centre of Arkema France S.A. for providing the needs to develop the research and London South Bank University for the academic support.

REFERENCES [1] M. Moan, J. Huitric, M. Mederic, J. Jarrin. Rheological properties and reactive compatibilization of immiscible polymer blends, J Rheol, Volume 44, 2000, pages 227–45 [2] Paul DR, CB Bucknall, Handbook of Polymer blends formulation, Wiley publications, Volume 1, New York, 2000 [3] DR Paul, CB Bucknall, Handbook of Polymer blends formulation, Wiley publications, Volume 2, New York, 2000

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 50

[4] CL Tucker III, P. Moldenaers, Microstructural evolution in polymer blends. Ann Rev Fluid Mech, volume 34, 2002, pages 177–210 [5] LA. Utracki, MR.Kamal, Melt rheology of polymer blends, Polym Eng Sci, volume 22, 1982, pages 96–114 [6] K. Urman, JU. Otaigbe, Novel phosphate glass/polyamide 6 hybrids: miscibility, crystallization kinectics, and mechanical properties, J Polym Sci, volume 44, 2006, pages 441-50 [7] K. Urman, J. U. Otaigbe, New phosphate glass/polymer hybrids—Current status and future

prospects, Prog Polym Sci, volume 32, 2007, pages 1462–1498 [8] LM. Sanford, Tick PA. Tin-phosphorous oxyfluoride glasses. 1982, US Patent: 4,314,031 [9] G.E.Schoolenberg, F.During and G. Ingenbleek, Coalescence and interfacial tension measurements for polymer melts: experiments on a PS-PE model system, Elsevier, volume 39, issue 4, 1998, and pages 765-772 [10] N. Chen, Y. Wan C, Y. Zhang, X. Zhang Y, Effect of nano-CaC03 on mechanical properties of PVC and PVCBlendex blend, Polym Test, 2004 23(2) 169-174

MANANO Seminar, 4th September 2013, Kingston University, London, UK Page 51