Processing and Properties of Zirconia-CNT Composites

Processing and Properties of Zirconia-CNT Composites A thesis submitted to Universitat Politècnica de Catalunya and Luleå University of Technology for the double degree of Doctor of Philosophy by

Latifa Melk

Dept. of Materials Science and Metallurgical Engineering at UPC Dept. of Engineering Sciences and Mathematics at LTU

Thesis directors: Prof. Marc Anglada and Ass. Prof. Marta-Lena Antti

May 2016

“Nothing in life is to be feared, it is only to be understood. Now is the time to understand more, so that we may fear less." ~Marie Curie~

iii

iv

ABSTRACT In the last decades there has been growing interest in developing ceramic materials with high fracture toughness (KIc) and strength for structural applications. In the specific case of 3 mol % yttria-doped tetragonal zirconia (3Y-TZP), can be increased by promoting phase transformation from tetragonal (t) to monoclinic (m) phase in front of a propagating crack tip referred to as transformation toughening. However, the stronger the tendency for stress induced transformation, the higher the risk for premature spontaneous t-m transformation in humid atmosphere. This phenomenon, which is referred to as ageing, hydrothermal degradation or low temperature degradation (LTD) induces microcracking and loss of strength and limits the use of 3Y-TZP. The resistance to LTD can be increased by reducing the grain size into the nanoscale by using Spark Plasma Sintering (SPS). However, the reduction of grain size may reduce t-m phase transformation in front of the crack tip and therefore fracture toughness may decrease. One way to enhance is the incorporation of a second phase as a toughening mechanism into zirconia matrix. In the present study, Multi-Walled Carbon Nanotubes (MWCNTs) were used to reinforce zirconia matrix. A novel method was developed in this project in order to measure the "true" fracture toughness of small cracks of 3Y-TZP/CNT composites. The method is based on producing a very sharp notch using Ultra-short Laser Ablation (UPLA). The same method was also applied to a high toughness zirconia ceramic 12Ce-ZrO2 with 300 nm grain size, which has much higher plateau fracture toughness than SPSed 3Y-TZP with 177 nm grain size. Moreover, the wear behaviour of zirconia/CNT composite was investigated by studying the effect of CNTs on the friction coefficient and the wear rate of the composites. The wear behaviour was investigated with scratch tests and reciprocating sliding. The machinability of zirconia/CNTs using Electrical Discharge Machining (EDM) was evaluated by studying the electrical conductivity, the thermal conductivity and the damage produced after machining. Besides that, the influence of grinding, thermal etching after grinding, and annealing of SPS zirconia with different grain sizes were studied. It has been found that by inducing a very sharp shallow notch using UPLA, the "true" of SPSed 3Y-TZP and 3Y-TZP/CNT composites for small cracks is low and independent of the added CNT amount. On the contrary, Vickers indentation is higher and increases with CNT content, which is attributed to the larger crack sizes studied in indentation and to an increase in the resistance to cracking under sharp v

contact loading induced by the presence of CNT. Therefore, indentation is not an appropriate method for analysing the influence of MWCNT on "true" fracture toughness. Moreover, only 10 % of difference in strength was found in 12Ce-ZrO2 and 3Y-TZP using UPLA method indicating that the "true" of both materials is almost similar. Thus, the beneficial effect of higher indentation in 12Ce-ZrO2 reported in literature has a very small effect on the "true" that determines the strength of unshielded small cracks. The incorporation of CNTs into zirconia matrix increases the friction coefficient and drastically decreases the wear rate when the amount of CNT reaches the percolation value (2 wt % CNT) under relatively low loads. However, during scratch test and under high loads, the composites develop chipping and brittle fracture. The addition of CNTs strongly enhances the electrical conductivity of the composite and induces slight changes in the thermal conductivity which results in successful EDM machining of the composites with 1 wt % and 2 wt % CNT. The material removal mechanisms in the composites are melting/evaporation and spalling. The thermal etching of ground SPS zirconia at 1100 °C for 1 hour in air induces a surface nanograin layer with crystallized grains of about 60 nm sizes and a thickness of less than few hundred nanometers, which is independent of the original grain size of the bulk material. If thermal etching is carried out at much higher temperature, 1575 °C for 1 hour, ground and polished zirconia reaches similar grain size.

vi

ACKNOWLEDGMENTS Although it is just my name on the cover, many people have contributed to the research in their own particular way and without them this project would never have been possible. First, I would like to express my gratitude to my principal supervisor Prof. Marc Anglada at UPC, in Spain, who has taught me a lot in Science and also in life. You allowed me to develop myself as a researcher in the best possible way. Then, I would like to acknowledge my second supervisor Ass. Prof. Marta-Lena Antti at LTU in Sweden. I greatly appreciate the freedom you have always given me to find my own path and the support you offered me when needed. I am grateful to the kind people at the division of Materials Science at LTU. I started with you when I was an undergraduate student and I grew up among all of you. Thank you wonderful people: Ragnar Tegman, Esa Vuorinen, Johnny Grahn and Lars Frisk for the heart-warming support. I´m thankful to the people I worked with in the department of Materials Science and Metallurgical Engineering at UPC. Your help and support made me realise that I am only a beginner in this exciting profession. I also would like to thank all my colleagues at both UPC and LTU and my friends from all over the world for your advices, constant enthusiasm and encouragement. Last but not least, I would like to thank my family for their encouragement, my Mom and Dad, my sister Sara and my brothers Zakaria and Yahya. I undoubtedly could not have done this without you. A special thanks to my partner Omar Benjelloun, you were always there for me with your quiet patience and unwavering love.

‫تحية خالصة للعائلتين ِملك و الِرسولي‬ "‫"لطيفة‬

vii

viii

APPENDED PAPERS Paper I L. Melk, J.J. Roa Rovira, M-L. Antti, M. Anglada, Coefficient of friction and wear resistance of zirconia–MWCNTs composites, Ceram. Int. 41 (2014) 459-468.

Paper II L. Melk, J.J. Roa Rovira, F. García-Marro, M.-L. Antti, B. Milsom, M.J. Reece, M. Anglada, Nanoindentation and fracture toughness of nanostructured zirconia/multiwalled carbon nanotube composites, Ceram. Int. 41 (2015) 2453-2461.

Paper III L. Melk, M. Turon-Vinas, J.J. Roa, M.L. Antti, M. Anglada, The influence of unshielded small cracks in the fracture toughness of yttria and of ceria stabilised zirconia, J. Eur. Ceram. Soc. 36 (2016) 147–153.

Paper IV L. Melk, M-L. Antti, M. Anglada, Material removal mechanisms by EDM of zirconia reinforced MWCNT nanocomposites, Ceram. Int. 42 (2016) 5792-5801.

Paper V L. Melk, J. Mouzon, F. Akhtar, M. Turon-Vinas, M-L. Antti, M. Anglada, Surface microstructural changes of Spark Plasma Sintered Zirconia after grinding and annealing, to be submitted.

ix

Conference contributions not included in the thesis L. Melk, F.G. Marro, A. Mestra, E. Jimenez-Piqué, and M. Anglada, Effect of MWCNT addition on the friction coefficient and electrical properties of 3Y-TZP, Poster presentation at the 13th European Inter-Regional Conference on Ceramics. Barcelona, Spain, September 2012. L. Melk, J. J. R. Rovira, F. G. Marro, M. Anglada, Mechanical and Electrical properties of 3Y-TZP/CNT composites, Abstract at the 7th EEIGM conference. Luleå, Sweden, March 2013. L. Melk. J. J. Rovira, F. G-Marro, M.L Antti, B. Milsom, M. J. Reece, and M. Anglada, Friction and Scratch Damage in 3Y-TZP/MWCNT composites, Oral presentation at the 13th International Conference of the European Ceramic society. Limoges, France, June 2013. L. Melk, J. J. R. Rovira, M-L Antti, M. Anglada, Coefficient of friction and wear resistance of 3Y-TZP/CNT nanocomposites, Oral presentation at the 14th European Inter-Regional Conference on Ceramics. Stuttgart, Germany, September 2014. L. Melk, M-L Antti, M. Anglada, the effect of SPS temperature on the mechanical properties of zirconia/CNTs composites, Oral presentation at 8th EEIGM Conference. Valencia, Spain, June 2015. M Turon-Vinas, L. Melk, J.J. Roa, M-L Antti, M. Anglada, Determination of fracture toughness from notches induced by ultra-short pulsed laser ablation, Abstract at European Ceramic Society conference. Toledo, Spain, June 2015. L. Melk, M-L Antti, M. Anglada, Mechanical and tribological properties of spark plasma sintered 3Y-TZP and 3Y-TZP reinforced MWCNTs composites, Oral presentation at European Congress and Exhibition on Advanced Materials and Processes. Warsaw, Poland, September 2015.

x

CONTENTS PART I CHAPTER 1-Introduction ........................................................................................3 1.1 Zirconia ceramics..............................................................................................3 1.1.1 Zirconia structure .....................................................................................3 1.1.2 Transformation from tetragonal (t) to monoclinic (m)................................4 1.1.3 The mechanism of toughening in zirconia ................................................5 1.1.4 R-curve behaviour....................................................................................7 1.1.5 Low thermal degradation of zirconia .........................................................8 1.1.6 Applications of zirconia ceramics ............................................................. 12 1.2 Carbon Nanotubes.......................................................................................... 13 1.2.1 Structure of Carbon Nanotubes .............................................................. 13 1.2.2 Synthesis of CNTs .................................................................................. 16 1.3 Processing of zirconia–CNT composites ......................................................... 19 1.3.1 Dispersion of CNTs ................................................................................ 19 1.3.2 Sintering ................................................................................................. 19

CHAPTER 2-Properties ..........................................................................................23 2.1 Mechanical properties ..................................................................................... 23 2.1.1 Reliability of indentation fracture toughness ........................................... 23 2.1.2 Comparison between indentation method and standardized fracture toughness tests .................................................................................................... 25 2.1.3 Effect of CNTs on the mechanical properties .......................................... 27 2.2 Tribological properties .................................................................................... 29 2.3 Electrical properties ........................................................................................ 29 2.4 Electrical discharge machining (EDM) ............................................................ 31 2.5 Thermal conductivity ..................................................................................... 32

CHAPTER 3-Materials and Methods .......................................................................33 3.1 Materials ......................................................................................................... 33 3.2 Powder processing and sintering ..................................................................... 34 3.3 Microstructural analysis ................................................................................... 35 xi

3.3.1 Density ................................................................................................... 35 3.3.2 Confocal microscope .............................................................................. 35 3.3.3 Scanning Electron Microscope ................................................................ 35 3.3.4 Focused ion beam (FIB) .......................................................................... 35 3.4 Phase characterization ..................................................................................... 37 3.4.1 X-Ray Diffraction (XRD) ...................................................................... 37 3.4.2 Raman spectroscopy ............................................................................... 37 3.5 Mechanical testing .......................................................................................... 37 3.5.1 Hardness and indentation fracture toughness ........................................... 37 3.5.2 True fracture toughness ........................................................................... 39 3.5.3 Nanoindentation ..................................................................................... 40 3.6 Tribological characterization ........................................................................... 42 3.6.1 Scratch testing ......................................................................................... 42 3.6.2 Wear test................................................................................................. 42 3.7 Electrical conductivity and Electrical Discharge machining (EDM) characterization....................................................................................................... 43 3.8 Thermal conductivity characterization ............................................................ 43

CHAPTER 4-Conclusions and Future work ............................................................45 4.1 Conclusions .................................................................................................... 45 4.2 Future work ................................................................................................... 46 REFERENCES ......................................................................................................... 49

PART II Summary of Papers ..................................................................................................... 61

PART III Paper I ....................................................................................................................... 69 Paper II ...................................................................................................................... 81 Paper III ..................................................................................................................... 93 Paper IV ................................................................................................................... 103 Paper V .................................................................................................................... 115

xii

Part I

2

CHAPTER 1 Introduction 1.1

Zirconia ceramics

1.1.1 Zirconia structure Zirconia is polymorphic of nature. It displays three different crystallographic structures depending on temperature. It exhibits a monolithic structure (m) from room temperature up to 1170 °C, a tetragonal structure (t) at intermediate temperatures (1170-2370 °C) and a cubic structure (c) when the temperature is above 2370 °C, see Fig. 1.1. The phase transformation from t to m is associated with a volume expansion of approximately 0.05 which induces microcracking. The t-m transformation starts at ≈950 °C ( ) on cooling in pure zirconia and is reversible on heating around 1150 °C [1]. The incorporation of some oxide dopants such as magnesium oxide (MgO), calcium oxide (CaO), cerium oxide (CeO2) and yttrium oxide (Y2O3) can suppress the phase transformation and stabilize zirconia in tetragonal form or even in cubic form. The major types of stabilized zirconia are formed depending on the final microstructure achieved: Fully Stabilized Zirconia (FSZ) is formed with full cubic structure. Partially Stabilized Zirconia (PSZ) contains cubic zirconia as the major phase while monolithic and tetragonal zirconia precipitate as minor phases. Tetragonal Zirconia Polycrystals (TZP) consists of tetragonal phase retained in a metastable state when cooled to room temperature. Y2O3 and CeO2 are the most promising stabilizers. They retain the tetragonal phase at room temperature in the form of TZP where the microstructure consists of equiaxed fine-grains [2]. The oxygen overcrowding around the small zirconium Zr4+ cations was shown to be responsible for the poor stability of tetragonal zirconia. Therefore, the use of oversized trivalent cation (Y3+) to stabilize tetragonal zirconia is more efficient than undersized trivalent cations [3,4]. 3

Introduction to Zirconia Ceramics

Fig.1.1 Schematics of the three polymorphs of ZrO2 and the corresponding space groups: (a) cubic (b) tetragonal and (c) monoclinic [4].

1.1.2 Transformation from tetragonal (t) to monoclinic (m) The transformation from t to m in zirconia is martensitic of nature. Although martensitic transformation is originally associated with transformation in quenched steels, it occurs as well in minerals and ceramics. A crystallographic correspondence of martensitic transformation exists between the parent (tetragonal) and the product (monoclinic) phase, described by habit planes and directions (shape strain) as shown in Fig. 1.2. The martensitic transformation is a change in crystal structure in the solid state through a diffusionless process. It is athermal and involves the simultaneous, cooperative movement of atoms over distances less than an interatomic distance which results in microscopic changes of shape of the transformed regions. This change in shape is associated with transformation toughening [5]. The thermodynamics of the t-m transformation (martensitic) in zirconia was first described by Lange [6] considering the ideal configuration of a spherical tetragonal ) associated to the particle in a matrix. The change of the total free energy ( transformation is given by equation (1.1): (

)

where (0) is the change in elastic strain energy associated to the transformation of particles and it depends on the modulus of the surrounding matrix, the size and shape of the particle and the presence of external stresses. Finally, the term (>0) refers to the change in the energy associated to the formation of new interfaces between the transformed particle and the matrix. In the case of 0, the tetragonal phase retains in the particle. A decrease in | 4

Introduction to Zirconia Ceramics followed by the addition of some oxides such as Y2O3 in Y-TZP, induce a decrease in the driving force of the t-m transformation and its temperature [3]. The transformation from (t) to (m) in zirconia is the source of enhanced toughness.

Fig.1.2 Schematic illustration of the crystallographic correspondences between the tetragonal (parent) and the monoclinic (product) phases during the martensitic t-m transformation.

1.1.3 The mechanism of toughening in zirconia Transformation toughening The two material classes better known to exhibit transformation toughening are transformation-induced plasticity (TRIP) steels and zirconia based ceramics [5]. Series of theoretical models have been developed to explain transformation toughening phenomena and they all agree on the development of a transformed “pro ess” zone associated with an advancing crack where a transformation of the metastable (t) phase takes place at the crack tip [4], see Fig.1.3. The tensile stresses generated induce a phase transformation from t to m around the crack tip and induce an increase in volume. The large volume generates compressive stresses around the crack and stops further crack propagation [3,4,7].

5

Introduction to Zirconia Ceramics

Fig.1.3 Illustration of transformation toughening in front of a propagation crack [8]. In 1980, McMeeking and Evans [7] developed a model of phase transformation toughening using linear elastic fracture mechanics. The model is based on the fact that the stress induced transformation toughening leads to a shielding, of the applied stress intensity factor which means that real stress intensity factor at the crack tip is lower than that applied by external forces according to the following relation: (

)

where is the stress intensity factor at the crack tip, is the applied stress intensity factor and is the shielding factor. Previous studies show that the higher the applied stress intensity factor, the larger the transformation zone and the larger shielding effect as shown in the following equations (1.3) and (1.4):

(

( )

)

(

√

)

(

)

(

)

where E is the elastic modulus, is the volume fraction of the transformable particles, is the dilatational strain associated with the transformation, is the Poisson ratio and is the critical stress leading to phase transformation.

6

Introduction to Zirconia Ceramics The toughness of zirconia depends directly on the critical local stress leading to transformation . A material with low and/or and a high will have low values of so that the contribution of transformation to shielding will be small ( ). On the contrary, with high and/or and low , a large (>>1) results in strong transformation, as far as the transformation is localised in front of the crack. The local stress transformation depends on the magnitude of the temperature of undercooling below the (t/m) temperature. The larger the undercooling below (t/m) the lower critical stress for stress assisted phase transformation, then the larger the transformation toughening [3,4].

Ferro-elastic toughening Besides phase transformation toughening, there is another mechanism of toughening referred to as induced ferro-elastic domain switching which also enhance the toughness of zirconia based ceramics. It is referred to as ferro-elastic toughening. In 1998, Virkar and Matsuimoto [9] reported for the first time, high values of toughness related to the ferro-elastic domain switching. The mechanism of the ferro-elastic toughening involves switching or alignment of the c-axis of the t-phase along the maximum stress axis, which induces a shape change of a pure shear type. Ferro-elastic toughening is possible in tetragonal zirconia produced by cooling from the cubic phase by a composition invariant displacive reaction. It can also occur when a direct deposition of tetragonalprime zirconia takes place by sputtering or electron-beam deposition for example [3].

1.1.4 R-curve behaviour R-curve, or crack resistance, is an increasing resistance of the material to crack growth under stress intensity factors higher than the crack lattice intrinsic fracture toughness. In general, R-curve behaviour is displayed by any material that exhibits nonlinearity in its stress-strain behaviour and experiences a crack-stabilizing effect [10]. The R-curve behaviour is more dominant for larger cracks because the effect of toughening mechanisms at the crack tip increases with the increase in crack size [11]. R-curve behaviour in ceramics is a result of extrinsic toughening mechanisms that are operating in the wake of a propagating crack, hence, an R-curve is referred as toughness-curve (t-curve) [12]. The increasing in toughness or R-curve for some ceramics is shown in Fig.1.4. Mg-PSZ exhibits a high crack resistance compared to Y-TZP which hardly has R-curve. The typical R-curve of Mg-PSZ is a result of transformation toughening in addition to crack deflection which both are crack shielding extrinsic toughening effects as reported by Hoffman [10]. A study by Eichler et al. [13] showed that the R-curve depends on the grain size and the testing environment. In the case of 2Y-TZP with grain size of 300 nm, the Rcurve reaches values up to 5.4 MPa√m in ambient air and 6.7 MPa√m in vacuum. 7

Introduction to Zirconia Ceramics Moreover, the authors reported no rising crack resistance with increasing crack length in air because of the smaller transformation zone. Gupta et al.[14] reported a fracture toughness of 5.7 for a grain size of 250 nm which is in line with the results reported before [13]. However, Fargas et al. [15] observed a small increase in plateau toughness in air from 3.9 to 4.3 MPa√m in 2.5Y-TZP with grain size of 300 nm. R-curves have been studied for many materials since the first detection of a rising crack growth resistance curve by Hübner and Jillek [16] in 1977. Then, most of the work on R-curves has been conducted on specimens with macrocracks and indentation cracks. Munz [17], in his review about "what we can learn from R-curve measurements" reported that the principal R-curve behaviour of a material can be measured with macrocracks. However, a quantitative prediction of the R-curve for natural flaws is not possible from measurements of specimens with macrocracks or with indentation cracks. Moreover, the investigation of materials with a rising R-curve results in increase in strength compared to a material with the same lattice intrinsic fracture toughness and a flat R-curve.

Fig.1.4 R-curves for some ceramic materials obtained from measurements of long cracks in compact tension samples [4].

1.1.5 Low thermal degradation of zirconia Low thermal degradation (LTD), referred to as hydrothermal degradation or aging, was first reported by Kobayashi et al. [18] in 1981 who discovered that zirconia samples could suffer from slow t-m transformation at the surface in humid atmosphere at 250 °C resulting in microcracking and loss of strength. Many investigations have been 8

Introduction to Zirconia Ceramics conducted to understand the LTD phenomena since its discovery, but it is still under debate. LTD was always considered important only at temperatures above human body or room temperature (37 °C) until 2001 when several of hundreds of hip prosthesis failed in a short time due to aging of zirconia femoral heads implanted in patients. This catastrophe limited the use of zirconia in medical applications and more deep investigations were needed to understand LTD of zirconia [1]. During LTD the degradation of properties is associated with the transformation from tetragonal to monoclinic phase. The reversible transformation (m-t) can occur, and then a total strength recovery is achieved, but only when aged specimens are annealed at high temperatures [19]. Yoshimura summarized experimental observations of LTD on zirconia as follows:     

LTD is accelerated at temperatures of 200-300 °C and it is time dependent. LTD is due to t-m transformation accompanied by micro- and macro-cracks. The transformation progresses from the surface to the interior of the sample. The transformation is enhanced by water or water vapour. A decrease of the grain size and increase of stabilizer retards the transformation.

The mechanism of LTD as reported in the work of Sato et al. [20] is a reaction between water and Zr-O-Zr bonds at the crack tip during the transformation in Y-and Ce-ZrO2. The results showed similar values for the reactions. Therefore, it was assumed that water react primary with Zr-O-Zr bonds on the surface and not the stabilizing oxide. Yoshimura et al. [21] reported the role of penetration of OH molecule on the degradation in Y-TZP in several steps as follows: Step 1: Chemical-adsorption of H2O at the surface. Step 2: The formation of Zr-OH and/or Y-OH bonds at the surface. Step 3: The migration of OH- ions at the surface and in the lattice so nucleating defects are prepared. Step 4: The nucleation of monoclinic phases in the tetragonal grains. Chevalier et al. [22] showed that LTD undergoes a nucleation and growth mechanism. The nucleation of transformation on one grain leads to a volume increase stressing up the neighbouring grains resulting in microcracking which offers a way for water to penetrate the specimens, see Fig. 1.5. The extension of transformation occurs preferentially in the neighbouring grains. Nucleation takes place on the most unstable grains that are subjected to the highest internal/applied tensile stresses. The unstable grains can be those with less Y2O3 or and with large size and/or subjected to higher internal stresses.

9

Introduction to Zirconia Ceramics

Fig.1.5 Illustration showing the transformation neighbour to neighbour (a), nucleation on a particular grain at the surface (b) and (c) growth of the transformed zone [23]. The number of nuclei increases with the stresses because of the penetration of water. Simultaneously, a growth mechanism takes places because the transformation of one grain puts its neighbours under tensile stresses, favouring their transformation under the effect of water. The kinetics of aging was described using Mehl-Avrami-Johnson (MAJ) laws in which time and the amount of monoclinic phase is described as the following: [ (

) ]

(

)

where, is the transformation fraction, is the time, and and are constants. The exponent is related to the nucleation and growth conditions. The parameter is related to the activation energy and it depends on temperature and follows an Arrhenius law in temperatures between 37-140 °C: [

]

(

)

where is a constant, the gas constant and the absolute temperature. An increase of monoclinic phase with increasing aging time has been reported by Chevalier at al. [22] at different temperatures (70-130 °C), see Fig.1.6. The results show an incubation-nucleation-growth mechanism of transformation because of the sigmoidal variation of the monoclinic phase with time for all temperatures.

10

Introduction to Zirconia Ceramics A linear relationship for each temperature with a constant value of equal to 3.6 was observed which fits very well with the experimental data. This value between 3 and 4 corresponds to a nucleation and three dimensional growth process according to the MAJ model. Therefore, the MAJ model gives a good prediction of monoclinic phase at the surface of aged 3Y-TZP for a given time and especially low temperature.

Fig.1.6 Monoclinic phase versus time from a temperature range 70-130 °C [22]. There are many factors influencing the LTD such as the density, grain size, the homogeneity of the phase distribution and the residual stresses on the surface. For instance, the low density together with the presence of open porosity allow water molecules to penetrate into the bulk resulting in microcracking and pores and then leading to a drop in the mechanical properties. Grain size has also a strong influence on LTD of zirconia. It was widely reported in the literature [24–26] that a decrease in the grain size (200 nm) limits aging. But, a reduction of grain size induces a decrease of toughness because of the transformation toughening. Indeed, a decrease of grain size influences the surface term in equation (1.1) where this latter leads to a dependence of the activation barrier for t-m transformation on the particle size. Hence, the activation barrier for the formation of critical nucleus is increased by a decrease in particle size. A bimodal microstructure of 3Y-TZP appears when it is sintered at high temperature and consists of cubic and tetragonal phases. The cubic grains are larger than the tetragonal ones which pump the Y2O3 out. The tetragonal grains are depleted and less stable. Thus, they act as nucleation sites [27–29]. The stress distribution plays a critical role in transformation. The microscopic tensile stresses in a grain trigger the t-m transformation. Schmauder and Shubert [30] reported 11

Introduction to Zirconia Ceramics that unconstrained grains were stable under humid conditions, whereas constrained grains transformed. The principal stresses from the thermal expansion anisotropy enhance the transformability of tetragonal grains to the monoclinic grains by increasing the t driving force and lowering the nucleation barrier. Increasing the amount of Y2O3 stabilizer increases the resistance to LTD. It was reported that 4 mol % yttria stabilized zirconia shows better resistance to LTD compared to the 3 mol % yttria stabilized zirconia [31].

1.1.6 Applications of zirconia ceramics Zirconia ceramics are used in wide range of applications: a dental bridge, oxygen sensors, grinding media, fuel cells, cutting blades, knifes and bearings for example. However, its main use is in the medical sector because of its outstanding mechanical properties, especially its unique balance of toughness and strength compared to other ceramic oxides, as well as biocompatibility. More than 60000 zirconia femoral heads have been implanted in the United States and Europe. Later on, 400 femoral heads have failed due to aging which resulted in a catastrophic impact on the use of zirconia in the medical sector. In the same time, results showed excellent behaviour of some heads after several years in vivo [23]. Therefore, the correlation between aging and clinical failures is still a matter of debate. The orthopaedic market sale decreased more than 90 % between 2001 and 2002 [3]. On the other hand, the aesthetic appearance in combination with excellent mechanical properties makes zirconia an excellent choice to be used in restorative dentistry. Zirconia has been used as implants and implant abutments, orthodontic brackets, cores for crowns, included endodontic posts and fixed partial denture prosthesis (FPDP) framework. As the grain size strongly affects the mechanical properties and aging, the grain size of 3Y-TZP used in dental applications consists of small equiaxed grains in the range of 0.2–0.5 μm depending on the sintering temperature [32]. The flexural strength of zirconia used in restorative dentistry is ≈ 1000 MPa and the indentation fracture toughness of ≈ 7 MPa√m [33].

12

Introduction to Carbon Nanotubes

1.2 Carbon Nanotubes Nowadays, carbon nanotubes (CNTs) are considered as one of the most important nanomaterials making the next industrial revolution. They are used in many applications such as: electronics, semi-conductors, aerospace, etc. Their extraordinary properties are the key behind the high demand.

1.2.1 Structure of Carbon Nanotubes In 1985, a new form of carbon C60 called the fullerenes [34] was discovered by Harold Kroto, Richard Smaley and Robert Curl (Nobel prize in chemistry in 1996). The main objective of their work was to determine the nature of carbon "clusters" present in interstellar space. Later on, intensive investigations were conducted from all over the world on the new carbon allotropes. In 1991, Iijima [35] observed for the first time multi wall CNTs (MWCNTs) and later in 1993, he discovered single wall CNTs (SWCNTs) [36]. Fig.1.7 shows different types of CNTs. CNT has a cylindrical form; it can be seen as a rolled plane of graphene. Its diameter is in the nanometric scale and its length can be in several centimetres which lead to a high aspect ratio (length/diameter) of more than 10 7. CNTs have unique properties. They have a high Young’s modulus (≈1500 GPa), high tensile strength (≈100 GPa), significantly higher than steel and carbon fibres, high thermal conductivity which is higher than diamond and an electrical conductivity similar to silver and platinum. The density of CNTs is much lower than aluminium [6][37]. Depending on the number of graphene cylinders (x), CNTs are referred to as SWCNTs (x=1), DWNT (x=2) and multi wall CNTs (MWCNT) with x> 2. For MWCNTs, the interspace between two successive CNTs is in the range from 0.344-0.36 nm and the carbon nanotube bond length is 0.144 nm [37]. SWCNTs can be visualized as a single sheet of graphene wrapped up to form a tube and for most observed SWCNTs, the diameter of the cylindrical graphene sheet is 1700 °C). For MWCNTs, arc discharge is based on using DC arc discharge between two graphite electrodes, usually water-cooled in a chamber filled with helium (He) at sub-atmospheric pressure. He atmosphere was used by Iijima et al. [35] for the synthesis of fullerenes in order to obtain first scale synthesis of CNTs. Zhao et al. [40,41] used He and methane gases and they resulted in different CNTs 16

Introduction to Carbon Nanotubes morphology. A study by Wang et al. [42] has confirmed that the use of different atmospheres strongly influences the morphology of CNTs. On the other hand, synthesis of CNTs in liquid solutions has also been studied. Thus, Jung et al showed that the use of arc discharge in liquid nitrogen results in high yield MWCNTs. In the case of SWCNTs, the synthesis by arc discharge implies the use of a composite anode in hydrogen or argon atmosphere. The anode is made of graphite and a metal such Ni, Fe or Co. The anode can also be a mixture of a metal and another element like: Co-Ni, Fe-Ni, Fe-No. Iijima and Ichihashi [36] reported the first synthesis of SWCNTs with a diameter of 1 nm. Then, Berthune et al. [43] reported the synthesis of SWCNTs by Co catalysis with a diameter of 1.2 nm. Saito et al. [44] reported growing of SWNTs radially from Ni fine particles. In 1996, they investigated the use of platinum group metals (Rh, Pd, Os, Ir, Pt) in the synthesis of SWCNTs by arc discharge [45]. Later on, a new approach was reported based on using a hydrogen DC arc discharge called ferrum-hydrogen which results in a high crystallinity of SWCNTs [46,47]. Moreover, a cheap method was reported for the synthesis of high purity SWCNTs with a diameter of 1.2 nm. The method was based on the synthesis of SWCNTs in argon DC arc discharge from charcoal as carbon source and FeS as catalyst [48].

Laser ablation Pulsed laser deposition (PLD) used for the synthesis of CNTs is similar to the arc discharge, where this time the energy is provided by a laser hitting a graphite pellet containing nickel or cobalt as catalyst material. Laser ablation is considered as a superior method for the production of SWCNTs with high quality and purity [49]. The lasers used for ablation have been Nd:YAG and CO2. Some studies reported that the use of a continuous wave of CO2 laser ablation without any additional heating to the target and with increasing the laser power will increase the average diameter of SWCNTs. A study by Kusaba and Tsunawaski [50] showed that using a laser ablation at 1623 K with oscillation wavelength of 308 nm irradiating a graphite containing Co and Ni at various temperatures resulted in the highest yield of SWCNTs with the diameter between 1.2 and 1.7 nm.

Chemical vapour deposition The catalytic chemical vapour deposition (CCVD) is considered as an economic and practical method for the production of CNTs with high purity and more controlled production compared to laser ablation. Thus, CVD is now a standard method for the production of CNTs. The models presented by Fotopoulos and Xanthakis [51] suggested that SWNTs are produced by based growth where the cap is formed first, and then the CNTs are fabricated by addition of carbon atoms at the base. 17

Introduction to Carbon Nanotubes The most used catalysts in CVD are Fe, Co, or Ni [52]. The catalysis contributes to the decomposition of carbon source by plasma irradiation or heat and to the nucleation of CNTs. The carbon source used in CVD is hydrocarbons while the substrates commonly used are Ni, Si, SiO2, Cu, Cu/Ti/Si, stainless steel or glass [53,54]. A study by Flahaut et al. [55] reported the influence of catalyst conditions on the synthesis of CNTs by CCVD. They observed that the combustion conditions in the case of using citric acid as a catalyst can induce either an increase in the CNTs walls or limit the formation of carbon nanofiber. Lyu et al. [56] used benzene as an ideal carbon source and Fe-Mo/Al2O3 as a catalyst at 900 °C and they succeeded to produce highquality and pure DWCNTs. Cui et al. [57]used acetonitrile as the carbon source and the ferrocene as the catalyst and they could produce thin walled, open-ended, and well-aligned N-doped CNTs on the quartz slides that can be used in many applications.

18

Processing of zirconia-CNT composites

1.3

Processing of zirconia–CNT composites

1.3.1 Dispersion of CNTs The synthesis methods reported in section 1.2.2 result in production of CNTs with different physical entanglement of the nanotubes where individual CNTs are entwined, interwoven and agglomerated. Therefore, the dispersion of CNTs is considered as a crucial step and a challenge in the processing of ceramic-CNT composites. A chemical entanglement can occur when a surface-surface attraction takes place. Moreover, the high aspect ratio, the high flexibility of CNTs increases rapidly the chance of entanglements. Besides that, attra tive Van der Waal’s for es between arbon surfa es increase the difficulty of CNTs to be dispersed. The molecular forces between carbon nanotubes are influenced by both chirality and surface curvature [58]. The objective of dispersion is to produce independent separated tubes that can be oriented either in one dimension (fiber), two dimensional (flat sheet) or three dimensions (bulk solid). The dispersion can be physical (mechanical) or chemical. Mechanical dispersion consists of separating the tubes by using ultrasonication for example. Chemical dispersion is based on functionalization using solvents or surfactants to change the surface energy of the nanotubes. The functionalization of the tubes may improve their adhesion characteristics to the matrix and helps to reduce agglomerations. On the other hand, the use of aggressive chemical or mechanical dispersion will result in changes in CNTs properties [58].

The ultrasonication Ultrasonication method is usually used to improve the dispersion of CNTs by shortening the tubes. However, there is a risk that the tube-wall will be damaged or broken. Therefore, an advance ultrasonication method based on the use of diamond crystals is developed where the SWCNTs bundles are destroyed but not the tubes. For MWCNTs, it was reported that the ultrasonication seems to destroy the external layers, so MWCNTs will not only get shorter but also thinner with time [59]. Moreover, ultrasonication can be used to remove impurities. A study showed that CNTs has been purified from ≈70% to ≈90% by ultrasonication-assisted filtration [60].

1.3.2 Sintering Sintering is a thermal treatment of powder compacts at temperature below the melting point resulting in dense polycrystalline solid. Thermodynamically, sintering is an irreversible process in which free energy decreases is brought by a decrease in surface energy. The main methods of sintering are described below. 19

Processing of zirconia-CNT composites Hot Isostatic Pressing (HIP) Hot Isostatic Pressing is a technique invented in United States in order to diffuse bonding nuclear fuel element assemblies [61]. HIP was first used to bond materials using high temperature and isostatic pressure. Therefore, it was called Gas Pressure Bonding [62]. High performance and high temperature ceramics such as silicon nitride and silicon carbide were the first to be produced commercially. Moreover, HIP was used to produce alumina cutting tools with excellent cutting properties at a reasonable cost. Later on, HIP was spread to oxide ceramics such as Mn-Zn ferrite, PZT and PSZ being one of the most promising technologies produce parts with good mechanical strength and reliability [63] . The advantages using HIP lies in the high densification, elimination of porosity, improvement of fatigue properties, improvement of creep properties, improvement of ductility and strength and recovery of defective parts. Besides that, it has been found that HIP can achieve a desired shape with a high degree of control and accuracy [61,64,65]. HIP equipment consists of an electric furnace contained in a pressure vessel. The high pressure transmitting gas which fills the pressure vessel is argon gas because the viscosity of the gas is very low and its density is very high. As a consequence, the heat generated by the heating is transferred by the natural convection of the gas. The furnace, the heating device and the heat insulation are made of heat resistant alloys or refractory metals [61].

Spark Plasma Sintering (SPS) Sintering under an electric current was first patented in 1906. In 1960, an invention of Spark sintering based on pulsed current was reported. However, the high equipment cost and the low sintering efficiency limited its wider use. The concept was further developed and in 1990s, Japanese companies, in particular Sumitomo Coal Mining Co. Ltd started the industrial production of Spark Plasma Sintering machines based on the use of pulsed direct current to heat the specimen. Later on, FCT System GmbH in Germany and Thermal Technology LLC, Inc. in the USA started producing similar equipment based on pulsed DC current. Today, the number of SPS machines installed in the world reaches 1750 with 2/3 in the industry. Besides that, more than 3000 publications have been reported using this technique and its applications according to the ISI Web of Science [66]. The configuration of SPS consists of a uniaxial pressure system, together with a high power electrical circuit placed in a controlled chamber. There is also a pulsed direct current (DC) generator, water cooled reaction chamber, a pressure-, a position- and temperature-regulating systems. The powder is located into the die, and a mechanical pressure is applied during the sintering process, see Fig.1.10. 20

Processing of zirconia-CNT composites A low voltage typically below 10 V is applied leading to an efficient Joule heating that goes through the sample. Even in the case of non-conductive ceramic powder the heat is efficiently transferred to the sample. Rapid heating rates could reach 1000 °C/min reducing the duration of the process and the energy costs [66]. Moreover, cooling rates of 150 °C/ min are also possible. The mechanical pressure applied to enhance the densification could be as high as 50 to 250 kN. The densification takes place under vacuum or protective gas. Finally the maximum temperature using standard graphite tools could attain 2400 °C where either thermocouples or axial/radial pyrometers are used to control the temperature. It was reported that when using non-conductive powder together with the use of an electrically conductive die, the current will be forced to pass through the powder generating the highest possible current density, see Fig.1.11. The stress applied in SPS induces changes in the morphology of contact between the particles and enhances the densification mechanisms. The grain growth during sintering is usually delayed and reduced. However, the grain size can be several times larger than the initial particle size as was reported for several oxides densified by SPS and HP [67]. It was reported that a finer starting particle size, in the opposite of coarser one, may induce a larger grain size at full density [68]. By increasing the pressure during SPS, the agglomeration breaks. At low temperature, the particle rearrangement increases reducing the pore size which enhances densification and limits the grain growth. At high temperature, other mechanisms take place such as plastic deformation or power law creep. Thus, low temperatures are needed for full densification [66].

21

Processing of zirconia-CNT composites

Fig.1.10 Schematic of SPS showing the different parameters involved during sintering [66].

Fig.1.11 Schematic of current flow in case of non-conductive powder and a conductive die [66].

22

CHAPTER 2 Properties 2.1

Mechanical properties

2.1.1 Reliability of indentation fracture toughness Measuring fracture toughness using indentation method has always been a matter of dispute. The origin of indentation fracture toughness was first initiated in 1970 by Evans and Charles [69] in a short communication. Then, the indentation technique achieved a high popularity because of its expediency. Moreover, it needs a short time for preparation and it is not costly. Therefore, the use of indentation method is practically ideal even if it is reported that it is not an accurate method for the determination of fracture toughness. The indentation technique described by Evans and Charles [69] suggests using the cracks which emanates from the corners under a high testing load. They presented a generalized equation and a normalized calibration curve that appeared to apply to many different materials with Palmqvist cracks or with median cracks. Later on, several authors have used similar curve fitting methods. However, many researchers have been confused with their results using the indentation method which have led to the issue of more than 30 equations presented in literature [70]. Indentation fracture toughness consists of the preparation of a high quality smoothly polished surface. Then, the polished surface is indented by a Vickers pyramidal hardness indenter. The sample is indented at high testing load until a deformed region is created beneath and in the vicinity of the indent. As a consequence cracks are generated from the four corners of the impression, see Fig.2.1. The indentation is not always perfect, for example it can lead to spalling chipping at high test loads. The parameters involved in calculating the indentation fracture toughness are: the lengths of the cracks, indentation load, impression size, the hardness and elastic modulus and a calibration constant. One should take care while measuring the lengths of the cracks Depending on the load and the multiple cracking, the material surrounding the 23

Properties of zirconia-CNT composites impression could be left with a complex residual stress state. Therefore, there are many questions raised concerning the sequence of crack growth from the Vickers indentation corners. It could for example be that the cracks first form Palmqvist cracks that can extend to median cracks or it could be that median cracks form directly from the deformation beneath the indenter. Furthermore, the cracks could be formed either during indenter loading or unloading. Besides that, the environment such as the water vapour may affect the results [70]. As mentioned before, there are numerous equations reported to calculate the indentation fracture toughness. Most of the equations are only a manipulation of previous equations with new calibration constants. Moreover, most of the equations proposed are highly questionable because of their lack of an accurate stress intensity factor solution. Furthermore, the complex network and the residual stress damage zone around indentations are not helpful to give a direct analysis as in traditional fracture mechanics tests. However, the most frequently used equations after Evans and Charles [69] are the ones of Marshall and Evans [71] who simplified the formula of indentation. Afterwards, there are Anstis et al. [72] who proposed additional modifications to the proposed equations. Then, two more papers were published by Niihara [73] describing the Palmqvist cracks. Finally, an adaptation of the equation of Anstis et al. [72] was reported by Miyoshi et al. [74] for median cracks.

Fig.2.1 Vickers indentation with cracks formed under 10 Kg load in SPSed zirconia.

24

Properties of zirconia-CNT composites

2.1.2 Comparison between indentation method and standardized fracture toughness tests The geometry in a standard fracture toughness test is simple and the crack has welldefined shape for which loading conditions in addition to the state of stress are known. For the indentation method, any geometry can be convenient as long as it can be mounted and polished. Moreover, there is no pre-crack in the specimen subjected to an indentation test. There are cracks that are generated when the indenter is forced into the polished specimen surface. However, in standard tests, there is a single well-defined pre-existing crack for which there is a formal stress intensity solution. The loading in standard test is done using a universal mechanical testing machine and a prismatic specimen where a sharp notch is machined through the specimen. For indentation method, under loading, plastic deformation occurs below and around the indentation which results in multiple cracks that propagate beneath the indenter in addition to high complex residual stresses. In fact, the method is based on a model of the residual elastoplastic stress field which depends on the material and on the assumption that this field can be represented by an opening force on the centre of the indentation. For quantitative estimations the equation of the stress intensity factor is calibrated by a constant from the fracture toughness determined using standard methods. Li et al. [75], Ponton and Rawlings [76] and Ghosh et al. [77] discussed the credibility of the different equations used for the indentation method and they revealed its weakness by summarizing indentation results using different equations for the same materials. The most critical conclusions of the four publications are summarized as follows: 1) The different proposed equations for the calculation of fracture toughness for the same or closely the same crack length in ceramic materials result in wide range of different KIC results. 2) The equations proposed are not able to generate accurate results for different materials. 3) The different equations result in increase or decrease of fracture toughness depending on the crack length for the same material. Moreover, the "R-curve" behaves differently for the same material when the different equations for the calculation of indentation fracture toughness are applied. The above conclusions illustrate clearly the inadequacy of using indentation method. The first conclusion shows that indentation method does not result in unique value of KIC. The wide range of KIC results obtained by using indentation method indicates its unreliability in measuring the true KIC of brittle materials. In addition, brittle materials deform and fracture differently underneath an indentation. Ceramics with covalent bonding such as silicon carbide deform differently than the ceramics with ionic bonding such as magnesium oxide or glasses. Furthermore, fine grain and coarse grain ceramics have different fracture mode as well. Moreover, the indentation cracks are three 25

Properties of zirconia-CNT composites dimensional networks. Therefore, the cracks cannot be idealized as Palmqvist, semicircular or median as it is assumed in the equations proposed. The variation of indentation fracture toughness with the crack length seems to be in a contradiction with the R-curve of brittle ceramics. The R-curve is either flat as for fine-grain ceramics or increases with crack length as for coarse grained polycrystalline ceramics. R-curve should not be decreasing with increasing the crack length, as was reported in conclusion 3 which makes the indentation method results confusing. As summary, it is concluded that the different equations proposed for the calculation of indentation fracture toughness in the literature are not reliable. Therefore, indentation fracture is not a suitable method for the calculation of fracture toughness. For this reason, we have in the current work developed, a new testing method from the measurement of the "true" fracture toughness of small cracks [78]. The new method is based on inducing a very sharp shallow surface notch using Pulsed Laser Pulsed Ablation (UPLA) on the surface of prismatic bars which was parallel to the original discs and determining the "true" KIC from the strength in four point bending [78]. More details on the experimental procedure will be discussed in Chapter 3. See Fig.2.2 shows an example of a notch induced by UPLA.

Fig.2.2 Side view of the notch induced by UPLA on SPSed zirconia.

26

Properties of zirconia-CNT composites

2.1.3 Effect of CNTs on the mechanical properties It was reported that a single CNT failed after 280 % stretching at high temperature [79]. A study by Treacy et al. [80] reported individual CNTs with Young’s modulus of more than 3 TPa. Lourie et al. [81] estimated that the stress required for producing buckling or collapse of CNTs is around 100-150 GPa. Therefore, it is evident that CNTs are the strongest and the stiffest fibres ever found. However, the efficiency of the addition of CNTs in improving the mechanical properties of ceramic matrix composites is still a matter of debate. Mukhopadhyay et al. [85] reported that the crack bridging provides a 150 % improvement of the fracture toughness compared to a monolithic material and that no fibre pull could be detected. A study carried out by Estili et al. [84] showed that the different toughening mechanisms such as crack deflection, fibre pull-out or crack bridging of CNTs induce an increase of 67 % of the fracture toughness with the addition of 3.5 vol % MWCNTs to alumina matrix. However, the hardness and elastic modulus were decreased with the addition of CNTs in the matrix. Moreover, An et al. [86] showed that the strong bonding between the matrix and MWNTs result in the improvement of elastic modulus from 74 GPa to 118 GPa by adding 6.4 vol % MWNT to silicon carbonitride (SiCN) matrix. On the other hand, as it was described in the previous section, the use of the indentation method results in a big dispersion of different KIC values in literature. A study by Zhan et al. [82] of the addition of SWCNTs in alumina nanocomposites showed a high increase of indentation fracture toughness (9.7 MPam 1/2) and a decrease in hardness of about 20 %. However, Wang et al. [83] had ambiguous results by the use of indentation method in the determination of fracture toughness. Therefore, they used single edge V-notch beam (SEVNB) on the same composites and they showed an improvement of only 3 % in fracture toughness. Nevertheless, they used a razor notch radius which is not sharp enough, see Fig.2.3. The study by Wang et al. [83] also reported that alumina-graphite composites prepared by SPS do not have a high fracture toughness, but they have a good contact-damage resistance. SWCNTs are highly sheardeformable which induce an intensive confined-shear field and causes a redistribution of the stress under the indenter which prevents the formation of cracks. Melk et. al [78] studied the addition of MWCNTs to zirconia matrix and they showed that the "true" fracture toughness of the composites is hardly increasing while the indentation fracture toughness is increasing which indicate that the incorporation of CNTs to zirconia matrix increases the resistance to cracking under contact loading. In 2007, Quinn and Bradt [84] suggested strictly to not use the indentation method for the determination of fracture toughness. Sun et al. [97] and Ukai et al. [98] showed that the addition of 1 wt % CNTs to 3YTZP/CNT nanocomposites induces no improvements of the indentation fracture toughness, fracture strength or hardness compared to 3Y-TZP matrix. They reported 27

Properties of zirconia-CNT composites that the low mechanical properties are due to the weak interfacial bonding between the matrix and the CNTs.

Fig.2.3 Fracture toughness of the SWCNT-alumina composites measured by (a) Indentation method measured by Zhan et al. [82] and (b) SEVNB method measured by Wang et al. [83]. Garmendia et al. [100] reported small improvements in indentation fracture toughness using slip casting for a better dispersion of carbon nanotubes in zirconia matrix. However, they found low Vickers hardness. Duszova et al. [6], [99] showed poor mechanical properties in hot pressed zirconia carbon-nanofiber composites. They reported that the composites have low density compared to monolithic zirconia. Dusza et al. [101] used hot pressing and SPS for the sintering route of 3Y-TZP/CNF. The use of SPS showed higher indentation fracture toughness and hardness than the samples sintered by hot pressing. They concluded that using SPS for a short sintering time of only 5 min can prevent the nanotubes from oxidation and damage. On the other hand, the addition of CNTs to alumina matrix has shown good results. It was reported by Fan et al. [88] that the addition of 1 % SWCNTs into alumina increases the indentation fracture toughness by 103 %. Moreover, Fan et al. [89] showed an improvement of 80 % of the indentation fracture toughness by studying hot pressed alumina-MWCNTs nanocomposites. However, Mo et al. used SPS for the processing of alumina-CNT and reported 7 % improvement of the Vi kers’s hardness and 10 % increase of indentation fracture toughness [87]. Chang et al. [92] and Siegel et al. [92] reported an improvement of 24 % of the indentation fracture toughness compared with a value of 4.2 MPa·m 1/2 of the monolithic alumina. However, Wei et al. [94] showed that by adding as little as 3 vol % CNTs to alumina matrix the indentation fracture toughness increases by 79 % and the bending strength increases by 13 %. 28

Properties of zirconia-CNT composites

2.2

Tribological properties

The effect of CNTs on the tribological behaviour of ceramic-CNTs composites has mainly been studied in alumina-CNT composites. The first study was carried out by An et al. [85] on the wear mechanism of alumina-CNT composites. They reported a decrease in the wear loss of 56 % and an improvement of 30 % in microhardness. A study by Axia et al. [86] showed how well-aligned CNTs can determine the friction coefficient (COF) of the composites using pin-on-disc geometry. The highly ordered nanotube ceramic composites exhibited low COF and high wear resistance by tuning the CNT thickness and buckling properties. The authors reported 80 % reduction of the COF. Studies of the tribological behaviour of CNT composites with zirconia as matrix material are fewer. In 2010, Hvizdos et al. [87] studied the wear behaviour of zirconiacarbon nanofiber (CNF) composites. They reported that the addition a little as 1.07 wt % CNF results in excellent frictional properties of the composites. The low friction coefficient achieved was related to the formation of a carbon-based transferred film on the surface which induces easy sliding and shear due to the lubricating effect of CNTs. Two years later, Hvizdos et al. [88] studied the effect of CNF/CNT on different matrix composites: ZrO2, Si3N4, Al2O3. They showed the benefits of the presence of CNFs on the friction coefficient of the composites. They concluded that a small amount of CNFs can decrease the friction coefficient while the CNTs are less effective. Then, it is necessary to incorporate high amount of CNTs for better frictional results. The authors reported that the percolation of CNF occurred at ≈1 wt % CNF in ZrO2 whereas, a percolation of 3 wt % and 5 wt % of CNT occurred for Si3N4 and Al2O3 respectively. Later, Kasperski et al. [89] used a pin on disc reciprocating flat geometry and an alumina ball as a counterpart for studying the tribological behaviour of 3Y-TZP/CNT composites in the range composition (0.55-5.16 wt % CNT). They reported that the composite with 5.16 wt % CNT had a coefficient of friction 3.8 lower than of monolithic zirconia and explained the obtained results by the exfoliation of MWCNTs due to the high shear stresses generated during sliding. In the current thesis we have performed detailed studies of the frictional behaviour of SPSed 3Y-TZP/ CNT composites where a range of CNT content from 0 to 2 wt % were added to a zirconia matrix. It was found that the incorporation of 2 wt % CNT results in high wear resistance which to the best of our knowledge has never been reported before [90]. More details are reported in Chapter 4.

2.3

Electrical properties

The electrical conductivity of CNTs was found to be as high as 2 107 S/m [80]. The electrical properties of CNTs depends on the type, chirality and defects in the structure.

29

Properties of zirconia-CNT composites Therefore, adding a small amount of CNT to a ceramic matrix which is an insulator can make it a good conductor. SWCNTs have better electrical conductivity properties than MWCNTs due to their perfect structure. The electron tunnelling between adjacent tubes in a percolated network of SWCNTs is believed to be the primary charge transport mechanism in CNT [91] . This mechanism has been associated with fluctuation-assisted tunnelling proposed by Sheng [92]. Sheng reported the observation of a new tunnelling conduction mechanism in carbon-PVC composite in which the modulation of tunnelling barriers by thermal fluctuations determines the dependency of conductivity on temperature and electric field [91]. Fonseca et al. [93] studied the electrical conductivity of 3YTZP/SWCNT composites in wide range of temperature. The authors attributed the high electrical conductivity of the CNT composites to the electron tunnelling between adjacent tubes in a percolated network. The electrical percolation threshold in ceramic-CNT composites was reported to be very low due to the excellent electrical properties of CNTs. The addition of only 0.3 vol % CNTs to SiC matrix results in 75 % reduction of electrical resistivity [94]. Ahmed et al. [95] reported a high electrical conductivity with a percolation threshold close to 0.79 vol % attributed to the high aspect ratio of MWCNTs. A study by Inam et al. [96] reported higher electrical conductivity of alumina-CNT composites as compared to alumina black carbon. The high electrical conductivity was attributed to the large aspect ratio which results in entangled network of conductive paths. Furthermore, the use of SPS technique allowed lower sintering temperature and shorter sintering time which preserves CNTs from damage [66]. Hvizdos et al. [88] studied the effect of the addition of CNF to zirconia matrix. The authors reported an increase of electrical conductivity with the addition of 3 wt % CNF due to the formation of a continuous intergranular network of CNFs. Ukai et al. [97] showed that the addition of only 1 wt % CNTs results in improvement of electrical conductivity due to the formation of a 3D network structure of MWCNTs in the 3YTZP matrix. In the current thesis, the electrical properties of zirconia-CNT composites produced using SPS have been studied. The incorporation of CNTs increases drastically the electrical conductivity of the composites when the percolation threshold is reached, which is only about s 1 wt % CNT [98]. More details are described in Chapter 4.

30

Properties of zirconia-CNT composites

2.4

Electrical discharge machining (EDM)

As been described before in the introduction, 3Y-TZP has a wide range of applications due to their extraordinary properties: high strength, high toughness, good chemical resistance and biocompatibility compared to other oxide ceramics [99]. However, the main drawback of machining zirconia ceramics remain being the brittleness which limits their ability to be machined with conventional techniques such as cubic boron nitride (cBN) or diamond tools [100]. Nowadays, a successful and attractive machining technology based on the use of electrical discharge machining (EDM) allows machining complex shapes of electrical conductive materials with high precision up to the microscale level, being independent of the material hardness. In EDM, series of controlled electrical discharges are generated between a tool electrode and a workpiece immersed in dielectric fluid resulting in removing material from the workpiece mainly through melting, evaporation and spalling [101–103]. Nevertheless, EDM requires an electrically conductive workpiece, which limits its applications to nonconductive ceramics such as ZrO2, Al2O3, Si3N4 and SiC [104–107]. It was reported that adding relatively large amounts of a conductive phase to 3Y-TZP such as WC, TiC, TiN, TiCN or TiB2 [108–111] makes the composite electrically conductive and therefore, susceptible to machining by EDM. During EDM, the temperature raises and the melting temperature can be eventually reached which contributes to material removal. Some of the molten material disappears and some fraction of the melt solidifies on the EDM surface forming the recast layer. Melting has been observed mainly in metals as the main material removal mechanism [112]. However, Bonny et al. [113] reported full melting as material removal mechanism in the EDM of ZrO2-WC composites. Melting was also observed in ZrO2–TiN and Al2O3–SiC–TiC using wire EDM [108]. Lauwers et al. [114] reported the effect of microstructure on wire EDM in zirconia where 40 vol % WC was used as conductive phase. It was found that a finer microstructure results in a lower thermal conductivity and then a higher cutting speed for materials where the predominant material removing mechanism (MRM) is melting. Another approach consists of using a conductive layer referred as assisting electrode (AE) and was used on the surface of the workpiece to make it conductive. In the current study, we investigated the machinability of zirconia-MWCNT using 1 wt % and 2 wt % CNT content. The composites showed high electrical conductivity and then a successful EDM machining. Moreover, the study of the surface integrity of the machined samples showed that melting/evaporation and spalling are the main MRM [98]. More details are found in Chapter 5.

31

Properties of zirconia-CNT composites

2.5

Thermal conductivity

Carbon nanotubes have a high thermal conductivity due to the large phonon mean free path in the strong carbon sp2 bond network of CNT walls. However, there is a relatively large data scatter in the literature for thermal conductivity values of CNTs at room temperature. The commonly reported values for the thermal conductivity of individual CNTs are 3,000 W mK−1 for MWCNTs and 3,500 W mK−1 for singlewalled carbon nanotubes (SWCNTs) at room temperature [115]. Bakshi et al. [116] showed that the increase of CNTs dispersion quality induces an increase in the thermal conductivity of Al2O3 coatings with 4 wt% MWCNTs. Kumari et al. [117], reported that the thermal properties of Al2O3-CNT composites are highly dependent on the CNT content, bulk density, and SPS conditions. However, in both studies [116,117] , the maximum increase in thermal conductivity was not as high as expected regarding the thermal conductivity of CNT. Zhang et al. [118] studied the thermal conductivity of SPSed bulk CNT samples. They showed that CNT results in low thermal conductivity (4.2 W mK−1) when they are sintered to bulk samples. The tube-tube interaction was the main reason behind the low thermal conductivity. Huang et al. [119] reported a decrease in thermal conductivity of BaTiO3-CNT composites due to the interfacial thermal barrier between CNTs and BaTiO 3. It was reported that the addition of MWCNTs increases the thermal conductivity of silica-MWCNT composites [120]. Jiang and Gao [121] found that the thermal conductivity increases with increasing the amount of MWCNTs in TiN-MWCNT composites. In the present work, the thermal conductivity of zirconia-MWCNTs sintered at two different temperatures 1350 °C and 1500 °C have been studied. To the best of our knowledge, there has not been any work on the thermal conductivity of 3YTZP/CNT before. It was shown that the thermal conductivity of the studied composites (1 wt % CNT) sintered at 1350 °C is decreasing with increasing MWCNTs content, whereas only a small decrease in the composite with 1 wt % CNT content is observed at higher SPS temperature 1500 °C. The decrease of thermal conductivity with CNT content may be explained by the increase in porosity and the drop of elastic modulus [98].

32

CHAPTER 3 Materials and Methods 3.1

Materials

Commercially available yttria stabilized zirconia powder with 3 mol % yttria (TZ3YSB-E, Tosoh Co, Japan) was used in the current study. The zirconia powder has a crystalline size of 30 nm and a particle size of 600 nm. The chemical composition of 3Y-TZP is illustrated in Table 3.1. Table 3.1 Chemical composition of 3Y-TZP powder as given by the producer [122].

Elements

Composition in wt %

Y2O3

4.95-5.35

Al2O3

0.15-0.35

SiO2

Max 0.02

Fe2O3

Max 0.01

NaO2

Max 0.04

The as received 3Y-TZP powder, see Table 1, was heat treated at 750 °C in order to burn of the organic additives. The carbon nanotubes used to reinforced zirconia were MWCNT (Graphistrength C100) supplied from Arkema, France, with an outer diameter 10-15 nm, inner diameter 2-6 nm and length 0.1-10 m (see Fig.3.1). The different amount of MWCNT contents chosen to reinforce zirconia were 0.5, 1 and 2 wt %. 33

Materials and Methods

Fig.3.1 Shape of as received MWCNTs bundles and zirconia powder.

3.2 Powder processing and sintering Multi-walled carbon nanotubes were initially dispersed in N, N dimethylformamid (DMF) using ultrasonication from 2 to 4 hours depending on the amount of MWCNT used. The heat treated zirconia powder was then added to the dispersed MWCNTDMF solution. After that, the mixture was milled using a planetary ball milling for 4 hours running at a velocity of 300 rpm using a mixture of different sized zirconia balls of diameter 10, 5 and 3 mm. The milled slurry was then dried at 70 °C for 24 hours. Finally, the powder-fibre mixture was passed through sieve of 250 µm. Another method of the composite powder processing was used in order to improve the dispersion of CNT within zirconia matrix. The processing method consists of dry mixing MWCNTs together with zirconia powder using "Turbula" mixer for 24 hours. However, the dry powder processing method was not as successful as the first method which consists of using DMF as a dispersant. Sintering is the next step after powder processing. The composite powder was sintered using Spark Plasma Sintering HPD 25/1 (FCT System, Germany) furnace at Queen Mary college, London, UK. SPS allows sintering to fully dense composites at lower temperatures and at shorter holding times compared with normal sintering, which preserves CNTs against damage as reported in section 1.3.2. The powder-fibre was filled into a graphite die and then heated with a heating rate of 100 °C/min to the holding temperature of 1350 °C. An axial pressure of 50 MPa was applied at 1350 °C for a dwell time of 5 min and then cooled down at a rate of 100 °C /min. After the sintering procedure, the final samples had a shape of flat discs with 3 mm thickness and a diameter of 40 mm and 50 mm. The samples were then polished using series of steps of diamond suspensions from 30 µm down to 3 m before a final polishing step with colloidal silica was made. The samples were then thermally etched 34

Materials and Methods in air at 1100 °C with a dwell time of 1 hour in order to reveal grain sizes and microstructure. The average grain size of each composite was determined using the line intercept method on scanning electron microscopy (SEM) images.

3.3

Microstructural analysis

3.3.1 Density The bulk density of 3Y-TZP/CNT composites was determined using Archimedes method. Since there was no reaction between CNTs and 3Y-TZP matrix, the theoretical density of the composites was calculated according to the rule of mixtures. The theoretical density of zirconia is 6.1 g·cm-3. However, the exact density of MWNTs is not known because it depends on their purity, number of walls and external diameter. Therefore, the value 1.8 g·cm-3 was considered for the density of CNTs based on literature data [123].

3.3.2 Confocal microscope A laser confocal microscope ( Olympus, LEXT OLS 3100) was used to characterize the Vickers indentations by measuring the crack lengths and the diagonal lengths of the imprints.

3.3.3 Scanning Electron Microscope (SEM) A Scanning Electron Microscope (JEOL JSM 6400) and a high resolution scanning electron microscope (Magellan 400, FEI Company) were used for initial microstructural characterization such as grain size measurements, the analysis of CNTs dispersion and fractography examinations. The sample surfaces were coated with a thin layer of carbon before introducing the samples in the SEM chamber to improve the electric conductivity of the sample surface for a better imaging of surface structures in the SEM.

3.3.4 Focused Ion Beam (FIB) Focus ion beam (FIB) operates in a similar way as the SEM. Both systems used focused beam to create a specimen image, ion beam for the FIB, an electron beam for the SEM. A reservoir of gallium (Ga) is positioned in contact with a sharp Tungsten (W) needle. A high extraction field is used to pull the liquid Ga into a sharp cone whose radius may be 5–10 nm. Ions are emitted and then accelerated down the FIB column. The reasons behind the use of Ga are that Ga has a melting temperature of only 30 °C, thus, it exists 35

Materials and Methods in the liquid state near room temperature. Moreover, Ga ions can be focused to a very small probe area of only 10 nm in diameter. The Ga+ ion beam is focused onto the area of interest on the surface. Once the beam hits the surface it enters the sample and creates cascade of events which results in the ejection of sputtered particles in the form of ions that leave the surface as secondary ions (i+ or i-) or neutral atoms (n0), as well as secondary electrons (e–), see Fig 3.2.The electron beam is used to raster the surface while the signals from the emitting ions are used to form the image. The primary ion penetration depth is approximately 20 nm for 25 keV Ga+. The high beam currents are used for milling large areas quite rapidly, and low beam currents are used for polishing the milled surface before imaging. FIBs typically operate with an accelerating voltage between 5 and 50 keV [124].

Fig.3.2 Principle of FIB imaging. In this thesis a dual beam focused ion beam (FIB) /SEM Microscope (Zeiss Neon 40) was used mainly to investigating the subsurface damage below nanoindentation and in front of the notch tip induced by laser ablation. A thin platinum layer was deposited on the sample prior to FIB machining in order to minimize ion-beam damage. A Ga+ ion source was used to mill the surface at a voltage of 30 kV. The final polishing of the cross-sections was performed at 10 pA.

36

Materials and Methods

3.4

Phase characterization

3.4.1 X-Ray Diffraction (XRD) The X-ray diffraction (XRD) was used to identify the crystallographic phases with Bragg-Brentano symmetric-geometry, using PANalytical Empyrean equipment with PIXcel-3D detector and Cu-Kα 45 kV and 30 mA radiation. The XRD spectra were obtained in a scan range of 20 ° ≤ θ ≤ 100 °, using a step size of 0.013 ° and an antiscatter slit of 1 °. The monoclinic volume fraction, , content was calculated by the equation proposed by Toraya et al. [125], ( ( (

)

(

(̅

(̅ )

) (

( )

)

)

) (

)

(

)

)

where and represent the intensities of the (hkl) peak of the monoclinic and tetragonal phases, respectively.

3.4.2 Raman spectroscopy In the current study, Raman spectroscopy (alpha300RA+ from WITec GmbH) was used at Universitat de Barcelona (UB) to analyse the effect of MWCNTs on the tribological properties of the composites where signals from both inside and outside the wear tracks were compared. Raman spectra were recorded using the 532 nm laser wavelength excitation and an acquisition range from 100 to 3000 cm-1.

3.5

Mechanical testing

3.5.1 Hardness and indentation fracture toughness Vickers indenter was used to indent the specimen´s surface inducing cracks at the corners of the residual indentation impressions. The crack lengths and the imprint area were then measured to determine the Vickers hardness (HV) and the indentation fracture toughness. Taking into account the Palmqvist morphology of the indentation cracks, the expression proposed by Niihara [73] for Palmqvist cracks is the most appropriate for determining the indentation fracture toughness, see Fig.3.3. Anstis et al. [72] equation for median cracks was also used to compare the present results with other published results as well as to compare the resistance to cracking by a Vickers indenter of composites with different amounts of CNTs. Moreover, in the literature, the equation 37

Materials and Methods for fracture toughness of median radial cracks is often employed independently of the shape of the cracks.

Fig.3.3 Vickers indentation with Palmqvist cracks. Vickers Hardness of the material is given by the following equation:

Fracture toughness

(

)

(

)

(

)

using the equation of Niihara et al [73] √

( )

(

)

Fracture toughness using Anstis et al equation [72] ( )

(

)

where refers to the crack length and is the elastic modulus and is the hardness calculated from the indentation load and the projected area of the imprint as follows:

38

Materials and Methods (

)

Indentation fracture toughness does not measure the true fracture toughness but rather the resistance to cracking by sharp contact loading as it was well explained in section 2.1.1. Therefore, in the current study, a new method is developed for the determination of the "true" fracture toughness of the nanocomposites.

3.5.2 True fracture toughness The method is based on inducing a sharp shallow notch on the surface of a prismatic bar by means of a femtolaser. The femtolaser system used to produce the notch was a commercial Ti: Sapphire oscillator (Tsunami, Spectra Physics) and a regenerative amplifier system (Spitfire, Spectra Physics) based on chirped pulsed amplification. The femtolaser delivered 120 fs linearly polarized pulses at 795 nm wave length with a repetition rate of 1 kHz. The pulse energy used was 5 mJ and the focusing system was an achromatic doublet lens with 50 mm focal length. The samples were placed on a XYZ motorized stage and moved along one of the horizontal axis with a scanning speed of 50 µm/s. Four passes were needed to achieve the desired notch depth ( 24 µm). The stress intensity factor, , was obtained by using the following expressions proposed by Munz and Fett [126]: (

√

[

)

√

(

)]

(

)

(

)

where and are the outer and inner spans respectively, is the thickness, the width, the applied load, is the crack length, which is taken as the depth of the ⁄ and notch plus the small damage region in front of the notch, . The tests were performed for the composites with 0, 0.5 and 2 wt % CNT content. Additional tests were carried out in standard 3Y-TZP with larger grain size (330 nm). Finally, the notched bar specimens (4 mm x 2.5 mm x 40 mm) were tested in a four point bending test device (DEBEN, Microtest, UK) in air with spans of 30/12 mm. The average stress rate was 2.4 MPa/s. Three specimens were used for each composition.

39

Materials and Methods

3.5.3 Nanoindentation Instrumented indentation is a technique in which an indenter is applied on the surface and both the applied load and the measured displacement are recorded simultaneously throughout the experiment. When the applied displacements and loads are very low the instrumented indentation technique is often referred to as nanoindentation. It is used for the determination of the near surface mechanical properties such as elastic modulus and hardness, among many other applications such as in thin films. The most used method to extract the mechanical properties is the one introduced by Olivier and Pharr in 1992 [127]. The experiment set up requires a sample with a flat surface and an indenter usually made of diamond that penetrates perpendicularly into the surface. The depth of penetration at a given load is a response to the material´s resistance to deformation. The most commonly used indenter geometry is Berkovich geometry which consists of three sided pyramid with an opening angle of 142.3 ° between one edge and the opposing face of the indenter. Depth from nanometres to millimetres can be covered to determine materials properties over a large or a local volume at the nanos ale. “Nanos ale” has been defined by the I O standard I O14577 as 0–200 nm in penetration depth, “mi ros ale” from 00 nm to an applied for e of N, and “ma ros ale” for for es larger than N (I O-14577-1 2002) [128]. The measured parameters during nanoindentation test are the applied load ( ), the total displacement relative to the initial undeformed surface ( ), the contact depth ( ), ⁄ . The contact stiffness is the final depth ( ), and the contact stiffness defined as the slope of the upper portion of the unloading curve during the initial stages of unloading, see also Fig.3.4.

Fig.3.4 Schematic of the a) unloading process showing the contact geometry and b) loaddisplacement curve [127].

40

Materials and Methods The contact depth is estimated from the load-displacement data using the following equation: (

)

(

)

where is the peak indentation load and is a constant which equal to 0.75 for a Berkovich indenter [128]. The projected area is estimated by evaluating the indenter shape function at the ( ). Then the hardness contact depth , and the effective elastic modulus are calculated as follows: (

√ where

)

is a constant equal to 1.034 in the case of a Berkovich indenter.

(

)

The effective elastic modulus, , takes into account the fact that elastic displacements occur in both the specimen with Young´s modulus E and Poisson´s ratio ν, and the indenter with and . In this thesis, nanoindentation tests were performed using a Nanoindenter XP from Agilent Technologies equipped with continuous stiffness measurements (harmonic displacement 2 nm and frequency of 45 Hz), using a pyramidal Berkovich diamond indenter. The strain rate was held constant at 0.05 s-1. The nanoindentation curves were analysed using the Oliver and Pharr method [127] in order to measure the ) and the elastic modulus ( ) as a function of the nanoindentation hardness ( penetration depth for each composition where was defined as the ratio of load and contact area at maximum load. The indenter shape was carefully calibrated for true penetration depths as small as 50 nm by indenting fused silica samples of well-known Young’s modulus (7 GPa). The indents were organized in a regularly spa ed array of 25 indentations (5 by 5) at a maximum penetration depth of 2000 nm or until reaching the maximum applied load, 650 mN. Each indentation was performed with a spacing distance of 50 μm in order to avoid any overlapping effect.

41

Materials and Methods

3.6

Tribological characterization

3.6.1 Scratch testing The scratch test can be carried out under a constant load and an incremental load. During scratch testing a stylus is moved over a specimen surface with a linearly increasing load until failure occurs at critical load , and the normal force and the tangential force are recorded. Acoustic Emission (AE) is also measured. An optical microscope is used to evaluate the damage induced after scratch testing, see Fig 3.5.

Fig.3.5 Schematic showing the different parameters involved during scratch test [129]. In the current study, nano-scratch tests were carried out with a nano-scratch attachment of the Nanoindenter XP that allows lateral force measurements. A Berkovich indenter was employed to scratch the surface under increasing load at a velocity of 10 μm/s for a total scratch length of 500 μm up to a maximum load of 40 mN. Three different scratches were performed on each sample. Lateral (friction) forces were calculated from the deflection of the loading column. The coefficient of friction (COF) was determined by taking the ratio of the lateral force measured by the equipment and the normal load applied on the material. Macro-scratch testing was conducted using a sliding Rockwell indenter with a diamond spherical tip radius of 200 m (automatic Scratch Tester, CSM-Instruments, Switzerland). Both normal and tangential forces were recorded. Applied load ranged from 1 to 150 N over a sliding distance of 7.5 mm at a sliding speed 5 mm/min.

3.6.2 Wear test Tribology tests were performed on an automatic tribometer (Wazau TRM1000, Germany) in reciprocating dry sliding conditions, using ball on disc geometry, at ambient temperature and pressure. A zirconia ball with 10 mm diameter was used as a 42

Materials and Methods counterpart. All the tests were carried out under a constant normal load of 5 N, a sliding velocity of 300 rpm and a stroke length of 4 mm. The total sliding distance was 100 m. At least four tests were performed for each condition and data represent their average. The COF was calculated by taking the ratio of the tangential and normal forces and it was reported versus the sliding distance. The volume removed was measured using a stylus profilometer where a hundred profiles along the width of the track were recorded. The wear rate, , was determined in terms of the volume loss per distance and applied load according to the following equation: (

3.7

)

Electrical conductivity and Electrical Discharge machining (EDM) characterization

In the current work, the electrical conductivities of all composites were measured by four-point method. A direct current runs through the sample through outer Pt current leads and the resistivity of the sample is measured using two "inner" Ni Voltage probes. The measurement was done with an accurate digital micro-ohmmeter (Keithley 580) using the four-point method on bar-shaped specimens (4 mm x 4 mm x 18 mm) prepared using a diamond cutting machine. The conductivity of the material is calculated from the current, the measured voltage and the geometry of the arrangement. Die sinking DM (“+GF+ AgieCharmilles”) was carried out with 3Y-TZP/ 1 wt% CNT and 3Y-TZP/ 2 wt% CNTs composites as workpiece. The tool used was a copper tool electrode with a surface of 0.12 cm². The average observed current was 0.5 A and the overall machining distance was set to a value of without rotation or planetary movement of the tool. The dielectric used lubricating oil named "IonoPlus IME-MH". To achieve a surface roughness of Ra=0.1, first impulse set 1 and afterwards impulse set 2 of the machine were used to machine the material under the boundary condition “Cu – Carbide Metal”. The omposites subje ted to DM were in the shape of one quadrant of the sintered discs, clamped by the sides and machined on one diametric plane.

3.8

Thermal conductivity characterization

In the present thesis, the thermal conductivity ( ) was calculated in the composites with 0.5, 1 and 4 wt % CNT by measuring the thermal diffusivity ( ), the density ( )

43

Materials and Methods and the specific heat capacity ( the equation below:

), for a temperature range of 25-950 °C, according to

(

)

The thermal diffusivity was determined by laser-flash method (Laser Flash Apparatus, Netzsch LFA 457, Germany) where disk-shaped specimens were thin coated with graphite, ASTM E2585-09 (2009) [130]. The specific heat capacity was measured under argon atmosphere using a differential scanning calorimeter (Netzsch DSC 404C, Germany). The measurements were carried out in Technical University of Denmark (DTU).

44

CHAPTER 4 Conclusions and Future work 4.1 Conclusions The processing and the properties of zirconia-MWCNT have been studied and the following conclusions can be made: 1. The use of SPS for sintering zirconia and zirconia-MWCNT composites results in high density ceramics with smaller grain size than those produced under conventional conditions by compacting and sintering. The addition of MWCNT decreases further the grain size. As the concentration of MWCNT increases above 0.5 % MWCNT, no full density is achieved and hardness and elastic modulus decrease. 2. The friction coefficient extracted from nano-scratch and wear tests decreases with the addition of MWCNTs. In macro-scratch tests using higher loads, there is a critical load over which the friction coefficient increases with CNT addition and brittle fracture starts. The low friction coefficient is attributed to the formation of a carbon tribo-film during reciprocation sliding. 3. A novel method for the calculation of the "true" fracture toughness of small cracks is successfully established. A very sharp notch in the scale of a few tens of microns and sub-micrometer tip radius less than 0.5 μm was induced in zirconia-MWCNT composite using Ultra-short Pulsed Laser Ablation (UPLA). 4. The "true" fracture toughness of zirconia and zirconia-MWCNT for small cracks is found to be low and practically independent of the CNT amount. Vickers indentation fracture toughness is higher and increases with increasing CNT. In this way it is shown that indentation fracture toughness is not an appropriate method for analysing the influence of MWCNT on true fracture toughness. Even in the absence of MWCNT,

45

Conclusions and Future work the fracture toughness of SPS zirconia measured from small surface cracks induced by UPLA is found to be lower than measured by conventional fracture toughness methods with long cracks. The increase observed in indentation fracture toughness with the addition of MWCNT is attributed to an increase in the resistance to cracking under sharp contact loading because of change in the elasto-plastic behaviour induced by the addition of MWCNT. 5. The addition of MWCNTs strongly increases the electrical conductivity of the studied composites. While, the thermal conductivity increases slightly with increasing CNT content and SPS temperature. 6. The composites with 1 wt % and 2 wt % MWCNT content can successfully be machined using EDM due to their high electrical conductivity. The material removal mechanism is found to be melting/evaporation and spalling. 7. A study of the effect of grinding on the SPS zirconia samples with a wide range of grain sizes shows that a thermal etching of ground zirconia at 1100 °C for 1 hour in air induces a nano-grain layer with grains of about 60 nm and a thickness of less than a few hundred nanometers, which is independent of the original grain size. At much higher temperature, 1575 °C for 1 hour, thermal etching of ground or polished zirconia induces similar grain size. 8. The influence of transformation toughening in the behaviour of small cracks was also conducted in conventionally sintered 12Ce-ZrO2 (300 nm grain size) because it has much higher plateau fracture toughness than 3Y-TZP (177 nm grain size). By inducing a similar short sharp notch using UPLA in both materials, the difference in strength is only about 10%. This means that the true fracture toughness difference for these small cracks is also less than about 10 % in comparison with plateau values of fracture toughness which are different by a factor ≈3. Therefore, the beneficial effect of higher indentation fracture toughness in 12Ce-ZrO2 reported in literature has a very small effect on the real effective fracture toughness that determines the strength of unshielded small cracks.

4.2 Future work 1. Dispersion is a crucial step in the processing of zirconia-CNT composites. Different dispersion solvents of CNTs should be studied and an optimized method for dispersion of MWCNTs in ceramic matrix composites should be established. Interfacial bonding should be improved by coating zirconia with functionalised CNTs for example.

46

Conclusions and Future work Homogeneous dispersion and interfacial bonding may increase the mechanical properties. 2. The addition of CNT up to 4 wt % results in high electrical conductivity, reported for the first time, in zirconia-MWCNT composites. It would be interesting to investigate the potential of using higher CNT content and high temperatures on the electrical conductivity and on the machining of the composites. 3. CNTs are in the nanoscale and they are located in the grain boundaries on zirconia matrix. Oxidizing CNTs leaves a porous nanoscale network in zirconia matrix. It would be of great interest to use CNTs to produce porous ceramic materials with nanosized long channels. 4. Since the thermal etching after grinding SPSed zirconia results in nanograin layer with few hundred nanometers in thickness independently of the bulk grain size, a study of LTD resistance of thermal etched ground zirconia samples with large bulk grain size could be interesting in order to see if this nano-layer protects large bulk grain size zirconia from LTD. 5. The use of SPS allows sintering the studied composites in short time and low temperature which results in small grain size. An alternative methods as Hot Isostatic Pressing (HIP) could be also used to compare with SPS and gain deeper understanding of the sintering mechanism.

47

48

REFERENCES [1] J. Chevalier, L. Gremillard, S. Deville, Low-Temperature Degradation of Zirconia and Implications for Biomedical Implants, Annu. Rev. Mater. Res. 37 (2007) 1-32. [2] P. Li, I.-W. Chen, J.E. Penner-Hahn, Effect of Dopants on Zirconia StabilizationAn X-ray Absorption Study: I, Trivalent Dopants, J. Am. Ceram. Soc. 77 (1994) 118–128. [3] J. Chevalier, L. Gremillard, A. V. Virkar, D.R. Clarke, The Tetragonal-Monoclinic Transformation in Zirconia: Lessons Learned and Future Trends, J. Am. Ceram. Soc. 92 (2009) 1901–1920. [4] R.H.J. Hannink, P.M. Kelly, B.C. Muddle, Transformation Toughening in Zirconia-Containing Ceramics, J. Am. Ceram. Soc. 83 (2004) 461–487. [5] P.M. Kelly, L.R.F. Rose, The martensitic transformation in ceramics and its role in transformation toughening, J. Am. Ceram. Soc. 47 (2002) 463–557. [6] F.F. Lange, Transformation toughening, J. Mater. Sci. 17 (1982) 225–234. [7] R.M. McMeeking, A.G. Evans, Mechanics of Transformation-Toughening in Brittle Materials, J. Am. Ceram. Soc. 65 (1982) 242–246. [8] P. Palmero, Structural Ceramic Nanocomposites: A Review of Properties and Powders Synthesis Methods, Nanomaterials. 5 (2015) 656–696. [9] A. V. Virkar, R.L.K. Matsumoto, Ferroelastic Domain Switching as a Toughening Mechanism in Tetragonal Zirconia, J. Am. Ceram. Soc. 69 (1986) C–224–C–226. [10] R.H.J. Hannink, P.M. Kelly, B.C. Muddle, Transformation Toughening in Zirconia Containing Ceramics, J. Am. Ceram. Soc. 83 (2004) 461–487. [11] H. Fischer, W. Rentzsch, R. Marx, R-Curve Behaviour of Dental Ceramic, J. Dent. Res. (2002) 547–551. [12] R.O. Ritchie, Mechanisms of fatigue crack propagation in metals, ceramics and composites: Role of crack tip shielding, Mater. Sci. Eng. A. 103 (1988) 15–28. [13] J. Eichler, M. Hoffman, U. Eisele, S. Rul, R-curve behaviour of 2Y-TZP with submicron grain size, J. Eur. Ceram. Soc. 26 (2006) 3575–3582. [14] N. Gupta, P. Mallik, B. Basu, Y-TZP ceramics with optimized toughness : new results, J. Alloy. Compd. 379 (2004) 228–232. [15] G. Fargas, D. Casellas, L. Llanes, M. Anglada, Thermal shock resistance of yttriastabilized zirconia with Palmqvist indentation cracks, J. Eur. Ceram. Soc. 23 (2003) 107–114. 49

References [16] H. Hobner, Sub-critical crack extension and crack resistance in polycrystalline alumina, Mater. Sci. 12 (1977) 117–125. [17] D. Munz, What can we learn from R-curve measurements, J. Am. Ceram. Soc. 90 (2007) 1–15. [18] K. Kobayashi, H. Kuwajima, T. Masaki, Phase change and mechanical properties of ZrO2-Y2O3 solid electrolyte after ageing, Solid State Ionics 4 (1981) 489–493. [19] R.L.K. Matsumoto, Strength Recovery in Degraded Yttria-Doped Tetragonal Zirconia Polycrystals, J. Am. Ceram. Sci. 213 (1985) 1984–1986. [20] E.M. Santos, S. Vohra, S.A. Catledge, M.D. Mcclenny, J. Lemons, K.D. Moore, Examination of Surface and Material Properties of Explanted Zirconia Femoral Heads, J. Arthroplasty. 19 (2004) 30–34. [21] M. Yoshimura, T. Noma, K. Kawabata, S. Somiya, Role of H 2O on the degradation process of Y-TZP, J. Mater. Sci. Lett. 6 (2000) 465–467. [22] J. Chevalier, B. Cales, J.M. Drouin, Low-Temperature Aging of Y-TZP Ceramics, J. Am. Ceram. Sci. 54 (1999) 2150–2154. [23] J. Chevalier, What future for zirconia as a biomaterial, Biomaterials. 27 (2006) 535–543. [24] E. Djurado, L. Dessemond, C. Roux, Phase stability of nanostructured tetragonal zirconia Polycrystals versus temperature and water vapour, Solid State Ionics 137 (2000) 1249–1254. [25] R. Chintapalli, A. Mestra, F. García Marro, H. Yan, M. Reece, M. Anglada, Stability of Nanocrystalline Spark Plasma Sintered 3Y-TZP, Materials 3 (2010) 800–814. [26] A. Paul, B. Vaidhyanathan, J.G.P. Binner, Hydrothermal Aging Behaviour of Nanocrystalline Y-TZP Ceramics, J. Am. Ceram. Sci. 2152 (2011) 2146–2152. [27] L. Ruiz, M.J. Readey, Effect of heat treatment on grain size, phase assemblage and mechanical properties of 3 mol % Y-TZP, J. Am. Ceram. Sci. 79 (1996) 23312340. [28] K. Matsui, H. Horikoshi, N. Ohmichi, M. Ohgai, H. Yoshida, Y. Lkuhara, Cubic-Formation and Grain-Growth Mechanisms in Tetragonal Zirconia Polycrystal, J. Am. Ceram. Sci. 8 (2003) 1401–1408. [29] S. Deville, M. Etienne, R. Jullian, Critical effect of cubic phase on aging in 3 mol % yttria-stabilized zirconia ceramics for hip replacement prosthesis, 25 (2004) 5539– 5545. [30] S. Schmauder, H. Schubert, Significance of Internal Stresses for the Martensitic Transformation in Yttria-Stabilized Tetragonal Zirconia Polycrystals during Degradation, J. Am. Ceram. Sci. 69 (1985) 534–540. [31] J.F. Jue, J. Chen, A. V Virkar, low-Temperature aging of t’-Zirconia: The Role of Microstructure, J. Am. Ceram. 20 (1991) 1811–1820.

50

References [32] M. Guazzato, M. Albakry, S.P. Ringer, M. V Swain, Strength, fracture toughness and microstructure of a selection of all-ceramic materials. Part II. Zirconia-based dental ceramics, Dent. Mat. (2010) 449–456. [33] I. Denry, J.R. Kelly, State of the art of zirconia for dental applications, Dent. Mat. 4 (2007) 299–307. [34] W. Krätschmer, L.D. Lamb, K. Fostiropoulos, D.R. Huffman, Solid C60: a new form of carbon, Nature. 347 (1990) 354–358. [35] S. Iijima, Helical microtubules of graphitic carbon, Nature. 354 (1991) 56–58. [36] S. Iijima, T. Ichihashi, Single-shell carbon nanotubes of 1 nm diameter, Nature. 363 (1993) 603–605. [37] F. Inam, Development of Ceramic-Carbon Nanotube (CNT) Nanocomposites, PhD thesis submitted to School of Engineering and Materials Science Queen Mary, University of London, 2009. [38] Y. Saito, Carbon Nanotube and Related Field Emitters, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2010. [39] E.T. Thostenson, Z. Ren, T.-W. Chou, Advances in the science and technology of carbon nanotubes and their composites: a review, Compos. Sci. Technol. 61 (2001) 1899–1912. [40] X. Zhao, M. Ohkohchi, M. Wang, S. Iijima, T. Ichihashi, Y. Ando, Preparation of high-grade carbon nanotubes by hydrogen arc discharge, Carbon N. Y. 35 (1997) 775–781. [41] X. Zhao, M. Wang, M. Ohkohchi, Y. Ando, Morphology of Carbon Nanotubes Prepared by Carbon Arc, J. Appl. Phys. 35 (1996) 4451–4456. [42] M. Wang, X. Zhao, M. Ohkohchi, Y. Ando, Carbon Nanotubes Grown on the Surface of Cathode Deposit by Arc Discharge, Fullerene. Sci. Tech. 4 (1996) 1027– 1039. [43] D.S. Bethune, C.H. Klang, M.S. de Vries, G. Gorman, R. Savoy, J. Vazquez, et al., Cobalt-catalysed growth of carbon nanotubes with single-atomic-layer walls, Nature. 363 (1993) 605–607. [44] Y. Saito, M. Okuda, N. Fujimoto, T. Yoshikawa, M. Tomita, T. Hayashi, SingleWall Carbon Nanotubes Growing Radially from Ni Fine Particles Formed by Arc Evaporation, J. Appl. Phys. 33 (1994) L526–L529. [45] Y. Saito, K. Nishikubo, K. Kawabata, T. Matsumoto, Carbon nanocapsules and single-layered nanotubes produced with platinum-group metals (Ru, Rh, Pd, Os, Ir, Pt) by arc discharge, J. Appl. Phys. 80 (1996) 3062. [46] B. Chen, X. Zhao, S. Inoue, Y. Ando, Fabrication and Dispersion Evaluation of Single-Wall Carbon Nanotubes Produced by FH-Arc Discharge Method, J. Nanosci. Nanotechnol. 10 (2010) 3973–3977. [47] J. Robertson, H. Kawarada, E. Kohn, R. Hamers, B. Chen, S. Inoue, et al., Raman spectroscopic and thermogravimetric studies of high-crystallinity SWNTs synthesized by FH-arc discharge method, Diam. Relat. Mater. 18 (2009) 975–978. 51

References [48] W. Fan, J. Zhao, Y. Lv, W. Bao, X. Liu, Synthesis of SWNTs from charcoal by arc-discharging, J. Wuhan Univ. Technol. Sci. Ed. 25 (2010) 194–196. [49] J. Prasek, J. Drbohlavova, J. Chomoucka, J. Hubalek, O. Jasek, V. Adam, et al., Methods for carbon nanotubes synthesis—review, J. Mater. Chem. 21 (2011) 15872. [50] Y. Watanabe, T. Asano, H.Y. Chang, H. Fujiyama, R. Mc Williams, T. Watanabe, et al., Production of single-wall carbon nanotubes by a XeCl excimer laser ablation, Thin Solid Films. 506 (2006) 255–258. [51] J. Robertson, N. Fotopoulos, J.P. Xanthakis, A molecular level model for the nucleation of a single-wall carbon nanotube cap over a transition metal catalytic particle, Diam. Relat. Mater. 19 (2010) 557–561. [52] O. Lee, J. Jung, S. Doo, S.-S. Kim, T.H. Noh, K.-I. Kim, et al., Effects of temperature and catalysts on the synthesis of carbon nanotubes by chemical vapour deposition, Met. Mater. Int. 16 (2010) 663–667. [53] A.S. Afolabi, A.S. Abdulkareem, S.D. Mhlanga, S.E. Iyuke, Synthesis and purification of bimetallic catalysed carbon nanotubes in a horizontal CVD reactor, J. Exp. Nanosci. 6 (2011) 248–262. [54] S. Dumpala, J.B. Jasinski, G.U. Sumanasekera, M.K. Sunkara, Large area synthesis of conical carbon nanotube arrays on graphite and tungsten foil substrates, Carbon N. Y. 49 (2011) 2725–2734. [55] E. Flahaut, C. Laurent, A. Peigney, Catalytic CVD synthesis of double and triplewalled carbon nanotubes by the control of the catalyst preparation, Carbon N. Y. 43 (2005) 375–383. [56] S.C. Lyu, T.J. Lee, C.W. Yang, C.J. Lee, Synthesis and characterization of highquality double-walled carbon nanotubes by catalytic decomposition of alcohol, Chem. Commun. (2003) 1404. [57] H. Cui, G. Eres, J.Y. Howe, A. Puretkzy, M. Varela, D.B. Geohegan, et al., Growth behaviour of carbon nanotubes on multi-layered metal catalyst film in chemical vapour deposition, Chem. Phys. Lett. 374 (2003) 222–228. [58] J. Hilding, E. a. Grulke, Z. George Zhang, F. Lockwood, Dispersion of Carbon Nanotubes in Liquids, J. Dispers. Sci. Technol. 24 (2003) 1–41. [59] S. Tsang, Mechanical damage of carbon nanotubes by ultrasound, Carbon N. Y. 34 (1996) 814–816. [60] K.B. Shelimov, R.O. Esenaliev, A.G. Rinzler, C.B. Huffman, R.E. Smalley, Purification of single-wall carbon nanotubes by ultrasonically assisted filtration, Chem. Phys. Lett. 282 (1998) 429–434. [61] M.H. Bocanegra-Bernal, Hot isostatic pressing (HIP) technology and its applications to metals and ceramics, J. Mater. Sci. 39 (2004) 6399–6420. [62] M. Koizumi, Hot Isostatic Pressing— Theory and Applications: Proceedings of the Third International Conference Osaka, Japan, 10-14 June 1991, Springer Science & Business Media, 2012. 52

References [63] R. Schaefer, Hot isostatic pressing : theory and applications : proceedings of the Second International Conference, Gaithersburg, Maryland, USA, 7-9 June 1989, ASM International, Materials Park. Ohio, 1991. [64] R. Tegman, Hot isostatic pressing of silicon nitride, Int. Ceram. 34 (1985) 22-28. [65] R. Tegman, Hot Isostatic Pressing of Ceramic Waste Forms, Int. Ceram. (1985) 35–38. [66] O. Guillon, J. Gonzalez-Julian, B. Dargatz, T. Kessel, G. Schierning, J. Räthel, M. Herrmann, Field-assisted sintering technology/spark plasma sintering: Mechanisms, materials, and technology developments, Adv. Eng. Mater. 16 (2014) 830–849. [67] G. Bernard-Granger, N. Monchalin, C. Guizard, Comparisons of grain sizedensity trajectory during spark plasma sintering and hot-pressing of zirconia, Mater. Lett. 62 (2008) 4555–4558. [68] J. Langer, M.J. Hoffmann, O. Guillon, Electric Field-Assisted Sintering in Comparison with the Hot Pressing of Yttria-Stabilized Zirconia, J. Am. Ceram. Soc. 94 (2011) 24–31. [69] A.G. Evans, E.A. Charles, Fracture Toughness Determinations by Indentation, J. Am. Ceram. Soc. 59 (1976) 371–372. [70] G.D. Quinn, R.C. Bradt, on the Vickers indentation fracture toughness Test, J. Am. Ceram. Soc. 90 (2007) 673–680. [71] B.R. Lawn, A.G. Evans, D.B. Marshall, Elastic/Plastic Indentation Damage in Ceramics : The Median/Radial Crack System, J. Am. Ceram. Soc. 63 (1980) 574– 581. [72] G.R. Anstis, P. Chantikul, B.R. Lawn, D.B. Marshal, A critical evaluation of indentation techniques for measuring fracture toughness: I Direct crack measurements, J. Am. Ceram. Soc. 46 (1981) 533–538. [73] K. Niihara, A fracture mechanics analysis of indentation-induced Palmqvist crack in ceramics, J. Mater. Sci. Lett. 2 (1983) 221–223. [74] T. Konno, Physical Chem. Chang. Med. Mix. with Adsorbents Solid State. IV. Study Reduced-Pressure Mix. Pract. Use Amorph. Mix. Flufenamic Acid, Chem. Pharm. Bull. 38 (1990) 2003-2007. [75] Z. Li, A. Ghosh, A.S. Kobayashi, R.C. Bradt, Indentation fracture toughness of sintered silicon carbide in the Palmqvist crack regime, J. Am. Ceram. Soc. 72 (1989) 904–911. [76] C.B. Ponton, R.D. Rawlings, Vickers indentation fracture toughness test. Part 1. Review of literature and formulation of standardised indentation toughness equations, Mater. Sci. Technol. 5 (1989) 865–872. [77] A. Ghosh, Z. Li, C.H. Henager, A.S. Kobayashi, R.C. Bradt, Vickers microindentation toughness of a sintered SiC in the median-crack regime, (1991) 1–33.

53

References [78] L. Melk, J.J. Roa Rovira, F. García-Marro, M-L. Antti, B. Milsom, M.J. Reece, Nanoindentation and fracture toughness of nanostructured zirconia/multi-walled carbon nanotube composites, Ceram. Int. 41 (2015) 2453–2461. [79] J.Y. Huang, S. Chen, Z.Q. Wang, K. Kempa, Y.M. Wang, S.H. Jo, G. Chen, M.S. Dresselhaus, Z.F. Ren, Superplastic carbon nanotubes, Nature. 439 (2006) 281. [80] M.M.J. Trea y, T.W. bbesen, J.M. Gibson, x eptionally high Young’s modulus observed for individual carbon nanotubes, Nature. 381 (1996) 678–680. [81] O. Lourie, D. Cox, H. Wagner, Buckling and Collapse of Embedded Carbon Nanotubes, Phys. Rev. Lett. 81 (1998) 1638–1641. [82] G-D. Zhan, J.D. Kuntz, J. Wan, A.K. Mukherjee, Single-wall carbon nanotubes as attractive toughening agents in alumina-based nanocomposites, Nat. Mater. 2 (2003) 38–42. [83] X. Wang, N.P. Pdture, H. Tanaka, Contact-damage-resistant ceramic/single-wall carbon nanotubes and ceramic/graphite composites, Nat. Mater. 3 (2004) 539–44. [84] G.D. Quinn, R.C. Bradt, On the Vickers Indentation Fracture Toughness Test, J. Am. Ceram. Soc. 90 (2007) 673–680. [85] J-W. An, D-H. You, D-S. Lim, Tribological properties of hot-pressed alumina– CNT composites, Wear. 255 (2003) 677–681. [86] Z.H. Xia, J. Lou, W.A. Curtin, A multiscale experiment on the tribological behaviour of aligned carbon nanotube/ceramic composites, Scr. Mater. 58 (2008) 223–226. [87] P. Hvizdoš, V. Pu hý, A. Duszová, J. Dusza, Tribologi al behaviour of arbon nanofiber–zirconia composite, Scr. Mater. 63 (2010) 254–257. [88] P. Hvizdoš, V. Pu hý, A. Duszová, J. Dusza, C. Balázsi, Tribologi al and ele tri al properties of ceramic matrix composites with carbon nanotubes, Ceram. Int. 38 (2012) 5669–5676. [89] A. Kasperski, A. Weibel, D. Alkattan, C. Estournès, V. Turq, C. Laurent, A. Peigney, Microhardness and friction coefficient of multi-walled carbon nanotubeyttria-stabilized ZrO2 composites prepared by spark plasma sintering, Scr. Mater. 69 (2013) 338–341. [90] L. Melk, J.J. Roa Rovira, M-L. Antti, M. Anglada, Coefficient of friction and wear resistance of zirconia–MWCNTs composites, Ceram. Int. 41 (2014) 459-468. [91] L. Berhan, A.M. Sastry, Modeling percolation in high-aspect-ratio fiber systems. I. Soft-core versus hard-core models, Phys. Rev. E - Stat. Nonlinear, Soft. Matter. Phys. 75 (2007) 1–8. [92] J.I.G. Ping Sheng, E.K. Sichel, Fluctuation-Induced Tunneling Conduction in Carbon-Polyvinylchloride, Phys. Rev. Lett. 40 (1981) 2–5. [93] F.C. Fonseca, R. Muccillo, D.Z. De Florio, L.O. Ladeira, A.S. Ferlauto, Mixed ionic-electronic conductivity in yttria-stabilized zirconia/carbon nanotube composites, Appl. Phys. Lett. 91 (2007) 2007–2009. 54

References [94] N. Song, H. Liu, Y.T. Yuan, J.Z. Fang, Fabrication and Characterization of SiCCoated Multi-Walled Carbon Nanotubes Reinforced Reaction Bonded SiC Composite, Adv. Mater. Res. 936 (2014) 176–180. [95] K. Ahmad, W. Pan, S-L. Shi, Electrical conductivity and dielectric properties of multiwalled carbon nanotube and alumina composites, Appl. Phys. Lett. 89 (2006) 133122. [96] F. Inam, H. Yan, D.D. Jayaseelan, T. Peijs, M.J. Reece, Electrically conductive alumina–carbon nanocomposites prepared by Spark Plasma Sintering, J. Eur. Ceram. Soc. 30 (2010) 153–157. [97] T. Ukai, T. Sekino, A.T. Hirvonen, N. Tanaka, T. Kusunose, T. Nakayama, K. Niihara, Preparation and Electrical Properties of Carbon Nanotubes Dispersed Zirconia Nanocomposites, Key. Eng. Mater. 317-318 (2006) 661–664. [98] L. Melk, M-L. Antti, M. Anglada, Material removal mechanisms by EDM of zirconia reinforced MWCNT nanocomposites, Ceram. Int. (2016) 1–10. 57625801. [99] I. Birkby, R. Stevens, Applications of Zirconia Ceramics, Key. Eng. Mater. 122124 (1996) 527–552. [100] O. Malek, J. Vleugels, Y. Perez, P. De Baets, J. Liu, S. Van den Berghe, B. Lauwers, Electrical discharge machining of ZrO2 toughened WC composites, Mater. Chem. Phys. 123 (2010) 114–120. [101] K. Ho, S. Newman, State of the art electrical discharge machining (EDM), Int. J. Mach. Tools Manuf. 43 (2003) 1287–1300. [102] M. Kunieda, B. Lauwers, K.P. Rajurkar, B.M. Schumacher, Advancing EDM through Fundamental Insight into the Process, CIRP Ann. Manuf. Technol. 54 (2005) 64–87. [103] H.C. Tsai, B.H. Yan, F.Y. Huang, EDM performance of Cr/Cu-based composite electrodes, Int. J. Mach. Tools Manuf. 43 (2003) 245–252. [104] A. Banu, M.Y. Ali, M.A. Rahman, Micro-electro discharge machining of nonconductive zirconia ceramic: investigation of MRR and recast layer hardness, Int. J. Adv. Manuf. Technol. 75 (2014) 257–267. [105] A. Muttamara, Y. Fukuzawa, N. Mohri, T. Tani, Effect of electrode material on electrical discharge machining of alumina, J. Mater. Process. Technol. 209 (2009) 2545–2552. [106] K. Liu, J. Peirs, E. Ferraris, B. Lauwers, D. Reynaerts, Micro Electrical Discharge Machining of Si3N4 -based Ceramic Composites, Energy Convers. (2008). [107] Y. Fukuzawa, N. Mohri, T. Tani, A. Muttamara, Electrical discharge machining properties of noble crystals, J. Mater. Process. Technol. 149 (2004) 393–397. [108] B. Lauwers, J.P. Kruth, W. Liu, W. Eeraerts, B. Schacht, P. Bleys, Investigation of material removal mechanisms in EDM of composite ceramic materials, J. Mater. Process. Technol. 149 (2004) 347–352. 55

References [109] K. Bonny, P. De Baets, J. Vleugels, A. Salehi, O. Van der Biest, B. Lauwers, W. Liu, Influence of secondary electro-conductive phases on the electrical discharge machinability and frictional behaviour of ZrO2-based ceramic composites, J. Mater. Process. Technol. 208 (2008) 423–430. [110] S. Salehi, O. Van der Biest, J. Vleugels, Electrically conductive ZrO 2–TiN composites, J. Eur. Ceram. Soc. 26 (2006) 3173–3179. [111] K. Bonny, P. De Baets, J. Vleugels, O. Van der Biest, A. Salehi, W. Liu, B. Lauwers, Reciprocating sliding friction and wear behaviour of electrical discharge machined zirconia-based composites against WC–Co cemented carbide, Int. J. Refract. Met. Hard. Mater. 27 (2009) 449–457. [112] T.H. Hocheng, Machining Technology for Composite Materials. Principles and Practice, Elsevier Science, 2011. [113] K. Bonny, P. De Baets, J. Vleugels, A. Salehi, O. Van Der Biest, B. Lauwers, W. Lieu, EDM machinability and frictional behaviour of ZrO2-WC composites, Int. J. Adv. Manuf. Technol. 41 (2009) 1085–1093. [114] B. Lauwers, K. Brans, W. Liu, J. Vleugels, S. Salehi, K. Vanmeensel, Influence of the type and grain size of the electro-conductive phase on the Wire-EDM performance of ZrO2 ceramic composites, CIRP Ann. Manuf. Technol. 57 (2008) 191–194. [115] A. Balandin, Thermal properties of graphene and nanostructured carbon materials, Nat. Mater. 10 (2011) 569–581. [116] S.R. Bakshi, K. Balani, A. Agarwal, Thermal Conductivity of Plasma-Sprayed Aluminum Oxide—Multiwalled Carbon Nanotube Composites, J. Am. Ceram. Soc. 91 (2008) 942–947. [117] L. Kumari, T. Zhang, G. Du, W. Li, Q.W. Wang, A. Datye. K.H. Wu, Thermal properties of CNT-Alumina nanocomposites, Compos. Sci. Technol. 68 (2008) 2178–2183. [118] H.L. Zhang, J.F. Li, K.F. Yao, L.D. Chen, Spark plasma sintering and thermal conductivity of carbon nanotube bulk materials, J. Appl. Phys. 97 (2005) 10–15. [119] Q. Huang, L. Gao, Multiwalled carbon nanotube/BaTiO nanocomposites: Electrical and rectification properties, Appl. Phys. Lett. 86 (2005) 123104. [120] R. Sivakumar, S. Guo, T. Nishimura, Y. Kagawa, Thermal conductivity in multi-wall carbon nanotube/silica-based nanocomposites, Scr. Mater. 56 (2007) 265–268. [121] L. Jiang, L. Gao, Densified multiwalled carbon nanotubes-titanium nitride composites with enhanced thermal properties, Ceram. Int. 34 (2008) 231–235. [122] http://www.tosoh.com/our-products/advanced-materials/zirconia-powders [123] C. Laurent, E. Flahaut, A. Peigney, The weight and density of carbon nanotubes versus the number of walls and diameter, Carbon N. Y. 48 (2010) 2994–2996. [124] L.A. Giannuzzi, F.A. Stevie, A review of focused ion beam milling techniques for TEM specimen preparation, Micron. 30 (1999) 197–204. 56

References [125] H. Toraya, M. Yoshimura, S. Somiya, Calibration Curve for Quantitative Analysis of the Monoclinic-Tetragonal ZrO2 System by X-Ray Diffraction, Commun. Am. Ceram. Soc. (1984) 119–121. [126] D. Munz, T. Fett, Ceramics: Mechanical Properties, Failure Behaviour, Materials Selection, Springer, 1999. [127] W.C. Oliver, G.M. Pharr, Measurement of hardness and elastic modulus by instrumented indentation: Advances in understanding and refinements to methodology, J. Mater. Res. 19 (2011) 3–20. [128] A. Tiwari, Nanomechanical Analysis of high performance materials, 2013. [129]http://www.adhesivestoolkit.com/Docs/test/MECHANICAL%20TEST%2 METHOD%201%20-%20Continued.xtp / [130] ASTM Standard E2585-09, ASTM Standard E2585-09: Standard Practice for Thermal Diffusivity by the Flash Method, ASTM Int. (2009).

57

58

Part II

59

60

Summary of Papers Paper I Coefficient of friction and wear resistance of zirconiaMWCNTs composites. Summary: The wear rate of 3Y-TZP /3Y-TZP is high in comparison to other ceramic pairs like alumina/alumina because of the surface fracture induced by microcracking during phase transformation. Moreover, the low thermal conductivity of zirconia induces a substantial increase in temperature in the contact zone and weakens the material. Therefore, the incorporation of relatively low weight fraction of CNTs in zirconia matrix can reduce the wear rate and lower the friction coefficient. In the present paper, the friction coefficient and the wear rate of SPS zirconiaMWCNT composites were determined. The addition of MWCNT from 0.5 to 2 wt % resulted in reduction of 3Y-TZP grain size from 174 to 148 nm respectively. The effect of the addition of MWCNT on the coefficient of friction (COF) was studied. Nanoscratch and macro-scratch tests were conducted using diamond Berkovich and Rockwell indenter, respectively. Furthermore, the wear rate was also investigated using reciprocating sliding under a load of 5 N. It was found that the COF decreased with the increase in MWCNT content. However, in macro-scratch testing, there was a critical load over which brittle fracture sets in and its value decreases as the MWCNT content increases. The wear resistance was found to be decreasing very slightly for MWCNT content less than 1 wt %. However, wear resistance increases strongly for the addition of 2 wt % MWCNT. Author contributions: The powder processing of the composites and all the experiments were performed by the author, except for the nanoindentation test which

61

Summary of Appended Papers was performed by co-author Joan Josep Rovira. Data evaluation and discussion were drawn by the author together with Prof. Marc Anglada. Writing of this article was accomplished by the author.

Paper II Nanoindentation and fracture toughness of nanostructured zirconia/multi-walled carbon nanotube composites. Summary: The fracture toughness ( ) in 3Y-TZP can be increased by promoting phase transformation from tetragonal (t) to monoclinic (m) phase in front of a propagating crack tip referred to transformation toughening. However, the stronger the tendency for stress induced transformation, the higher the risk for premature spontaneous t-m transformation in humid atmosphere. This is called hydrothermal degradation or low temperature degradation (LTD) and can result in microcracking and loss of strength. This phenomenon is the main drawback for the wider use of 3Y-TZP. The resistance to LTD can be increased by reducing the grain size into the nanoscale by using Spark Plasma Sintering (SPS). However, the reduction of grain size will reduce the transformation toughening and the fracture toughness will decrease. In the current study MWCNTs have been used as a toughening phase due to their excellent mechanical properties. On the other hand, the "true" of the composites is still under debate since the indentation method is not an accurate method as it overestimates the values of . In the current study, a novel and accurate method for the calculation of "true" was developed. The method is based on producing a very sharp notch with submicrometer radius less than 0.5 μm using Ultra-short Pulsed Laser Ablation (UPLA). It was found that "true" of the composites with CNT content ranging from 0.5 up to 2 wt % CNT is hardly increasing with the addition of CNTs content while the indentation fracture toughness is increasing. Hence, the resistance to indentation cracking of the composites by adding CNTs to 3Y-TZP matrix does not indicate higher true fracture toughness. The reasons of the low of the composites could be related to the drop in hardness and elastic modulus found using Berkovich nanoindentation. Author contributions: The preparation of the composite powders was accomplished by the author. Mechanical testing and damage analysis were accomplished by the author. The nanoindentation test was performed by co-author Joan Josep Rovira. The discussion and the analysis of the results were drawn together by the author and Prof. Marc Anglada. 62

Summary of Appended Papers

Paper III The influence of unshielded small cracks in the fracture toughness of yttria and of ceria stabilised zirconia used in medical applications. Summary: Based on fracture mechanics criteria, if the critical natural flaw and the fracture toughness are precisely determined, then it is possible to predict the strength of advanced ceramics. In the present work, the novel method Ultra-short Pulsed Laser Ablation (UPLA) was used to investigate the influence of transformation toughening on the strength and on the fracture toughness of both 12Ce-ZrO2 and 3Y-TZP with two different grain sizes; 300 nm and 177 nm. A sharp notch induced using UPLA in 300 nm-3Y-TZP, 177 nm-3Y-TZP and 12Ce-ZrO2 results in similar and small cracks in both materials. Three regions can be distinguished from the fracture surface of the samples; the region of the notch (zone A), the micro-crack region (zone B) and the final fracture (zone C). The notch plus micro-cracked region act as unshielded sharp crack and the effective fracture toughness was found to be similar for all studied materials in spite of the large differences in plateau and steepness of their R-curves. The strength of 330 nm-3Y-ZrO2 is much higher compared to the strength of 12CeZrO2 which has higher grain size and higher transformation induced fracture in specimens with natural cracks. However, by inducing a very sharp notch using UPLA, the initial crack size from which fracture takes place is practically the same in both materials and the difference in strength is only about 10 %. Therefore, the beneficial effect of higher indentation fracture toughness in 12Ce-ZrO2 in comparison with 3YZrO2, widely referred in the literature, has a very small effect on the real effective fracture toughness that determines the strength of unshielded small cracks. Author contributions: The preparation of the specimens was accomplished by the author and co-author Miquel Turron-Vinas. The processing of the laser notch using UPLA method was manufactured by Dr. Pablo Moreno Pedraz at the University of Salamanca. Mechanical testing and all characterization techniques were carried out by the author. The discussion and the writing were accomplished by the author and Prof. Marc Anglada.

63

Summary of Appended Papers

Paper IV Material removal mechanisms by EDM of zirconia reinforced MWCNT nanocomposites. Summary: Zirconia ceramic has good mechanical properties and therefore a wide range of applications. However, its high hardness and brittleness limits its ability to be machined with conventional techniques. Electrical Discharge Machining (EDM) is one of the most promising techniques to machine ceramic matrix composites. Nevertheless, a crucial requirement in the use of EDM is the need for an electrically conductive work piece. In the current paper, the effect of MWCNT on the electrical conductivity and the machinability of the composites were studied. Moreover, the thermal conductivity of the composites at two different SPS temperatures; 1350 °C and 1500 °C, was investigated. The damage produced and the material removal mechanisms were investigated after EDM machining. It was found that the studied zirconia-MWCNT composites have high electrical conductivity which results in successful EDM machining of the composites with 1 and 2 wt % CNT content. However, there was only a relatively small increase of thermal conductivity with the addition of MWCNTs. A recast layer was found on the surface after machining where zirconium carbide (ZrC) was detected. The material removal mechanisms were found to be melting/evaporation and spalling. Author contributions: The composites powder processing was done by the author. Damage characterization after EDM was done by the author. Electrical and thermal conductivities were measured by the author with the help of co-authors Andreas Kaiser and Nikolaos Bonanos at Technical University of Denmark (DTU). EDM was performed by the company Zentrum für Mechatronik und Automatisierungstechnik (ZeMA) in Germany. The writing of the paper was accomplished by the author.

Paper V Surface microstructural changes of Spark Plasma Sintered zirconia after grinding and annealing Summary: Machining zirconia ceramics is considered as a crucial step in the manufacturing of long lasting and strong zirconia components. The damage induced by machining zirconia ceramics could affects it´s integrity and reliability. In the present study, an investigation of the effect of grinding was conducted on 3YTZP sintered using SPS at different temperatures; 1350 °C, 1450 °C and 1600 °C. A 64

Summary of Appended Papers thermal etching at 1100 °C for one hour of the ground samples revealed the presence of a nano-grain layer of about 60 μm grain sizes independent of the SPS temperature. The thickness of the nano-grain layer was in the order of few hundred nanometers. This nanograin layer is formed by recrystallization of a very thin highly deformed surface layer produced during grinding. Moreover, a heat treatment in air at 1200 °C of the ground samples showed that the surface grain size increases fast but it still remains smaller than in the starting polished specimens. Finally, when the ground layer is exposed to 1575 C annealing temperature, the surface grains grow to a size which is roughly similar to that achieved in polished specimens by heat treatment to the same temperature. The formation of the nano-grain layer could increase the resistance to LTD of zirconia ceramics. Furthermore, it may be of interest when a rough surface is beneficial as for example for implants since roughness favours osseointegration. Author contributions: The powder processing of the composites was done by the author. The SPS of samples was conducted by co-author Farid Akhtar. The experiments were performed by the author. The SEM image analysis was carried out together by the author and co-author Johanne Mouzon. The writing of the paper was completed by the author. All the co-authors participated in the improvements of the paper.

65

66

Part III

67

68

Paper I Coefficient of friction and wear resistance of zirconia–MWCNTs composites

Authors: Latifa Melk, Joan Josep Roa, Marta-Lena Antti and Marc Anglada

Paper Published in: Journal of Ceramics International, 41 (2014) 459-468.

69

70

Available online at www.sciencedirect.com

CERAMICS INTERNATIONAL

Ceramics International 41 (2015) 459–468 www.elsevier.com/locate/ceramint

Coefficient of friction and wear resistance of zirconia–MWCNTs composites Latifa Melka,b,n,1, Joan Josep Roa Roviraa, Marta-Lena Anttib, Marc Angladaa a

Department of Materials Science and Engineering, Universitat Politècnica de Catalunya, Barcelona 08028, Spain b Department of Engineering Sciences and Mathematics, Luleå University of Technology, Luleå, Sweden Received 13 August 2014; received in revised form 21 August 2014; accepted 23 August 2014 Available online 4 September 2014

Abstract Composites of 3 mol% yttria-doped tetragonal zirconia (3Y-TZP) reinforced with multiwalled carbon nanotubes (MWCNT) up to 2 wt% content have been produced using spark plasma sintering (SPS). The theoretical densities of the studied composites were found to be between 99.4% and 97.4%. The addition of MWCNT content resulted in reduction of 3Y-TZP grain size from 174 to 148 nm. The effect of MWCNT on the friction coefficient (COF) was studied by performing nano- and macro-scratches using diamond Berkovich and Rockwell indenters, respectively. Moreover, the COF and the wear rate were also investigated in reciprocating sliding against a zirconia ball under a load of 5 N. The results showed that the COF decreased upon the increase in MWCNT content. However, in macro-scratch testing, there was a critical load over which brittle fracture sets in and its value decreases as the MWCNT content increases. The wear resistance was found to be decreasing very slightly for less than 1 wt% MWCNT, while it increases strongly for the addition of 2 wt% MWCNT under the conditions studied. The results were discussed in terms of material properties, scanning electron microscopy observations of the wear track and nanoindentation tests. & 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.

Keywords: B. Nanocomposites; C. Wear resistance; D. Zirconia; Carbon nanotubes; Coefficient of friction

1. Introduction Tetragonal zirconia doped with 3 mol% yttria (3Y-TZP) has an excellent combination of high hardness, strength and fracture toughness. Regarding the wear behaviour, there is experimental evidence that the tribological couple 3Y-TZP/ 3Y-TZP has a high wear rate in comparison to other ceramic pairs like alumina/alumina [1,2]. The reason for this has been associated to surface fracture induced by microcracking during tetragonal (t) to monoclinic (m) phase transformation [3] and to the low thermal conductivity of zirconia which induces a substantial increase in temperature in the contact zone and weakens the material [4]. The addition of carbon nanotubes (CNT) or carbon nanofibers (CNF) to 3Y-TZP matrix using spark plasma sintering n Corresponding author at: Department of Materials Science and Engineering/ ETSEIB, Diagonal 647. 08028-Barcelona, Spain. Tel.: þ 34 934016701; fax: þ 34 934016706. E-mail addresses: [email protected], [email protected] (L. Melk). 1 Also at: Department of Engineering Science and Mathematics, Division of Material Science, E309, 97187 Luleå, Sweden. Tel.: þ 46 920493217.

http://dx.doi.org/10.1016/j.ceramint.2014.08.092 0272-8842/& 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.

(SPS) results in composites with significant reduction in grain size. Thus, t–m transformation is strongly reduced and, in principle, other mechanisms of toughening which are associated to the high strength and large aspect ratio of CNTs could be activated. At the same time, if the percolation is achieved, which occurs with relatively low weight percentages of CNTs (less than 2 wt%) [5], the higher thermal conductivity could also influence the wear behaviour. Therefore, it is interesting to assess the effect of CNTs on the tribological behaviour and to compare the wear resistance of monolithic 3Y-TZP and 3YTZP/CNT composites. The studies of the wear behaviour of zirconia–CNT composites are scarce and most of the work has been concentrated on alumina–CNT. According to Wang et al. [6] single wall carbon nanotubes (SWCNT)/alumina composites with 10 wt% SWCNT are highly contact-damage resistant, as shown by the lack of crack formation in indentation tests, but they also showed that they are as brittle as dense Al2O3, as revealed by the low fracture toughness determined by the single-edge V-notch beam method. Regarding the tribological properties, it has been reported an enhanced wear resistance in alumina–CNT composites for the

460

L. Melk et al. / Ceramics International 41 (2015) 459–468

addition of up to 2.25 vol% CNT content [7] or up to 4 wt% [8], while Ahmad et al. [9] reported a low friction coefficient (COF) and an improved wear resistance for 2, 5 and 10 wt% CNTs for low and moderate sliding loads. However, for high sliding loads (35 N), the nanocomposite with higher concentration of CNT exhibited lower wear resistance. On the other hand, it was reported in the literature that 3YTZP/CNT and 3Y-TZP/CNF composites have lower COF than 3Y-TZP matrix. Hvizdos et al. [10] studied the friction and wear behaviour of ZrO2 þ 1.07 wt% CNF composite and of monolithic ZrO2 using the ball-on-disk technique with an alumina ball as friction counterpart in dry condition at room temperature. They found that the COF of the composite was significantly smaller compared to monolithic zirconia, but the wear rate was slightly higher. The main wear mechanism in the nanocomposite was abrasion accompanied with pull-out of the carbon nanofibres which acted in the interface as a sort of lubricating media. Recently, Kasperski et al. [11] have reported that 3Y-TZP/CNT composites with a carbon content of 5.16 wt% have a COF of about 3.8 lower than that of monolithic ZrO2. In this study, the low COF was explained by the exfoliation of the MWCNTs due to the high shear stresses generated during sliding and to the formation of a lubricating film over the contact area. However, the wear rate was not studied in details. A similar study of the tribological behaviour of composites of CNT and CNFs in Si3N4, ZrO2 and Al2O3 matrices has shown a decrease in wear resistance with the exception of Si3N4–5 wt% CNT [12]. The purpose of the present paper is to study the tribological behaviour of 3Y-TZP/CNT composites produced by SPS. The study is carried out by measuring the friction and the wear in scratch tests under diamond tips and under reciprocating sliding using a ball of 3Y-TZP as counterpart. 2. Experiments 2.1. Starting materials and sample preparation Composites with four different MWCNT content (0, 0.5, 1 and 2 wt%) were investigated. The starting materials were high purity zirconia powder (TZ-3YSB-E, Tosoh, Japan) with 36 nm crystalline size and MWCNTs (Graphistrength C100, Arkema, France) with an outer diameter 10–15 nm, inner diameter 2–6 nm and length 0.1–10 μm. The details of the processing route used are described elsewhere [13]. The composites were sintered by spark plasma sintering (SPS FCT HP D25I, FCT System Gmβh) at maximum temperature of 1350 1C, dwell time of 5 min, and with heating and cooling rates of 100 1C/min. A pressure of 50 MPa was maintained during the sintering cycle. The final samples had the shape of discs with 3 mm thickness and 40 mm diameter and they were polished using series of diamond suspensions from 30 to 3 μm before a final polishing step with colloidal silica. Some of the polished samples were thermally etched at 1100 1C with a dwell time of 1 h in order to reveal their microstructure. The average grain size of each composite was determined using the line intercept method on scanning electron microscopy (SEM) images and the density was determined by the Archimedes method.

2.2. Characterisation techniques Nanoindentation tests were performed using a Nanoindenter XP (MTS) equipped with continuous stiffness measurements (harmonic displacement 2 nm and frequency of 45 Hz), using a pyramidal Berkovich diamond indenter. The strain rate was held constant at 0.05 s  1. The nanoindentation curves were analysed using the Oliver and Pharr method [14] in order to measure the nanoindentation hardness (HBerk) and the elastic modulus (EBerk). The indenter shape was carefully calibrated for true penetration depths as small as 60 nm by indenting fused silica samples of well known Young's modulus (72 GPa). The indents were organised in a regularly spaced array of 25 indentations (5 by 5) at a maximum penetration depth of 2000 nm or until reaching the maximum applied load, 650 mN. Each indentation was performed with a spacing distance of 50 μm in order to avoid any overlapping effect. Nano-scratch tests were carried out with a nano-scratch attachment of the Nanoindenter XP that allows lateral force measurements. A Berkovich indenter was employed to scratch the surface under increasing load at a velocity of 10 μm/s for a total scratch length of 500 μm up to a maximum load of 40 mN. Three different scratches were performed on each sample. Lateral (friction) forces were calculated from the deflection of the loading column. The COF was determined by taking the ratio of the lateral force measured by the equipment and the normal load applied on the material. One method used to assess the response to contact loading was by means of macro-scratch testing using a sliding Rockwell indenter with a diamond spherical tip radius of 200 μm (automatic Scratch Tester, CSM-Instruments, Switzerland). Both normal and tangential forces were recorded. Applied load ranged from 1 to 150 N over a sliding distance of 7.5 mm at a sliding speed 5 mm/min. Tribology tests were performed on an automatic tribometer (Wazau TRM1000, Germany) in reciprocating dry sliding conditions, using ball on disc geometry, at ambient temperature and pressure. A zirconia ball with 10 mm diameter was used as a counterpart. All the tests were carried out under a constant normal load of 5 N, a sliding velocity of 300 rpm and a stroke length of 4 mm. The total sliding distance was 100 m. At least four tests were performed for each condition and data represent their average. The COF was calculated by taking the ratio of the tangential and normal forces and it was reported versus the sliding distance. The volume removed was measured using a stylus profilometer where a hundred profiles along the width of the track were recorded. The wear rate, W, was determined in terms of the volume loss V per distance L and applied load F according to the following equation: W¼

V LF

ð1Þ

The surfaces of the wear tracks as well as sections perpendicular to the surface inside the tracks were analysed in detail using a dual beam Focused Ion Beam (FIB)/SEM Microscope (Zeiss Neon 40). For the observation of the cross sections a thin platinum layer was deposited on the sample prior to FIB machining in order

L. Melk et al. / Ceramics International 41 (2015) 459–468

to minimise ion-beam damage. A Ga þ ion source was used to mill the surface at a voltage of 30 kV. The final polishing of the cross-sections was performed at 10 pA and the surface was observed by SEM at regions below the wear track. Wavelength Dispersive X-ray spectrometry (JEOL JXA-8230) was used to compare the carbon content inside and outside the wear track at a voltage of 10 kV. Raman spectroscopy (alpha300RAþ from WITec GmbH) was used to analyse the effect of MWCNTs on the tribological properties of the composites where signals from both inside and outside the wear tracks were compared. Raman spectra were recorded using the 532 nm laser wavelength excitation and an acquisition range from 100 to 3000 cm  1.

3. Results Main properties of the composites together with code designation are given in Table 1. The increase in MWCNT content in the composites diminishes the final density from 99.4% for 3Y-TZP to 97.4% for 2 wt% MWCNT. Regarding the grain size, the average value is 177 nm in ZM0 (0 wt% MWCNT), which is nearly half that of 3Y-TZP sintered conventionally at 1450 1C ( 330 nm) [5]. In the composites, the average grain size of the zirconia matrix decreases slightly with MWCNT content down to 148 nm for ZM2 (2 wt% MWCNT).

461

The Vickers indentation response in terms of hardness and indentation fracture toughness (KIc) has been studied elsewhere [5]. Hardness diminishes and indentation KIc increases with MWCNT content making the material more tolerant to damage by contact loading as revealed by the higher indentation fracture toughness (Table 1). However, we have recently measured the true KIc on four point bending single edge cracked specimens with a very sharp notch induced by ablation with ultra-short laser pulses. The result shows that the true KIc in ZM0 is smaller than in conventionally sintered 3Y-TZP and it does not practically change with MWCNT content [5,16]. The COF measured in the nanoscratch test at very small loads and with a diamond Berkovich tip always decreases with increasing MWCNT content (see Fig. 1b). The results of the COF in macro-scratch testing with a sliding Rockwell indenter at increasing loads are shown in Fig. 1(a). In general, the friction coefficients for the different composites are not very different. Below about 40 N, the COF is clearly smaller for the composites with higher MWCNT content. While, beyond a critical load, about 120, 95 and 80 N for 0.5, 1 and 2 wt% MWCNT respectively, it shows strong oscillations except for ZM0. This transition starts earlier for the composites with larger MWCNT content. SEM pictures of the macro-scratch track are shown in Fig. 2 where the load is increasing from right to left. It can be seen that brittle fracture debris are hardly observed in ZM0 near the scratch

Table 1 Properties of the 3Y-TZP/CNT nanocomposites. Specimen code Specimen

Relative density Grain size (nm) H (GPa) (TD %)

HBerk (GPa) EBerk (GPa) KIc (MPa√m)a KIc (MPa√m)b

ZM0 ZM0.5 ZM1 ZM2

99.47 0.2 99.27 0.3 98.27 0.3 97.47 0.2

20.07 0.4 15.97 0.4 15.97 0.7 12.77 0.7

a

3Y-TZP 3Y-TZPþ 0.5 wt% CNT 3Y-TZPþ 1 wt% CNT 3Y-TZPþ 2 wt% CNT

177714 176710 161711 14879

14.2170.09 12.9870.08 11.3270.10 9.5270.05

263 75 224 74 215 77 181 76

3.5770.09 4.0270.05 4.5670.08 4.9770.06

2.770.1 2.770.1 2.870.1 2.870.1

Indentation method of Anstis et al. [15]. SENB specimen with notch machined with UPLA.

b

Fig. 1. COF of the composites during (a) macroscratch test under an incremental load (1–150 N) and (b) nanoscratch test under a maximum load of 40 mN.

462

L. Melk et al. / Ceramics International 41 (2015) 459–468

Fig. 2. Tracks of the Rockwell scratch test in the composites for various MWCNT contents: (a) ZM0, (b) ZM0.5, (c) ZM1 and (d) ZM2.

Fig. 3. (a) COF versus sliding distance and (b) average values of COF in different sliding intervals in terms of MWCNTs content.

tip, while for the composites ZM1 and ZM2 brittle fracture is present at the tip as well as along the track in the region above the critical load. The beginning of brittle fracture is located at distances from the scratch tip of about 2.75 mm (80 N, for ZM2) and of 3.5 mm (95 N, for ZM1) where oscillations in the COF start to appear as can be observed in Fig. 1. Fig. 3(a) shows that under reciprocating sliding conditions using ball on disc geometry and a zirconia ball as a counterpart, the COF increases with sliding distance. In the first 20 m sliding distance, the average value of the COF for all compositions is  0.45, with the exception of 3Y-TZP matrix for which the COF increases from 0.36 to 0.54. With further sliding, during the next 30 m, approximately, there is a strong raise in the COF for all compositions, but the increase is smaller for the composites with higher MWCNT content. After 50 m sliding distance, there are only slight changes in the COFs with further sliding. It is interesting to notice that after 100 m sliding distance, the smaller COF corresponds to ZM2 (the highest MWCNT content ) as it can be also appreciated in Fig. 3(b) where the average values of the COFs are plotted for the different sliding

distance intervals mentioned before. COFs are finally in the range between 0.85 (ZM0) and 0.57 (ZM2). Moreover, the COF of conventionally sintered 3Y-TZP (with  330 nm grain size) was found to be larger than the COF of ZM0 (spark plasma sintered with  177 nm grain size). With respect to the wear rate, the effect of CNTs is clearly visible in the measured track profiles at half track length (Fig. 4a). Once the wear rate is calculated and plotted (Fig. 4b) it is observed that for ZM0.5 and ZM1 the change in wear rate is relatively small with respect to ZM0, while a strong decrease takes place for ZM2. Fig. 5 shows the images of the tracks in reciprocating sliding after 100 m sliding distance. Many flat patches were observed and a clear decrease in the depth and the width of the track profile of ZM2. The appearance of the surface of the track in the SEM at higher magnification is shown in Fig. 6. The main difference observed is the presence of grain pull out and debris in ZM0 together with flat patches in the composites with higher CNT content.

L. Melk et al. / Ceramics International 41 (2015) 459–468

463

Fig. 4. (a) Wear track profiles measured at half track length and (b) wear rate of the studied nanocomposites.

Fig. 5. Wear track profiles after 100 m sliding (a) ZM0, (b) ZM0.5, (c) ZM1 and (d) ZM2.

One specimen of ZM2 was fractured perpendicularly to the track in order to observe the surface just below the track (Fig.7a). At higher magnification the track shows a similar appearance as the lower part of Fig. 6(c), showing the presence of small grains and debris (Fig. 7b). The surface just below the track has a slight different appearance close and away from the wear track (Fig. 7c and d). At long distances from the track, the fracture surface has the usual appearance as that of four point bending specimens of ZM2 with some agglomerates of

MWCNT which have not been affected by the sliding contact forces. However, close to the track, the appearance of the fracture surface shows that the composite is slightly more compacted. The centre of the track of ZM2 was milled by FIB in order to study the cross-section just below the wear track. In some places it was detected the presence of strong contrast by electron back scattering as revealed by the white contrast in the surroundings of the cut (see Fig. 8). After polishing, the zone

464

L. Melk et al. / Ceramics International 41 (2015) 459–468

Fig. 6. SEM images of the wear tracks for (a) ZM0, (b) ZM0.5, (c) ZM1 and (d) ZM2.

Fig. 7. High magnification of ZM2: (a) general view, (b) inside wear track, (c) fracture surface just below the wear track and (d) fracture surface far away from track.

of white contrast of the cross section was analysed by both imaging modes and by energy dispersive spectroscopy (EDS). In comparison with the surroundings, no grain structure is observed inside this region and the EDS chemical analysis gives much higher (nearly twice) Zr content than the average composition and O2 which indicates maybe the formation of an

amorphous phase of zirconia and which to the best of our knowledge has not been shown before for CNT/ZrO2 composites, but warrants further studies. A chemical analysis by wavelength dispersive spectroscopy (WDS) was attempted for the determination of carbon amount inside and outside the track of ZM2, because of its lower

L. Melk et al. / Ceramics International 41 (2015) 459–468

465

Fig. 8. (a) FIB cross section in the middle of the wear track of ZM2 and (b) selected zone below the wear track (c) a high magnification of (b).

Fig. 9. Result of WDS analysis of carbon inside and outside the wear track. The carbon content at any point inside the track is higher than outside. (a) Points 1 and 2 where the analysis was carried out and the result is shown in (b).

Fig. 10. (a) Nano-hardness and (b) elastic modulus inside and outside the wear track.

friction coefficient and higher wear resistance. In any place inside the wear track, the carbon content detected was significantly higher than outside the track. This is shown in Fig. 9 where the analysis was carried out in the zones indicated in Fig. 9(a) and the results are shown in Fig. 9(b). The surface inside the track was not flat enough in order to obtain absolute values of the carbon content. Nanoindentation tests were carried out inside and outside the wear track in order to detect differences in the nanohardness and the elastic modulus. Inside the wear track, the surface is not perfectly flat so that large scatter is expected in the results.

Fig. 10 shows the nanohardness and elastic modulus. Independently of the composition, these properties reach higher values outside the wear track than inside. Fig. 11 shows the variation of Raman spectra inside (worn surface) and outside (unworn surface) of the wear track in ZM2. Raman spectra show the typical D (  1350 cm  1), G ( 1585 cm  1) and G´ (  2700 cm  1) characteristic bands of CNTs. The G band is associated with stretching vibrations of C–C bond in graphene layers and indicates the formation of well-graphitized carbon nanotubes [17]. The D-band is originated from atomic displacement and disorder induced features

466

L. Melk et al. / Ceramics International 41 (2015) 459–468

caused by lattice defect [18], and the G´ or (2D) band is an overtone mode of the D-band [19]. The intensity of the G and D bands are a commonly accepted indicator of the sample crystallinity [20] and the broadening of the bands is related to the amorphous character [21]. As one can see in Fig. 12(a), the ratio between intensities of D and G bands, ID/IG, is larger inside the wear track (1.17), where there is more damage, than outside (0.49). Besides, inside the wear track, there is a negative shift in the peaks of  36 cm  1 (D band),  3 cm  1 (G band), and  7 cm  1 (G´ band) indicating an increase of defects in the nanotubes. Fig. 12(b) shows a Lorentzian fitting analysis of the features in the centre of the wear track. It is observed that the peaks at 1395 cm  1 and 1595 cm  1 are the D and the G bands respectively. However, a shoulder identified at 1544 cm  1 suggests the loss of the nanotube structure. Fig. 13 shows Raman analysis of some regions in the centre of ZM2 wear track where a broad spectra called “amorphous” similar to that of carbon thin film [22]. The spectrum consists originally of two peaks corresponding to D and G bands. D band is centred at around 1360 cm  1 and G band at around

Fig. 11. Raman spectra taken from inside (worn surface) and outside (unworn surface) of the wear track of ZM2.

1570 cm  1 which confirms the presence of an amorphous carbon film as it was reported in [23]. 4. Discussion Regarding the COF of the pair zirconia–zirconia, the main results can be summarised as follows. Under the load studied here, the COF for ZM0 is lower than for conventionally 3Y-TZP. Similarly, the addition of MWCNT diminishes the COF of ZM0, which is in the range 0.52–0.73 as compared to values reported in the literature of 0.45–0.6 [11] for zirconia–CNT composites and  0.4–0.5 for 3Y-TZP–1.07% CNF composites [10]. In both cases, an alumina ball was used as counterpart. The analysis of the different sliding distances shows that the COF of ZM1 after 25 m sliding distance is still far away from reaching a saturation value. Then it looks as there is a progressive deterioration of the material that increases the COF with sliding distance, except for the compositions ZM0 that correspond to the 3Y-TZP matrix and ZM2 where the MWCNT content has reached a percolation

Fig. 13. Raman spectra of the carbon film taken from the wear track showing D peak is centred at 1360 cm  1 and G peak is centred at 1570 cm  1. The solid lines are the Lorentzian fit to the experimental data.

Fig. 12. (a) Raman spectra taken from the centre and outside the wear track of ZM2 and (b) Lorentzian fitting of the MWNTs D and G band related peaks in the centre of the wear track.

L. Melk et al. / Ceramics International 41 (2015) 459–468

value. This could be related to a decrease in abrasion by an increase in thermal conductivity with percolation and/or the formation of some tribo-film during sliding which could be related to the flat patches observed on the wear track. The flat patches are still distributed on a relatively small fraction of the area of the wear track, but it´s frequency increases with MWCNT content. However attempts to measure the amount of carbon by WDS on the flat patches of the wear track were unsuccessful taking into account the probably small thickness of this layer. Observations of flat patches on the wear track have been seen before by Hvizdoš et al. [10] and Kasperski et al. [11], and they have been associated to the existence of a smeared carbon based-transferred film along the wear track due to the exfoliation of MWCNTs which results in easy shear and lower friction coefficient in the composites. The minimum shear stress at the sample surface during sliding can be calculated for 5 N load as it was reported in [11] reaching values higher than 100 MPa which is very high compared to the measured shear strength of 0.04 MPa for MWNTs [24]. Broken (exfoliated) MCWNTs has also been reported in CNT/copper composites and explained to contribute to the formation of a mechanically mixed carbonaceous layer which lower the friction coefficient [25]. However, full evidence has not been given. In the present case additional evidence is given for the existence of this tribo-film by raman spectroscopy analysis, which could explain the low wear and friction coefficient found for ZM2. The hardness and elastic modulus for ZM2 are lower than in ZM0 because of the existence of some clusters of MWCNT in ZM2 and close to the surface as can be evidenced in Figs. 7 and 8. However, in both compositions these parameters inside the wear track are smaller than outside. This may appear contradictory as it seems to be some compaction below the wear track (see Fig. 7). Thus, one expects the hardness to be higher. This effect appears independently of the MWCNT content and could be associated to the existence of a very thin layer of wear debris and grains not fully attached to the wear track and that have therefore higher compliance than the internal parts below the wear track. It can be seen that, in spite of the large scatter in the results, the difference in elastic modulus inside and outside the wear track is smaller since now the elastic modulus is not as much sensitive to the properties of this very thin surface layer as it´s case of the hardness. On the other hand, the appearance of microcracking below the track gives only an important contribution to the increase in compliance as compared to outside of the wear track. In the macroscopic scratch test, the load at which the COF increases with MWCNT content coincides with the appearance of brittle fracture on the scratch track. At small loads, the lower COF of the composite with the highest MWCNT content may be understood by the lubricating effect of MWCNTs trapped just below the tip of the indenter. It is interesting to note, that the composite in which brittle fracture appeared at smaller loads in the macroscopic scratch test (ZM2) is the one with the highest resistance to cracking under Vickers static indentation [5]. This could be associated to the high tensile stress perpendicular to the surface generated

467

behind the scratch test differently of the static Vickers indentation where the tensile stress near the surface at the contact are small. The tensile stress easily propagates the lateral crack system and probably favoured by the presence of more MWCNT agglomerates in the highest composition. The decrease in wear rate for ZM2 under the conditions studied here has not been reported before in 3Y-TZP-CNT composites. The reason for this behaviour cannot be associated to an increase in mechanical properties of the composites since the true KIc does not suffer significant changes with addition of MWCNTs [5]. If the expression developed by Evans for W is considered [16]  4=5 F 1=8 E W ¼ a 1=2 ð2Þ 5=8 H K Ic H where F is the applied load, KIc the fracture toughness, H the hardness, a is constant independent of material type. As E/H and KIc practically do not change and H decreases with increasing MWCNTs, one should expect a higher W in the composites, which is opposite to the present findings for ZM2. Therefore, the reason for this behaviour should lie in the change of other parameters that may affect wear. One of the parameters is the higher thermal conductivity of the material for ZM2 since the percolation value is reached and higher amount of heat transfer is possible. Moreover, the tendency to phase transformation in the wear track of ZM2 diminishes due to the small grains size. However, in the present case, the main difference between all the compositions is the prominent appearance of flat patches in the wear track of ZM2. The analysis by WDS of the flat patches of the track shows that the C content in some few monolayers would be compatible with the amount measured. Hvizdos et al. [10] also detected a small increase in the wear rate by using an alumina ball as counterpart with the addition of 1 wt% CNF to zirconia. However, the absolute values of wear rate were lower than in the present investigation. 5. Conclusion From the study of the COF and the wear of spark plasma sintered 3Y-TZP reinforced with up to 2 wt% MWCNTs, it is concluded that Under relatively low loads, the addition of MWCNTs increases the COF and clearly decreases the wear rate when the amount of MWCNT reaches about the percolation value (2 wt% CNT). While, under high loads, the composites develop chipping and brittle fracture as MWCNT content increases. The low COF and low wear rate at small load is related to the formation of a tribofilm during reciprocating sliding. The onset of chipping at high loads is probably related to the weak interface between zirconia and MWCNTs and to the tensile stresses developed during the scratch test. Acknowledgements The authors gratefully acknowledge Prof. Michael Reece of Nanoforce Ltd. for the use of SPS facilities and the financial support given by the “Ministerio de Ciencia e Innovación” of

468

L. Melk et al. / Ceramics International 41 (2015) 459–468

Spain through research grant MAT2011-23913. L.M. acknowledges the fellowship award received from the European Joint Doctoral Programme in Materials Science and Engineering (DocMASE) of the European Union. All authors thank Dr. Trifon Trifonov and Dr. Emilio Jiménez for their assistance in the FIB/SEM equipment and nanoindenter, respectively. References [1] C. Piconi, G. Maccauro, Zirconia as a ceramic biomaterial, Biomaterials 20 (1999) 1–25. [2] W.M. Rainforth, The wear behaviour of oxide ceramics – a review, J. Mater. Sci. 39 (2004) 6705–6721. [3] I. Birkby, P. Harrison, R. Stevens, The effect of surface transformation on the wear behaviour of zirconia TZP ceramics, J. Eur. Ceram. Soc. 5 (1989) 37–45. [4] W.M. Rainforth, R. Stevens, Transmission electron microscopy of worn zirconia surfaces, J. Mater. Res. 13 (2011) 396–405. [5] L. Melk, J.J. Roa Rovira, F. García-Marro, M.-L. Antti, B. Milsom, M.J. Reece, Nanoindentation and fracture toughness of nanostructured zirconia/multi-walled carbon nanotube composites (2014) (submitted for publication). [6] X. Wang, N.P. Padture, H. Tanaka, Contact-damage-resistant ceramic/ single-wall carbon nanotubes and ceramic/graphite composites, Nat. Mater. 3 (2004) 539–544. [7] S.W. Kim, W.S. Chung, K.-S. Sohn, C.-Y. Son, S. Lee, Improvement of wear resistance in alumina matrix composites reinforced with carbon nanotubes, Metall. Mater. Trans. A. 41 (2009) 380–388. [8] J.-W. An, D.-H. You, D.-S. Lim, Tribological properties of hot-pressed alumina – CNT composites, Wear 255 (2003) 677–681. [9] I. Ahmad, A. Kennedy, Y.Q. Zhu, Wear resistant properties of multiwalled carbon nanotubes reinforced Al2O3 nanocomposites, Wear 269 (2010) 71–78. [10] P. Hvizdoš, V. Puchý, A. Duszová, J. Dusza, Tribological behavior of carbon nanofiber–zirconia composite, Scr. Mater. 63 (2010) 254–257. [11] A. Kasperski, A. Weibel, D. Alkattan, C. Estournès, V. Turq, C. Laurent, et al., Microhardness and friction coefficient of multi-walled carbon nanotube-yttria-stabilized ZrO2 composites prepared by spark plasma sintering, Scr. Mater. 69 (2013) 338–341.

[12] P. Hvizdoš, V. Puchý, A. Duszová, J. Dusza, C. Balázsi, Tribological and electrical properties of ceramic matrix composites with carbon nanotubes, Ceram. Int. 38 (2012) 5669–5676. [13] B. Milsom, G. Viola, Z. Gao, F. Inam, T. Peijs, M.J. Reece, The effect of carbon nanotubes on the sintering behaviour of zirconia, J. Eur. Ceram. Soc. 32 (2012) 4149–4156. [14] W.C. Oliver, G.M. Pharr, Measurement of hardness and elastic modulus by instrumented indentation: Advances in understanding and refinements to methodology, J. Mater. Res. 19 (2011) 3–20. [15] G.R. Anstis, P. Chantikul, B.R. Lawn, D.B. Marshall, A critical evaluation of indentation techniques for measuring fracture toughness: I, direct crack measurements, Am. Ceram. Soc. 64 (1981) 533–538. [16] A.G. Evans, D.B. Marshall, Wear mechanisms in ceramics, in: Fundamentals of Friction and Wear of Materials. ASM Int., 1981. [17] M.S. Dresselhaus, G. Dresselhaus, R. Saito, A. Jorio, Raman spectroscopy of carbon nanotubes, Phys. Rep. 409 (2005) 47–99. [18] H.C. Choi, J. Park, B. Kim, Distribution and structure of N atoms in multiwalled carbon nanotubes using variable-energy X-ray photoelectron spectroscopy, J. Phys. Chem. B 109 (2005) 4333–4340. [19] L.G. Bulusheva, A.V. Okotrub, I.A. Kinloch, I.P. Asanov, A.G. Kurenya, A.G. Kudashov, et al., Effect of nitrogen doping on Raman spectra of multi-walled carbon nanotubes, Phys. Status Solidi 245 (2008) 1971–1974. [20] J. Schwan, S. Ulrich, V. Batori, H. Ehrhardt, S.R.P. Silva, Raman spectroscopy on amorphous carbon films, J. Appl. Phys. 80 (1996) 440. [21] B. Rebollo-Plata, R. Lozada-Morales, R. Palomino-Merino, J.A. DávilaPintLe, O. Portillo-Moreno, O. Zelaya-Angel, S. Jiménez-Sandoval, Amorphous carbon thin films prepared by electron-gun evaporation, AZo J. Mater. Online (2005). [22] M. Iwaki, K. Takahashi, A. Sekiguchi, Wear and Raman spectroscopy of ion-bombarded glass-like carbon, Diam. Relat. Mater. 3 (1994) 47–51. [23] P.K. Chu, L. Li, Characterisation of amorphous and nanocrystalline carbon films, Mater. Chem. Phys. 96 (2006) 253–277. [24] A. Kis, K. Jensen, S. Aloni, W. Mickelson, A. Zettl, Interlayer forces and ultralow sliding friction in multiwalled carbon nanotubes, Phys. Rev. Lett. 97 (2006) 025501. [25] C. Guiderdoni, E. Pavlenko, V. Turq, A. Weibel, P. Puech, C. Estournès, et al., The preparation of carbon nanotube (CNT)/copper composites and the effect of the number of CNT walls on their hardness, friction and wear properties, Carbon N.Y. 58 (2013) 185–197.

Paper II Nanoindentation and fracture toughness of nanostructured zirconia/multi-walled carbon nanotube composites

Authors: Latifa Melk, Joan Josep Roa, Fernando García-Marro, Marta-Lena Antti, Ben Milsom, Mike J. Reece and Marc Anglada.

Paper Published in: Journal of Ceramics International, 41 (2015) 2453-2461.

81

82

Available online at www.sciencedirect.com

CERAMICS INTERNATIONAL

Ceramics International 41 (2015) 2453–2461 www.elsevier.com/locate/ceramint

Nanoindentation and fracture toughness of nanostructured zirconia/multi-walled carbon nanotube composites Latifa Melka,b,n, Joan J. Roa Roviraa, Fernando García-Marroa, Marta-Lena Anttib, Ben Milsomc,d, Michael J. Reecec,d, Marc Angladaa a

Department of Materials Science and Engineering, Universitat Politècnica de Catalunya, Barcelona 08028, Spain b Department of Engineering Sciences and Mathematics, Luleå University of Technology, Luleå, Sweden c Department of Materials, Queen Mary College, University of London, London E1 4NS, UK d Nanoforce Technology Limited and School of Engineering and Materials Science, Queen Mary, University of London, London E1 4NS, UK Received 21 September 2014; received in revised form 9 October 2014; accepted 10 October 2014 Available online 19 October 2014

Abstract Multi-walled carbon nanotubes (MWCNTs)/3 mol% yttria-doped tetragonal zirconia (3Y-TZP) composites were produced using spark plasma sintering (SPS) with MWCNT content ranging within 0–2 wt%. In the present paper, it was shown that the addition of MWCNTs results in a refinement of the composites microstructure. Moreover, nanoindentation tests were performed in order to monitor the change in elastic modulus and hardness with MWCNT content and it was found that both properties decrease with the addition of MWCNT content. A novel method was used to measure the true fracture toughness of the composites by producing a shallow surface sharp notch machined by ultra-short pulsed laser ablation on the surface of beam specimens. The true fracture toughness obtained on this laser machined single edge V-notch beam (SEVNB) specimens tested in four point bending was compared to the indentation fracture toughness measured using a Vickers indenter. It was found that the indentation fracture toughness increases with increasing MWCNT content, while the true fracture toughness determined with SEVNB was practically independent of the composition. Finally, it was concluded that the increase in the resistance to indentation cracking of the composites with respect to 3Y-TZP matrix cannot be associated to higher true fracture toughness. The results were discussed in terms of transformation toughening, damage induced in front of the notch tip, microstructure of the composites, and fracture toughness of 3Y-TZP. & 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.

Keywords: C. Mechanical properties; Ceramic–matrix composites (CMCs); Carbon nanotubes; Fracture toughness; Short pulse laser machining

1. Introduction In the last decades there has been growing interest in developing ceramic materials with high fracture toughness (KIc) and strength for structural applications. In the specific case of 3 mol% yttriadoped tetragonal zirconia (3Y-TZP), KIc can be increased by promoting phase transformation from tetragonal (t) to monoclinic (m) phase in front of a propagating crack tip (so called n Corresponding author at: Department of Materials Science and Engineering/ ETSEIB, Diagonal 647, 08028-Barcelona, Spain. Tel.: þ34 934016701; fax: þ34 934016706. E-mail address: [email protected] (L. Melk).

http://dx.doi.org/10.1016/j.ceramint.2014.10.060 0272-8842/& 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.

transformation toughening) [1]. However, as stronger is the tendency for stress induced transformation, as higher is the risk for premature spontaneous t–m transformation on the external surface in contact with moisture in the environment. This effect is referred to as hydrothermal degradation or low temperature degradation (LTD) and it is accompanied by surface microcracking. This phenomenon is the main drawback for the wider use of 3Y-TZP. It is more severe at higher temperatures, for larger grain sizes, and under the presence of tensile residual stress or low concentrations of yttria [1]. The resistance to LTD can be strongly increased by reducing the grain size into the nanoscale. However, this will reduce the transformation toughening capability, so that the fracture

2454

L. Melk et al. / Ceramics International 41 (2015) 2453–2461

toughness will diminish [2]. One strategy to counteract LTD is the incorporation of a second phase as a toughening mechanism in the form of particles, grains, whiskers or fibres into the nano-grain zirconia matrix. Recent investigations have been focused on the addition of carbon nanotubes (CNTs) as reinforcements into a ceramic matrix because of their high aspect ratio and outstanding mechanical properties [3]. Moreover, it was reported that the addition of CNT to TZP-based composites improves the fracture toughness without affecting the resistance to LTD [1]. On the other hand, there is still a debate about the efficiency of CNTs in the mechanical properties of the ceramic matrix nanocomposites. Zhan et al. [4] reported an important improvement in indentation KIc (KIc  9.7 MPa m1/2) with addition of 10 vol% SWNTs to Al2O3 matrix, but Wang et al. [5] found a much smaller improvement (only 3%) when KIc was measured by the single edge V-notch beam (SEVNB) method on the same nanocomposites. A key observation made on these alumina composites is that they can be more contactdamage resistant than alumina, as it was shown by the lack of crack formation during indentation tests, and, in the same time, they were brittle as dense Al2O3 with low KIc measured by the SEVNB method [5]. Therefore, it was concluded that for the alumina composites studied, a high tolerance to indentation loading does not necessarily represent an increase in KIc. Some studies about the incorporation of multi-walled carbon nanotubes (MWCNTs) to 3Y-TZP matrix reported an increase in indentation KIc using different techniques for the processing of 3Y-TZP/CNT composites [6–9]. Values up to 13 MPa m1/2 have been reported for the addition of 1 wt% SWNTs [10]. Other authors have found that the increase in indentation KIc only occurs for the addition of relatively small CNTs content: 0.5 wt% ([11,12]), 1.0 wt% [11]. However, an increase of KIc was also reported for the addition of 4 wt% of functionalized CNTs [12]. While, a decrease in indentation KIc was usually observed for concentrations higher than 2 wt% CNT [11–15] and explained by the poor dispersion of CNTs and the weak bonding between CNTs and zirconia matrix. Moreover, a decrease in KIc was reported in 1.07 wt% CNF/CNT [16,17] as well as by adding a small fraction of  0.69 wt% carbon nanofibers (CNF) [18]. Regarding the influence of CNT on the Vickers hardness to 3Y-TZP, most of the studies have concluded that with the addition of either small CNT content (  0.5%) ([6,7,9]) or above relatively high content, the hardness decreases with the addition of CNT or CNF to monolithic zirconia [8,10,11, 14–18]. One exception is the work of Mazaheri et al. [7], who reported a slight increase for the addition of 5 wt% MWCNT. The reason of the disparity of KIc results in the literature, mentioned in the previous part, is probably related to the use of indentation method. It was reported [19] that the indentation method can give unreliable results as shown before for SWNTAl2O3 composites [5]. Thus, alternative methods to indentation are necessary to be used to measure the true KIc. Single edge V-notch beam method has been used as a reliable method to measure KIc. Fracture toughness values of 8.1 MPa m1/2 for 5 wt% 3Y-TZP/CNT have been reported

[7,20]. A critical issue with this method is the sharpness of the notch tip. It is well known that notches induce more moderate stress fields than cracks and this may lead to an overestimation of KIc, However, since sharp microcracks are produced in front of the notch during machining, the notch tip radius should be smaller than the size of machining-induced defects in order that the SEVNB method can be applied for the determination of KIc [21]. As the length of these microcracks are proportional to the grain size, they can be long enough in coarse grain structures, but in fine grain ceramics this condition is difficult to fulfil even by sharpening the notch tip with a razor blade impregnated with diamond paste. This point was investigated experimentally in a European round robin where it was concluded that the radius of the starting notch in 3Y-TZP (grain size  0.4 mm) should be less than about 1 mm [22]. Therefore, in both 3Y-TZP and 3Y-TZP/CNT processed by spark plasma sintering (SPS) with grain size even much smaller than conventionally processed 3Y-TZP, the required notch tip radius is too small to be introduced by honing. Recently, the effect of notch tip radius on KIc has been studied by Fischer et al. [23] in 3Y-TZP of 400 nm grain size. They obtained large values of KIc ranging from 5.9 to 13.6 MPa m1/2 for notch root radii between 18 and 167 μm. The purpose of this paper is to study the influence of MWCNT content on the mechanical properties of 3Y-TZP and to calculate the true KIc of 3Y-TZP matrix and 3Y-TZP/CNT composites, both produced by SPS, by applying a novel method based on inducing a shallow surface sharp notch machined by ultra-short pulsed laser ablation (UPLA) and finally to compare the results with those obtained by Vickers indentation on the same materials.

2. Experimental procedure 2.1. Processing of nanocomposites Composites with four different amounts of MWCNTs (0, 0.5, 1 and 2 wt%) were prepared. The starting materials were high pure (99.9%) zirconia powder (TZ-3YSB-E, Tosoh, Japan) with crystalline size of 36 nm and MWCNTs (Graphistrength C100, Arkema, France) sintered by SPS (SPS FCT HP D25I, FCT System Gmβh, Germany). The details of the processing and preparation of samples are described elsewhere [24]. All samples were sintered at a temperature of 1350 1C with a dwell time of 5 min, with heating and cooling rates of 100 1C/min. A pressure of 50 MPa was maintained during the sintering cycle. The final samples were ceramic discs with 3 mm thickness and 40 mm diameter. The discs were polished using series of diamond suspensions from 30 to 3 μm then with colloidal silica. Finally the discs were cut into bar (4 mm  2.5 mm  40 mm) specimens for mechanical testing. The cut faces were ground and polished. Some of the polished samples were thermally etched at 1100 1C during 1 h in order to reveal the microstructure. The average grain size was determined using the line intercept method on scanning electron microscopy (SEM) images and the density was determined by the Archimedes method.

L. Melk et al. / Ceramics International 41 (2015) 2453–2461

2.2. Mechanical properties The response to sharp contact loading was examined by indentation fracture testing method. The specimen’s surface was indented by a Vickers indenter under 1 kg (9.81 N) load and the Meyer hardness (H) and the indentation KIc were determined. Taking into account the Palmqvist morphology of the indentation cracks, the expression proposed by Niihara et al. [25] for Palmqvist cracks is the most appropriate for determining the indentation KIc. However, since in the literature the equation of Anstis et al. [26] for indentation KIc of median radial cracks is often employed independently of the shape of the cracks, it will be used in this study in order to compare the present results with other published results. As the grain size of the composites tested in this study is in the range  148–174 nm, a very sharp notch is needed in order to ensure that microcracks left in front of the notch tip are larger than the notch tip radius [21,27]. A novel method has been used for measuring KIc based on introducing a sharp shallow surface notch by UPLA on the surface of prismatic bars which was parallel to the original discs and determining KIc from the strength in four point bending. The femtolaser system used to produce the notch was a commercial Ti: Sapphire oscillator (Tsunami, Spectra Physics) and a regenerative amplifier system (Spitfire, Spectra Physics) based on chirped pulsed amplification. The femtolaser delivered 120 fs linearly polarized pulses at 795 nm with a repetition rate of 1 kHz. The pulse energy used was 5 mJ and the focusing system was an achromatic doublet lens with 50 mm focal length. The samples were placed on a XYZ motorized stage and moved along one of the horizontal axis with a scanning speed of 50 mm/s. Four passes were needed to achieve the desired notch depth (  24 mm). The stress intensity factor, KI, was obtained by using the following expressions proposed by Munz and Fett [28]: 3FðS1  S2 Þ pffiffiffi Y a 2BW 2  pffiffiffi  1:1215 π 5 5 1 2 6 3 6:1342α  α þ α β þ expð Þ Y¼ 8 12 8 8 β β3=2 KI ¼

ð1Þ ð2Þ

where S1 and S2 are the outer and inner spans, respectively, B is the thickness, W the width, F the applied load, a is the crack length, which is taken as the depth of the notch plus the small damage region in front of the notch, α ¼ a/W and β ¼ 1 α. The tests were performed for 0, 0.5 and 2 wt% CNT. Additional

2455

tests were carried out in standard 3Y-TZP with larger grain size ( 330 nm). Finally, the notched bar specimens (4 mm  2.5 mm  40 mm) were tested in a four point bending test device (DEBEN, Microtest, UK) in air with spans of 30/12 mm. The average stress rate was 2.4 MPa/s. Three specimens were used for each composition. The method used in this study has recently been applied to conventionally sintered 3Y-TZP (330 nm grain size) [29] and the results for KIc were found to be similar to values determined by other methods using sharp starting cracks. Nanoindentation tests were performed using a Nanoindenter XP from MTS equipped with continuous stiffness measurements (harmonic displacement 2 nm and frequency of 45 Hz), using a pyramidal Berkovich diamond indenter. The strain rate was held constant at 0.05 s  1. The nanoindentation curves were analysed using the Oliver and Pharr method [30] in order to measure the nanoindentation hardness (HBerk) and the elastic modulus (EBerk) as a function of the penetration depth for each composition where HBerk was defined as the ratio of load and contact area at maximum load. The indenter shape was carefully calibrated for true penetration depths as small as 50 nm by indenting fused silica samples of well-known Young’s modulus (72 GPa). The indents were organized in a regularly spaced array of 25 indentations (5 by 5) at a maximum penetration depth of 2000 nm or until reaching the maximum applied load, 650 mN. Each indentation was performed with a spacing distance of 50 μm in order to avoid any overlapping effect.

2.3. Damage The surface damage associated with residual nanoindentation imprints was visualized by atomic force microscopy (AFM, Dimension D3100 from Bruker), in tapping mode. All the images were processed with the WSxM software [31]. The subsurface damage induced during the nanoindentation tests as well as in front of the notch tip by UPLA were characterized using a dual beam Focused Ion Beam (FIB)/SEM Microscope (Zeiss Neon 40) and the fracture surfaces were characterized by field emission scanning electron microscope (SEM) (JSM-7001F, JEOL, Japan). A thin platinum layer was deposited on the sample prior to FIB machining in order to minimize ion-beam damage. A Ga þ ion source was used to mill the surface at a voltage of 30 kV. The final polishing of the cross-sections was performed at 10 pA.

Table 1 Properties of the 3Y-TZP/CNT nanocomposites. Specimen

Relative density (TD %)

Grain size (nm)

H (GPa)

HBerk (GPa)

EBerk (GPa)

KIca(MPa m1/2)

KIcb (MPa m1/2)

3Y-TZP 3Y-TZPþ0.5 wt% CNT 3Y-TZPþ1 wt% CNT 3Y-TZPþ2 wt% CNT

99.470.2 99.270.3 98.270.3 97.470.2

1777 14 1767 10 1617 11 1487 9

14.2170.09 12.9870.08 11.3270.10 9.5270.05

20.070.4 15.970.4 15.970.7 12.770.7

263 75 224 74 215 77 181 76

3.577 0.09 4.027 0.05 4.567 0.08 4.977 0.06

2.770.1 2.770.1 2.870.1 2.870.1

a

Indentation method of Anstis et al. [26]. SENB specimen with notch machined with UPLA.

b

2456

L. Melk et al. / Ceramics International 41 (2015) 2453–2461

3. Results and discussion 3.1. Processing of nanocomposites The measured bulk densities are given in Table 1 for the different composites where it can be seen that they drop from 99.4% for 3Y-TZP to 97.4% for the addition of 2 wt% CNT. For a given composition, the theoretical density (TD) was calculated according to the rule of mixtures. The zirconia TD was taken as 6.1 g/cm3 and that of MWCNTs as 1.8 g/cm3 since their exact density was not known and it depends of purity, number of walls and external

diameter of CNTs [32]. The drop in density of the composites seen in Table 1 is related to the presence of porosity connected to the CNT agglomerates. Fig. 1 shows the fracture surface of the composite with 2 wt% CNT content where some pores related to the clusters of MWCNTs could be observed. Moreover, the low density reported in this study is a common observation of practically all studies on 3Y-TZP/CNT composites [6–10,14–18]. Fig. 2 shows the fracture surfaces of the sintered nanocomposites for all studied compositions. The carbon nanotubes are well dispersed in the composite (blue arrows) with the addition of 0.5 wt% MWCNTs but increasingly agglomerated with the increase of MWCNT content. Fig. 3 shows the final microstructure of the SPSed composites after etching at high temperature in air. As MWCNT content increases, the microstructure reveals smaller grain size. A refinement of the grain size was observed from 174 to 148 nm for 0 up to 2 wt% CNT respectively (see Table 1). The decrease of grain size with adding MWCNTs content has been attributed to the presence of MWCNTs in grain boundaries which slows [9] or hinders the grain growth of zirconia grains [7]. 3.2. Hardness and elastic modulus

Fig. 1. Fracture surface showing the presence of porosity in the composite with 2 wt% CNT content.

Fig. 4 shows the Berkovich hardness and the elastic modulus of all composites in terms of penetration depth. It is clearly seen that the values are stabilised for penetration depth

Fig. 2. Fracture surfaces of the sintered nanocomposites; (a) 0.5 wt% CNT, (b) 1 wt% CNT, (c) 2 wt% CNT composites.

L. Melk et al. / Ceramics International 41 (2015) 2453–2461

2457

Fig. 3. Microstructure of (a) 0 wt% CNT, (b) 0.5 wt% CNT, (c) 1 wt% CNT, (d) 2 wt% CNT composites after etching in air at high temperature.

Fig. 4. Berkovich hardness and elastic modulus of all composites in terms of penetration depth.

larger than 400 nm, while for lower penetration depths, these properties are affected by surface defects and roughness. Moreover, both HBerk and EBerk of all composites are smaller than for monolithic 3Y-TZP (see Table 1). On the other hand, the Meyer hardness of 3Y-TZP/CNT diminishes in a similar way to HBerk, but with less scatter (see Table 1). Therefore, the trend is the same whether a small or relatively large surface area is tested. Since the ratio (EBerk/H)1/2 practically does not change with MWCNT content and the length of the indentation cracks is shorter with increasing MWCNT amount, the indentation KIc

increases (see Table 1). The value of KIc depends on whether the equation of Anstis et al. [26] or Niihara et al. [25] is used. Values in the range of 3.5–5.0 MPa m1/2 were found using Anstis et al. [26] (see Table 1), while the values were about 20% higher if KIc was calculated using the equation of Niihara et al. [25] Therefore, it was concluded that the indentation KIc increases for the compositions studied independently of the equation used. With respect to the 3Y-TZP matrix, the observed contribution of 0.5 wt% MWCNT to the indentation KIc, is in a good agreement with similar composition investigated by Mazaheri

2458

L. Melk et al. / Ceramics International 41 (2015) 2453–2461

et al. [7]. In the same direction, Ukai et al. [13] showed that the addition of 0.5 wt% CNTs improves indentation KIc in about 0.5 MPa m1/2, similar to the present results. However, our absolute values of indentation KIc are noticeably smaller even for the starting 3Y-TZP matrix.

3.3. True fracture toughness Fig. 5 shows the morphology of the notch induced by UPLA in the composite with 0 wt% CNT content. In order to get a close overview of the notch, the top surface of the notch (surface on which the notch was induced) was grinded step by step, then polished till approximately the mid depth is reached. The damage zone below the notch is hardly seen before fracture. However, the damage could easily be detected after fracture. See Fig. 6. Fig. 6 shows the fracture surface of the notched in four point bending specimens for 0, 0.5 and 2 wt% MWCNT, where three different regions are well differentiated. The first one (region A), at the top, corresponds to the notch, region B has a length of about  24 μm with a different fracture surface appearance, and region C at the bottom has the usual common surface fracture appearance of zirconia. The transition between regions B and C is clearly defined by a relatively sharp straight border normal to the crack advance (see Fig. 7).

Fig. 5. View of the notch induced by UPLA on specimen with 0 wt% CNT content.

The extension of region B increases slightly with the MWCNT content. This corresponds to the laser ablation damage region in front of the notch tip [29]. The crack length used in the calculation of KIc is easily identified as the notch length (region A) plus the length of the microcracked damaged zone (region B) clearly observed on the fracture surface. The effective length of the crack was taken as the length of the notch plus the length of region B where the damage was observed. Therefore, the different appearance of region B is associated, in the current study, to the coalescence of the damage induced by laser ablation in front of the notch during fracture. The damage is restricted to the length of region B and to a maximum average depth of 2 μm below the fracture surface. This can be appreciated in Fig. 8 where sections of region B (Fig. 8a) and of the border between regions B and C (Fig. 8b) are shown. On the left (region B) microcracking and porosity can be observed below the surface while just to the right in region C no trace of microcracking is observed. Similar observations using a notch tip induced by UPLA in conventionally sintered 3Y-TZP have been observed [29]. The true KIc values of the notched specimens using UPLA are given in Table 1, where it can be appreciated that KIc values hardly change with the addition of MWCNT content. Moreover, the true KIc of monolithic SPSed 3Y-TZP is found to be smaller (KIc ¼ 2.7 MPa m1/2) than the conventionally sintered monolithic 3Y-TZP of larger grain size ( 330 nm) where a KIc slightly higher than 4 MPa m1/2 was reported previously by the present authors [29]. It is interesting to notice that the higher values of KIc reported in the literature using the standard SEVNB method for the same composites include also a high KIc value for the starting monolithic 3Y-TZP matrix of about 6 MPa m1/2 [20]. It should be noticed that this is a very high value for 3Y-TZP with a grain size of 145 nm. Since the starting powder, grain size and density of SPS 3Y-TZP reported in [20] are very similar to the present study, the difference regarding the KIc reported here could be associated to the sharpness of the starting notches. The present low KIc for 177 nm grain size 3Y-TZP (2.7 MPa m1/2) is in line with the results of Eichler et al. [33] where they reported values between 2.8 and 3.1 MPa m1/2 for 3Y-TZP with grain sizes in the range between 110 and 210 nm. Therefore, it may be concluded that at least part of the low KIc values measured for

Fig. 6. Fracture surface of notched specimens: (a) 0 wt% CNT, (b) 0.5 wt% CNT and (c) 2 wt% CNT composites. A is the surface of the notch, B the microcracked region in front of the notch and C is the final fracture.

L. Melk et al. / Ceramics International 41 (2015) 2453–2461

2459

Fig. 7. High magnification of the transitions between regions A, B and C.

Fig. 8. (a) FIB cross section of the fracture surface below region B: (b) fracture surface with a trench at the border of regions B and C. The border coincides with the disappearance of damage below the surface.

the composites with respect to [20] is related to the matrix and possibly to the sharpness of the notch and length of microcracks induced in front of the notch. As is shown above, here the damage in the form of microcracks in front of the notch is clearly revealed on the fracture surface. It is much longer than the notch radius, and it ends abruptly at some distance of the notch tip. This is the reason for taking the length of the initial crack as the length of the notch plus the length of the microcracked zone as it has been shown in [29]. If only the notch length was considered as the crack length, then even much lower values of KIc would be

found. On the other hand, it is unlikely that the low value of KIc is related to the short depth of the shallow starting notch since the R-curve in nano grain size 3Y-TZP is relatively weak and very steep [33]. As seen in Table 1, the true KIc does not practically increase with MWCNT content. However, the contact damage resistance measured by indentation KIc increases. There were slight differences in the morphology of the crack paths with the addition of MWCNTs. However, the crack lengths are small and do not contribute significantly to the indentation fracture toughness. We rather associate the enhanced contact-damage resistance related to

2460

L. Melk et al. / Ceramics International 41 (2015) 2453–2461

Fig. 9. AFM topography image (3D view, 20  20 μm2) of nanoindentation imprints performed at maximum penetration depth on (a) 0 wt% CNT, (b) 0.5 wt% CNT, (c) 1 wt% CNT, (d) 2 wt% CNT composites.

a reduction in the driving force for indentation crack growth related to change in plastic deformation behavior below the indentation. The intense confined-shear under Vickers indentation in the presence of shear-deformable MWCNTs in the small agglomerates increases irreversible deformation by this mechanism as well as by compaction of porosity below the indenter instead of by plastic deformation. The reason for this behaviour is probably related to the weak interfacial bonding between MWCNTs and zirconia matrix.

3.4. Damage No fracture events like radial cracks at the corners of the imprints were detected under the Berkovich indentations (see Fig. 9). This indicates that the presence of fine grain microstructure (see Fig. 3) improves the damage tolerance under contact loading and it blocks the deformation mechanisms induced during the indentation process, which is similar to the behaviour reported in particle-dispersion-toughening of ceramic based nanocomposites [34]. The amount of t–m phase transformation below the indenter has not been measured in the present study. It was reported in a previous work [6], that t–m transformation is reduced in 3YTZP/CNT composites because of the small grain size. With the addition of 2 vol% MWCNT with grain size similar to the present study, t–m transformation at the fracture plane of Vickers indentation cracks was less than 5% at distances less than E2.0 mm below the fracture surface. Therefore less

transformation would be expected below the indenter since the main stress state is compression. 4. Conclusions True KIc of spark plasma sintered 3Y-TZP/CNT composites was successfully calculated using a novel method based on inducing a very sharp notch by UPLA. The true KIc for the SPSed 3Y-TZP matrix studied is smaller as compared to pressureless conventional sintered 3Y-TZP. Moreover, the true KIc hardly increases with the addition of MWCNT while the indentation KIc of the starting 3Y-TZP matrix is higher and it increases with MWCNTs content. It is concluded that the resistance to indentation cracking of the composites by adding MWCNTs to 3Y-TZP matrix does not indicate higher true fracture toughness. It is suggested that this is related to the drop in hardness and elastic modulus with the addition of MWCNTs content as it was detected using Berkovich nanoindentation. Acknowledgements The authors gratefully acknowledge the financial support given by the “Ministerio de Economía y Competitividad” of Spain through research grant MAT2011-23913. L. M. acknowledges the fellowship received from the Joint Doctoral Programme in Materials Science and Engineering (DocMASE). All authors thank Dr. T. Trifonov and Dr. E. Jiménez for their assistance in the FIB/SEM equipment and nanoindenter, respectively.

L. Melk et al. / Ceramics International 41 (2015) 2453–2461

References [1] J. Chevalier, L. Gremillard, A.V. Virkar, D.R. Clarke, The tetragonalmonoclinic transformation in zirconia: lessons learned and future trends, J. Am. Ceram. Soc. 92 (2009) 1901–1920. [2] R. Chintapalli, A. Mestra, F. García Marro, H. Yan, M. Reece, M. Anglada, Stability of nanocrystalline spark plasma sintered 3Y-TZP, Materials, 3, 2010, p. 800–814. [3] E. Zapata-Solvas, D. Gómez-García, A. Domínguez-Rodríguez, Towards physical properties tailoring of carbon nanotubes-reinforced ceramic matrix composites, J. Eur. Ceram. Soc. 32 (2012) 3001–3020. [4] G.-D. Zhan, J.D. Kuntz, J. Wan, A.K. Mukherjee, Single-wall carbon nanotubes as attractive toughening agents in alumina-based nanocomposites, Nat. Mater. 2 (2003) 38–42. [5] X. Wang, N.P. Padture, H. Tanaka, Contact-damage-resistant ceramic/ single-wall carbon nanotubes and ceramic/graphite composites, Nat. Mater. 3 (2004) 539–544. [6] R.K. Chintapalli, F.G. Marro, B. Milsom, M. Reece, M. Anglada, Processing and characterization of high-density zirconia-carbon nanotube composites, Mater. Sci. Eng., A. (2012). [7] M. Mazaheri, D. Mari, R. Schaller, G. Bonnefont, G. Fantozzi, Processing of yttria stabilized zirconia reinforced with multi-walled carbon nanotubes with attractive mechanical properties, J. Eur. Ceram. Soc. 31 (2011) 2691–2698. [8] N. Garmendia, I. Santacruz, R. Moreno, I. Obieta, Slip casting of nanozirconia/MWCNT composites using a heterocoagulation process, J. Eur. Ceram. Soc. 29 (2009) 1939–1945. [9] A. Datye, K.-H. Wu, G. Gomes, V. Monroy, H.-T. Lin, J. Vleugels, et al., Synthesis, microstructure and mechanical properties of Yttria Stabilized Zirconia (3YTZP)–Multi-Walled Nanotube (MWNTs) nanocomposite by direct in-situ growth of MWNTs on zirconia particles, Compos. Sci. Technol 70 (2010) 2086–2092. [10] J.-H. Shin, S.-H. Hong, Microstructure and mechanical properties of single wall carbon nanotube reinforced yttria stabilized zircona ceramics, Mater. Sci. Eng., A 556 (2012) 382–387. [11] J.P. Zhou, Q.M. Gong, K.Y. Yuan, J.J. Wu, Y.F. Chen, C.S. Li, et al., The effects of multiwalled carbon nanotubes on the hot-pressed 3 mol% yttria stabilized zirconia ceramics, Mater. Sci. Eng., A 520 (2009) 153–157. [12] J. Yi, T. Wang, Z. Xie, W. Xue, Zirconia-based nanocomposite toughened by functionalized multi-wall carbon nanotubes, J. Alloys Compd 581 (2013) 452–458. [13] T. Ukai, T. Sekino, A.T. Hirvonen, N. Tanaka, T. Kusunose, T. Nakayama, et al., Preparation and electrical properties of carbon nanotubes dispersed zirconia nanocomposites, Key Eng. Mater 317-318 (2006) 661–664. [14] J. Sun, L. Gao, M. Iwasa, T. Nakayama, K. Niihara, Failure investigation of carbon nanotube/3Y-TZP nanocomposites, Ceram. Int. 31 (2005) 1131–1134. [15] N. Garmendia, S. Grandjean, J. Chevalier, L.a. Diaz, R. Torrecillas, I. Obieta, Zirconia—multiwall carbon nanotubes dense nano-composites with an unusual balance between crack and ageing resistance, J. Eur. Ceram. Soc. 31 (2011) 1009–1014.

2461

[16] A. Duszová, J. Dusza, K. Tomášek, G. Blugan, J. Kuebler, Microstructure and properties of carbon nanotube/zirconia composite, J. Eur. Ceram. Soc. 28 (2008) 1023–1027. [17] P. Hvizdoš, V. Puchý, A. Duszová, J. Dusza, Tribological behavior of carbon nanofiber—zirconia composite, Scr. Mater 63 (2010) 254–257. [18] J. Dusza, G. Blugan, J. Morgiel, J. Kuebler, F. Inam, T. Peijs, et al., Hot pressed and spark plasma sintered zirconia/carbon nanofiber composites, J. Eur. Ceram. Soc. 29 (2009) 3177–3184. [19] G.D. Quinn, R.C. Bradt, On the vickers indentation fracture toughness test, J. Am. Ceram. Soc. 90 (2007) 673–680. [20] M. Mazaheri, D. Mari, Z.R. Hesabi, R. Schaller, G. Fantozzi, Multiwalled carbon nanotube/nanostructured zirconia composites: outstanding mechanical properties in a wide range of temperature, Compos. Sci. Technol 71 (2011) 939–945. [21] R. Damani, R. Gstrein, R. Danzer, Critical notch-root radius effect in SENB-S fracture toughness testing, J. Eur. Ceram. Soc. 16 (1996) 695–702. [22] J. Kübler, Fracture Toughness of Ceramics Using the SEVNB Method; Round Robin. [23] H. Fischer, A. Waindich, R. Telle, Influence of preparation of ceramic SEVNB specimens on fracture toughness testing results, Dent. Mater. 24 (2008) 618–622. [24] B. Milsom, G. Viola, Z. Gao, F. Inam, T. Peijs, M.J. Reece, The effect of carbon nanotubes on the sintering behaviour of zirconia, J. Eur. Ceram. Soc. 32 (2012) 4149–4156. [25] K. Niihara, A fracture mechanics analysis of indentation-induced Palmqvist crack in ceramics, J. Mater. Sci. Lett. 2 (1983) 221–223. [26] G.R. Anstis, P. Chantikul, B.R. Lawn, D.B. Marshall, A critical evaluation of indentation techniques for measuring fracture toughness: I, Direct crack measurements, J. Am. Ceram. Soc. 64 (1981) 533–538. [27] T. Fett, Estimated stress intensity factors for semi-elliptical cracks in front of narrow circular notches, Eng. Fract. Mech 64 (1999) 357–362. [28] D. Munz, T. Fett, Ceramics: Mechanical Properties, Failure Behaviour, Materials Selection, Springer, 1999. [29] M. Turon-Viñas, M. Anglada, Fracture toughness of zirconia from a shallow notch produced by ultra-short pulsed laser ablation, J. Eur. Ceram. Soc. 34 (2014) 3865–3870. [30] W.C. Oliver, G.M. Pharr, Measurement of hardness and elastic modulus by instrumented indentation: advances in understanding and refinements to methodology, J. Mater. Res. 19 (2011) 3–20. [31] I. Horcas, R. Fernández, J.M. Gómez-Rodríguez, J. Colchero, J. Gómez-Herrero, A.M. Baro, WSXM: a software for scanning probe microscopy and a tool for nanotechnology, Rev. Sci. Instrum 78 (2007) 013705. [32] C. Laurent, E. Flahaut, A. Peigney, The weight and density of carbon nanotubes versus the number of walls and diameter, Carbon (N. Y.) 48 (2010) 2994–2996. [33] J. Eichler, J. Rödel, U. Eisele, M. Hoffman, Effect of grain size on mechanical properties of submicrometer 3Y-TZP: fracture strength and hydrothermal degradation, J. Am. Ceram. Soc. 90 (2007) 2830–2836. [34] H. Awaji, S.-M. Choi, E. Yagi, Mechanisms of toughening and strengthening in ceramic-based nanocomposites, Mech. Mater. 34 (2002) 411–422.

Paper III The influence of unshielded small cracks in the fracture toughness of yttria and of ceria stabilised zirconia

Authors: Latifa Melk, Miquel Turón-Viñas, Joan Josep Roa, Marta-Lena Antti and Marc Anglada.

Paper Published in: Journal of the European Ceramic Society, 36 (2016) 147–153.

93

94

Journal of the European Ceramic Society 36 (2016) 147–153

Contents lists available at www.sciencedirect.com

Journal of the European Ceramic Society journal homepage: www.elsevier.com/locate/jeurceramsoc

The influence of unshielded small cracks in the fracture toughness of yttria and of ceria stabilised zirconia Latifa Melk a,b,c , Miquel Turon-Vinas a,b , Joan Josep Roa a,b , Marta-Lena Antti c , Marc Anglada a,b,∗ a

CIEFMA—Department of Materials Science and Metallurgical Engineering, ETSEIB, Universitat Politècnica de Catalunya, 08028 Barcelona, Spain CRnE, Campus Diagonal Sud, Edifici C’, Universitat Politècnica de Catalunya, 08028 Barcelona, Spain c Department of Engineering Sciences and Mathematics, Luleå University of Technology, 97187 Luleå, Sweden b

a r t i c l e

i n f o

Article history: Received 24 August 2015 Accepted 13 September 2015 Available online 26 September 2015 Keywords: Fracture resistance curves Fracture stress Zirconia 3Y–ZrO2 12Ce–ZrO2

a b s t r a c t The fracture toughness, KIC of two 3Y–ZrO2 with different grain size (177 and 330 nm) and 12Ce–ZrO2 were determined from a sharp micro-machined notch by Ultra-Short Pulsed Laser Ablation (UPLA) where a micro-cracked zone and non-transformed is generated in front of the notch. The notch plus the damage behaved as an unshielded edge surface crack. The fracture stress,  f of both 330 nm-3Y–ZrO2 and 12Ce–ZrO2 with similar short crack sizes were found to be comparable in despite of their different published R-curves. The results of KIC were discussed in terms of the type of cracks induced and by using a simple R-curve model. It was concluded that for the development of high strength composites with 12Ce–ZrO2 as the matrix, the relevant KIC that controls the  f with surface unshielded short cracks is much closer to the intrinsic KIC than to the indentation KIC or to the plateau KIC of long cracks. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction Fracture in advanced ceramics is caused by flaws with variable dimensions, generally of several tens of micrometers. In principle, it is possible to predict their fracture stress,  f , from fracture mechanics if the critical natural flaw and the fracture toughness, KIC , are precisely determined. However, an accurate prediction requires a specific knowledge of the shape of the critical flaw, which is difficult to be precisely modeled as a two dimensional crack. Regarding the actual KIC of zirconia ceramics with phase transformation toughening, KIC is characterized by a fracture resistance curve (KR ), which depends on crack length and which is also referred to as R-curve [1]. The R-curve is usually known for long cracks of several millimeters in length. It is often obtained from fracture testing prismatic specimens by the single edge V-notch beam (SEVNB) method. Nowadays, one of the zirconia ceramics widely used in biomedical applications is zirconia doped with 3 mol.% of yttria known as 3Y–ZrO2 . It has a high strength (>1000 MPa), but only a moderate √ KIC (4–5 MPa m) and it suffers from slow subcritical crack growth

∗ Corresponding author at: CIEFMA—Department of Materials Science and Metallurgical Engineering, ETSEIB, Universitat Politècnica de Catalunya, 08028 Barcelona, Spain. E-mail address: [email protected] (M. Anglada). http://dx.doi.org/10.1016/j.jeurceramsoc.2015.09.017 0955-2219/© 2015 Elsevier Ltd. All rights reserved.

and low temperature degradation (LTD) after exposure to humid environments [2]. One possible way to increase the KIC and the resistance to LTD is to develop new composites in which the matrix of zirconia is doped with ceria and the reinforcement is a harder phase such as alumina [3]. Although subcritical crack growth is still present [4], zirconia doped with ceria is resistant to LTD and it has a large transformability under stress, as measured by indentation methods, and a high plateau in the R-curve for ceria concentrations in the range of 9–12 mol.% [4–6]. However, when cracks are of dimensions similar to natural cracks, it is not obvious to which extend the R-curve and the crack size could determine the strength [7]. If the flaw population size and the transformation stress both decrease, then fracture may be controlled by phase transformation and it can be flaw independent in the very small flaw range. As the size of the critical flaw and transformation stress increase, fracture is controlled by the flaw size and the R-curve. There are experimental results from literature showing that natural small cracks tend to have lower R-curves [7]. The objective of the present work is to investigate the influence of transformation toughening on the  f and on the KIC of 12Ce–ZrO2 and 3Y–ZrO2 with similar small two dimensional artificial surface cracks of dimensions as close as possible to natural cracks. For this purpose, we have machined very shallow single edge notches by a novel method using Ultra-short Pulsed Laser Ablation (UPLA) on the surface of rectangular bending bars on two 3Y–ZrO2 with different

148

L. Melk et al. / Journal of the European Ceramic Society 36 (2016) 147–153

grain size, and on 12Ce–ZrO2 . These two zirconia ceramics have Rcurves with completely different plateau levels and steepness. The  f and the KIC are measured from a similar small crack induced in each material, and they are discussed in terms of their published R-curves. 2. Experimental procedure 2.1. Preparation of specimens Commercial powders of 3Y–ZrO2 (ZrO2 with 3 mol.% Y2 O3 ) from Tosoh (gradeTZ-3YSB-E) and 12Ce–ZrO2 (ZrO2 with 12 mol.% CeO2 ) from Daiichi Kigenso Kagaku Kogyo of Japan (grade CEZ-12SD) were cold and uniaxially pressed at 100 MPa in the form of bars (3 × 4 × 50 mm3 ), then sintered in an alumina tube furnace (model ST-18, Forns Hobersal S.L., Spain) at 1450 ◦ C with heating and cooling rates of 3 ◦ C/min and a holding time of 1 h and 2 h for 3Y–ZrO2 and 12Ce–ZrO2 , respectively. Using the Archimedes’ principle, the bulk materials result in a density of above 99.5% for 330 nm-3Y–ZrO2 and 99% for 12Ce-TZP of the theoretical density. The specimens were polished following the standard grinding and polishing methods down to 3 ␮m. Additional specimens of 3Y–ZrO2 were prepared by Spark Plasma Sintering (SPS) (SPS FCT HP D25I, FCT System Gm␤h) with a finer grain size (177 nm), and in the present study they will be referred to as 177 nm-3Y–ZrO2 . The samples were sintered at maximum temperature of 1350 ◦ C with heating and cooling rates of 100 ◦ C/min and a holding time of 5 min. A pressure of 50 MPa was maintained during the sintering cycle. Further information on sintering details can be found elsewhere [8]. The final samples were ceramic discs with 3 mm thickness and 40 mm diameter cut into bars (4 × 2.5 × 35 mm3 ) with a density of 99.4% of the theoretical density. The average grain size was determined by the conventional linear intercept method using a total of ten micrographs per specimen in order to have statistical significance. The micrographs were obtained by using a field emission scanning electron microscope (FESEM, JEOL 7001F). 2.2. The starting defect size A single edge V-notch was produced by UPLA on the center of the surface of prismatic specimens perpendicularly to the 4 mm width. Laser pulses (120 fs, 795 nm) were induced by means of a commercial Ti:sapphire oscillator (Tsunami, Spectra Physics) and a regenerative amplifier system (Spitfire, Spectra Physics) based on

chirped pulsed amplification. The pulses were linearly polarized and the repetition rate was 1 kHz. The pulse energy used was 5 ␮J and the focusing system used was an achromatic doublet lens with focal length of 50 mm. The samples were placed on a XYZ motorized stage and moved along one of the horizontal axis with a scanning speed of 50 ␮m/s. Four passes were needed to achieve the desired notch depth. The shape of the final notch was similar to a V-notch and its depth was ranged between 24 and 30 ␮m. Fracture surface analysis was done in a field emission scanning electron microscope (FESEM). The damage induced by UPLA below the surface was characterized using a dual beam focused ion beam (FIB)/FESEM (Neon40, Carl Zeiss AG, Germany). A thin platinum layer was deposited on the sample prior to FIB machining in order to flatten the surface and minimize ion-beam damage and curtain effect. A Gallium ion (Ga+ ) source was used to mill the surface at a voltage of 30 kV. The final polishing of the cross-sections was performed at 500 pA and the surface was observed by FESEM at regions around the notch. Raman spectra were collected with a spectrometer Jobin-Yvon LabRam HR 800 coupled to an Olympus optical microscope BXFM with an objective of 50× and to a CCD detector (liquid nitrogen cooled). A laser wavelength of 532 nm was used as source of excitation. The spectrum integration time was 30 s, with averaging the recorded spectra over two successive measurements. 2.3. Strength and fracture toughness measurements Finally, the single edge V-notched bar specimens were tested in a servohydraulic fatigue testing machine (model 8511, Instron, USA) using a four point bending test configuration with spans of 40/20 mm and under a high constant stress rate equal to 130 MPa s−1 in laboratory air. The number of specimens tested was 5 for the 330 nm-3Y–ZrO2 samples. In the case of 177 nm3Y–ZrO2 , the notched bar specimens were tested in a four point bending test device (DEBEN, Microtest, UK) in air with spans of 30/12 mm. The average stress rate was 2.4 MPa s−1 . The influence of strain rate on the  f of the cracked bars was studied in the case of 330 nm-3Y–ZrO2 specimens. No significant effect of strain rate has been found when two additional specimens were tested at different strain rates by a factor of 50. Three specimens were used for each composition, while only two specimens were used for the measurement of the  f of 12Ce–ZrO2 . The stress intensity factor (KI ) was obtained by using the following expression [9]: √ KI = Y a (1)   √  1.1215  5 1 3 6.1342ı 5 Y= (2) − ı + ı2 6 + exp − 8 12 8 8  3/2 where  f is the maximum stress on the surface which is given by: f =

Fig. 1. FESEM micrographs showing the grain size of (a) 177 nm-3Y–ZrO2, (b) 330 nm-Y–ZrO2 and (c) 12Ce–ZrO2 after thermal etching.

3F(S1 − S2 ) 2BW 2

(3)

S1 and S2 are the outer and inner spans respectively, B is the thickness, W is the width, F is the applied load, a is the crack length, ı = a/W, and  = 1 − ı. On the other hand, the residual imprint and the damage related to Vickers’ indentation imprints were assessed by inspecting the indentations through confocal laser scanning microscopy (CLSM, LEXT OL3 100, Olympus). The response to contact loading by Vickers indentation was determined by calculating the indentation fracture toughness (Kind ) with the purpose of comparing with other studies. Hence, the equation of Anstis et al. [10] for median radial cracks was used since it has often been used in the literature independently of the shape of the cracks. Kind was reported only for analysing the response to contact loading for comparative purposes, but it does

L. Melk et al. / Journal of the European Ceramic Society 36 (2016) 147–153

Fig. 2. Vickers cracks at imprints in (a) 177 nm-3Y–ZrO2, (b) 330 nm-3Y–ZrO2 and in (c) 12Ce–ZrO2.

Fig. 3. View of the notch and the micro-cracked zone in front of the notch in the center of the specimen [11].

149

150

L. Melk et al. / Journal of the European Ceramic Society 36 (2016) 147–153

Fig. 4. Fracture surface of 12Ce–ZrO2. (a) General view; (b) notch surface (zone A), fracture surface micro-cracked zone (zone B) and final fracture surface (zone C). (c) Detail of border between B and C zones. Table 1 Grain size, hardness and indentation fracture toughness.

a

Material

Grain size (nm)

HV10 (GPa)

Indentation KIC a (MPa

330 nm-3Y–ZrO2 177 nm-3Y–ZrO2 12Ce–ZrO2

330 ± 10 177 ± 14 1200 ± 28

13.5 ± 0.1 14.2 ± 0.1 9.2 ± 0.1

4.5 ± 0.1 3.6 ± 0.1 –

√ m)

Indentation method of Anstis et al. [10].

not represent the true KIC as it has been widely reported in the literature (see, for example, Quinn and Bradt [11]). 3. Results 3.1. Microstructure, hardness and Vickers cracks The microstructure of the three ceramics is shown in Fig. 1 where it can be observed the large difference in grain sizes. In Table 1, the average values of grain size are presented together with 10 kgf indentation Vickers hardness (HV10) and indentation fracture toughness (Kind ). The latter could be determined in both 330 nm-3Y–ZrO2 and 177 nm-3Y–ZrO2 as indentation cracks at the corners of the Vickers’s imprints were long enough under a load of 10 kgf. However, in the case of 12Ce–ZrO2, the indentation cracks could not be measured even under a load of 50 kgf (Fig. 2). The large amount of transformation that occurs near the indent impedes the indentation crack growth. 3.2. Small surface edge cracks One important point to be first considered is the radius of the notch tip and the damaged volume induced by UPLA. This has been previously reported in 330 nnm-3Y–ZrO2 where it was shown that a sharp notch with sub-micrometer notch tip radius less than ∼0.5 ␮m can be micro-machined. In front of the notch tip, there is a micro-cracked narrow region of ∼2 ␮m width which extends to

Fig. 5. The fracture surface of 12Ce–ZrO2 and its corresponding Raman spectra taken from different regions.

L. Melk et al. / Journal of the European Ceramic Society 36 (2016) 147–153 Table 2 Total crack length, fracture stress, and KIC of the materials investigated. Material

177 nm-3Y–ZrO2

330 nm-3Y–ZrO2

12Ce–ZrO2

Initial crack size (␮m) Fracture stress(MPa) √ KIC (MPa m)

52 ± 8 193 ± 6 2.7 ± 0.1

48 ± 10 309 ± 32 4.1 ± 0.4

47 ± 1 340 ± 13 4.5 ± 0.2

∼13 ␮m from the notch tip, see Fig. 3 reproduced from Ref. [12]. This region is highly micro-cracked with micro-cracks separated one from the neighbor by distances generally smaller than their size. The fracture surface of the notch (region A), the micro-cracked region (region B), and the final fracture surface (region C) are clearly distinguished, as can be appreciated in Ref. [12] for 330 nm3Y–ZrO2 and in Ref. [13] for 177 nm-3Y–ZrO2 . For 12Ce–ZrO2 , the fracture surface of 12Ce–ZrO2 presents also similar characteristic regions and they are illustrated in Fig. 4. It was important to analyze whether any structural change besides the appearance of the highly concentrated micro-cracked region was induced by UPLA on the specimens that could affect their fracture response. First, it can be seen in the micrograph view of the notch illustrated in Fig. 3(b) that the grains in the micro-cracked zone have the same size as the initial grains and as those located far away from the micro-craked zone. Then, by analyzing the three regions of the fracture surface using micro-Raman confocal spectroscopy in 177 nm-3Y–ZrO2 and 330 nm-3Y–ZrO2 , no significant presence of monoclinic phase in any of the regions could be found [12,13]. Similar result was obtained in 12Ce–ZrO2 for regions A and B, but in C the volume fraction of monoclinic phase was high, 57% (spectrum 4) and 67% (spectrum 5) as it is shown in Fig. 5. The average length of A + B regions (the crack length) and the average  f of the tested cracked specimens are shown in Table 2. The length of micro-cracked region B ranged generally between 40 and 25% of the total crack length under the specific conditions of UPLA used here to micro-machine the notch. KIC was calculated from Eqs. ((1)–(3)) and the result is shown in Table 2. 4. Discussion 4.1. Model KR (a) curves for 3Y–ZrO2 and 12Ce–ZrO2 The main mechanism for the high KIC of zirconia ceramics is toughening by phase transformation [14]. This leads to an increase of fracture toughness, KR , with crack extension, a, from an initial unshielded crack size a0 [7]. For any crack start propagating, the applied stress intensity factor, Kapp , has to be higher than the cracktip fracture toughness, K0 . KR (a) usually reaches a maximum value in the form of a plateau represented by Kmax . The total increase in fracture toughness for long cracks, Kmax = Kmax − K0 , as well as the steepness of the KR (a)-curve, depend on the capacity to shield the crack tip by transformation toughening. These two parameters of the R-curve play a central role in the measured fracture toughness and the  f . Thus, for 330 nm√ 3Y–ZrO2 , Kmax is relatively small (≤1.0 MPa m) and the initial part of the KR -curve is very steep so that Kmax is reached after crack extensions of a few tenths of micrometers [15,16]. In 177 nm3Y–ZrO2 , Kmax is not expected to be significantly higher than K0 , since for this small grain size there is hardly any contribution of transformation toughening. On the other hand, 12Ce–TZP is much more transformable than 330 nm-3Y–ZrO2 as it was shown either from the length of indentation cracks (see Fig. 2), the amount of monoclinic phase on the fracture surface using Raman spectroscopy, or from a previous detailed study by Deville et al. [17] on the transformation toughening in the region surrounding propagating cracks using atomic force microscopy. Therefore, Kmax may be comparatively large and the increase in KR is much less steep

151

than for 330 nm-3Y–ZrO2, so that crack extensions for reaching the plateau toughness are in the millimeter range [4,6]. A simple equation for KR (a) has been widely used in the literature to describe different steepness and plateaus levels [18]:

 a−a  0

KR (x) = Kmax − Kmax exp −

(4)



The applied stress intensity factor can generally be written as, √ Kapp (x) = Y a (5) Here a is the crack length and  represents the value of crack extension, a, for an increase of 63.2% of Kmax . Therefore, Kmax / is proportional to the steepness of the KR -curve,  is the applied stress and Y is a parameter that depends on the ratio between crack length and specimen width and on the geometry of the specimen. This ratio is very small, so that in the present discussion the parameter Y would be considered independent of crack length. Both Eqs. (4) and (5) can be written considering the following parameters: x≡

a − a0 ; a0

a0 ; 

˛=

ˇ=

K Kmax

(6)

Then, the normalized fracture resistance curve, KR /Kmax , and the normalized applied stress intensity factor, Kapp /Kmax , are given by: KR (x) = 1 − ˇ exp(−˛x) Kmax

(7)

Kapp (x)  = (1 − ˇ)(1 + x)1/2 Kmax 0

(8)

In the last equation,  0 is the theoretical fracture stress of a specimen with crack size a0 in the absence of a KR (a) curve, that is, when KR = K0 . The conditions for unstable fracture are given by [7]: Kapp (x) KR (x) = Kmax Kmax d dx



Kapp (x) Kmax



=

(9) d dx

 K (x)  R

(10)

Kmax

By applying both conditions to Eqs. (7) and (8), we obtain an equation from which x can be solved: exp(˛x) − ˇ(2˛(1 + x) − 1) = 0

(11)

By combining Eqs. (9) and (10), we can find an expression for the normalised fracture stress,  f / 0 , in terms of the normalized crack growth x*, of Eq. (11):

  2 f

0

=



2˛ (1 − ˇ)

2



ˇ exp(−␣x∗) − ˇ2 exp(−2␣x∗)

(12)

In Eq. (12), ˇexp (−␣x*) must be always smaller than 1 since  f / 0 must be positive. The expression in the square brackets has a maximum value of only 1/4 at exp (−˛x) = (1/2␤), so that a necessary condition for  f >  0 is: ˛ > 2(1 − ˇ)

2

(13)

For an explanation of the experimental results, we analyze the solutions of Eqs. (11) and (12) of the model of KR (a)-curve used here in Eq. (1) for several values of ˇ (0.2, 0.5, 0.65 and 0.80) and ˛ (0.05, 0.1, 0.5, 1, 5 and 10). Higher values of ˇ represent higher normalised increase in fracture resistance, while higher values of ˛ represent shorter values of  for a fixed initial crack length. Firstly, it can be seen that there are pairs of values of ˛ and ˇ which do not satisfy Eq. (13), so that only for specific combinations of ˛ and ˇ we have  f / 0 > 1. For the rest of combinations,  f should be equal to  0 as it can be seen in Fig. 6 where a semi-log plot of  f / 0 in terms of ␣ is presented for different values of ˇ.

152

L. Melk et al. / Journal of the European Ceramic Society 36 (2016) 147–153

4.2. Crack length for fracture toughness determination Because of the high density of micro-cracks in region B in front of the notch, their influence should be considered in the measurement of KIC . The effect of a distribution of micro-cracks in front of a notch in a brittle solid was analysed by Ortiz [19]. They concluded that if the density of micro-cracks is above a critical value at the start of loading, they will coalesce under external loading in front of the notch forming a larger crack before complete fracture occurs. We assume that this is also the case in the current study because of the high micro-crack density observed. Therefore, the crack induced by UPLA is taken as a surface edge crack with a length equal to the notch depth plus the length of the micro-cracked zone (regions A + B). 4.3. Fracture toughness The measured values of KIC for 177 nm-3Y–ZrO2 , 330 nm3Y–ZrO2 and 12Ce–ZrO2 are relatively small in comparison with values reported in the literature. Firstly, KIC for 177 nm-3Y–ZrO2 √ (2.7 MPa m) is smaller than for conventionally sintered 330 nm3Y–ZrO2 (see Table 2), presumably, because of smaller grain size, hence, there is much less transformation toughening. Moreover, the value found for KIC is close to the values found by Eichler et al. [16] √ where KIC of 2.8 and 3.1 MPa m for 3Y–ZrO2 with grain sizes in the range between 110 and 210 nm were reported. The KR (a) curve is practically absent because of very weak transformation toughening, so that practically KR (a) ∼ K0 , and no stable crack growth takes place before final fracture. KIC for 330 nm-3Y–ZrO2 reported here is close to usual values √ found in the literature (4–5 MPa m) but in the lower side of the scatter band. This is because, the plateau of the KR -curve of 330 nm3Y–ZrO2 is reached after small crack extension ( ≤ 50 ␮m) and √ √ Kmax is small (∼1 MPa m) with K0 ∼3 MPa m [14,15]. As the initial crack is unshielded and ∼50 ␮m in length, then by applying the above model we can assume that ˛ is in the range between 1 ( = 50 ␮m) and 5 ( = 10 ␮m) with ˇ ∼1/4 = 0.25. We can see then that Eq. (13) does not hold for ˛ = 1, but it does for ˛ ≥ 1.12 so that it is possible for  f and KIC to be higher than  0 and K0 , respectively if  is only slightly shorter than 50 ␮m. This may qualitatively explain that the experimental value for KIC from Table 2 √ (∼4.1 ± 0.4 MPa m) is higher than the crack-tip fracture toughness measured in 177 nm-3Y–ZrO2 . Therefore, the model is useful to explain the increase in fracture toughness of 330 nm-3Y–ZrO2 with respect to 177 nm-3Y–ZrO2 if  is smaller than 50 ␮m. The plateau of the KR -curve for 12Ce–ZrO2 has been measured for grain sizes of ∼1.5–1.6 ␮m [4,6] which are higher than in the present study (1.25 ␮m). The first studies reported extremely high

Fig. 6. Ratio of fracture stresses in terms of ˛ for different normalized maximum increase in fracture toughness (ˇ).

√ values of the plateau for 12Ce–ZrO2 of about 12 MPa m [4], while more recent studies show that for similar grain size the plateau √ is much lower ∼7 MPa m [6]. The corresponding maximum crack extension reported was ranged between just below the millimeter [6] to well over the millimeter [4]. Even with this strong discrepancy in the KR -curves, we shall try to justify our result, by assuming that Kmax is between 1 and 2 times K0 . For Kmax ∼2K0 and  ∼1000 ␮m, then ˛ ∼ 0.05 and ˇ ∼ 2/3 and for these pair of values Eq. (13) is not satisfied. This means that no crack extension will take place before fracture and  f will be given by  0 . If, on the other hand, Kmax ∼ K0 (ˇ = 0.5), we can see from Eq. (13) that the necessary condition for stable crack growth with ˇ = 0.5 is ˛ ≥ 0.5, so that no crack extension will take place. Even if we take the maximum crack extension of the KR -curve as half of the reported value in Ref. [6], that is,  ∼500 ␮m and ˛ = 0.1, there is no solution with  f >  0 . Therefore, this explains the reason for the low measured value of KIC for 12Ce–ZrO2 , which is only slightly higher than for 330 nm-3Y–ZrO2 . This is in line with the results of Fett et al. [20] who reported low values for the fracture toughness from small cracks induced in 9Ce–ZrO2 by FIB. A convenient graphical representation of the condition for stable crack growth can be obtained by plotting the squares of normalized fracture resistance and normalized stress intensity factor (Fig. 7). From Eq. (5), it can be seen that (Kapp (x)/Kmax )2 is a straight line with a slope given by (/ 0 ) (1 − ˇ). This is plotted for  =  0 , against normalized crack extension for two combinations of ˛ and ˇ which represent approximately the situation for 330 nm-3Y–ZrO2 and 12Ce–ZrO2 . (Kapp (x)/Kmax )2 for  >  0 and for an unshielded short crack of about 50 ␮m is always higher than the tangent to the square of the normalized crack resistant curve for any crack extension. It is important to recall that in smooth specimens with natural cracks, the strength of 330 nm-3Y–ZrO2 is much greater than that of 12Ce–ZrO2 which is usually attributed to a higher grain size in 12Ce–ZrO2 and transformation induced fracture [14]. In the present case, as the crack size from which fracture takes place is about twice the size of critical natural cracks, fracture is nucleated by the artificial short crack in all materials. In addition, they have similar length and geometry in all specimens and the difference in  f between 330 nm-3Y–ZrO2 and 12Ce–ZrO2 is only about 10%. Hence, the beneficial effect of higher indentation fracture toughness in 12Ce–ZrO2 in comparison with 3Y–ZrO2, which is widely referred in the literature, has a very small effect on the real relevant fracture toughness that determines the strength of unshielded small cracks of the dimensions studied. However, 12Ce–ZrO2 has a much higher damage tolerance to contact loading compared to

Fig. 7. Plots of the squares of normalized fracture resistance and normalized stress intensity factor (for  =  0 ) in terms or normalized crack extension for two KR -curves which approximately represent the behaviour of 330 nm-3Y–ZrO2 (˛ ∼1 and ˇ ∼1/4) and 12Ce–ZrO2 (˛ ∼0.1 and ˇ ∼2/3).

L. Melk et al. / Journal of the European Ceramic Society 36 (2016) 147–153

3Y–ZrO2 , since residual compressive stresses around the transformation indent in 12Ce–ZrO2 contribute to resist crack extension from the indentation corners (Fig. 2). Finally, in the determination of stable crack extension, the term Y of Eq. (5) was considered as independent of the crack length. Actually Kapp increases faster with crack extension than assumed, but this means that the common tangency condition of Eq. (10) would be even more difficult to satisfy for 12Ce–ZrO2 since then the slope of (Kapp (x)/Kmax )2 of Fig. 7 will be higher than assumed (see Fig. 7). For 330 nm-3Y–ZrO2 , with steeper KR -curves, crack extensions before fracture are very small so that the change in Y is also small and does not affect significantly the present discussion. 5. Conclusion As a summary, a very short and sharp notch can be induced in 3Y–ZrO2 and 12Ce–ZrO2 by a novel method using ultra-short laser ablation (UPLA) where a micro-cracked zone is generated in front of the notch tip. The notch plus the micro-cracked zone act as an unshielded sharp crack and the effective fracture toughness determined was found to be similar for both materials in spite of their large difference in the plateau and steepness of their R-curves. From the analysis of the results, it was concluded that for development of composites with a 12Ce–ZrO2 matrix, the effective relevant fracture toughness that controls the  f of the matrix is much closer to the crack-tip fracture toughness than to the high values determined by indentation or by using fracture mechanics standard long cracks. Acknowledgements The authors gratefully acknowledge Prof. Michael Reece of Nanoforce Ltd. and Dr. T. Trifonov of CRnE (UPC) for the use of SPS facilities and his assistance in the FIB/FESEM equipment, respectively. The present work was carried out within the scope of MAT2011-23913 project, supported by the Spanish “Ministerio de Economía y Competitividad”. We are grateful to “Direcció General de Recerca del Comissionat per a Universitat i Recerca de la Generalitat de Catalunya” for financial support (2014-SGR-130). L.M. acknowledges the fellowship award received from the European Joint Doctoral Program in Materials Science and Engineering (DocMASE) of the European Union. J.J.R acknowledges the Juan de la Cierva program (Grant number: JCI-2012-14454) fellowship award received from “Ministerio de Economía y Competitividad”. References [1] D. Marshall, M. Swain, Crack resistance curves in magnesia-partially-stabilized zirconia, J. Am. Ceram. Soc. 71 (1988) 399–407, http://dx.doi.org/10.1111/j.1151-2916.1988.tb05885.x.

153

[2] J. Chevalier, L. Gremillard, A.V. Virkar, D.R. Clarke, The tetragonal-monoclinic transformation in zirconia: lessons learned and future trends, J. Am. Ceram. Soc. 92 (2009) 1901–1920, http://dx.doi.org/10.1111/j.1551-2916.2009. 03278.x. [3] P. Palmero, M. Fornabaio, L. Montanaro, H. Reveron, Towards long lasting zirconia-based composites for dental implants. Part I: Innovative synthesis, microstructural characterization and in vitro stability, Biomaterials 50 (2015) 38–46, http://dx.doi.org/10.1016/j.biomaterials.2015.01.018. [4] C.S. Yu, D.K. Shetty, Transformation yielding, plasticity and crack-growth-resistance (R-curve) behaviour of CeO2-TZP, J. Mater. Sci. 25 (1990) 2025–2035, http://dx.doi.org/10.1007/BF01045759. [5] G. Rauchs, T. Fett, D. Munz, R-curve behaviour of 9Ce-TZP zirconia ceramics, Eng. Fract. Mech. 69 (2002) 389–401, http://dx.doi.org/10.1016/S00137944(01)00073-X. [6] H. El Attaoui, M. Saâdaoui, J. Chevalier, G. Fantozzi, Static and cyclic crack propagation in Ce-TZP ceramics with different amounts of transformation toughening, J. Eur. Ceram. Soc. 27 (2007) 483–486, http://dx.doi.org/10.1016/ j.jeurceramsoc.2006.04.108. [7] D. Munz, What can we learn from R-curve measurements? J. Am. Ceram. Soc. 90 (2007) 1–15, http://dx.doi.org/10.1111/j. 1551-2916.2006.01447.x. [8] R.K. Chintapalli, F.G. Marro, B. Milsom, M. Reece, M. Anglada, Processing and characterization of high-density zirconia–carbon nanotube composites, Mater. Sci. Eng. A (2012), http://dx.doi.org/10.1016/j.msea.2012.03.115. [9] D. Munz, T. Fett, Mechanical properties, failure behavior, Mater. Sel. 36 (1999), http://dx.doi.org/10.1007/978-3-642-58407-7. [10] G.R. Anstis, P. Chantikul, B. Lawn, D.B. Marshall, A critical evaluation of indentation techniques for measuring fracture toughness: I, direct crack measurements, J. Am. Ceram. Soc. 64 (1981) 533–538, http://dx.doi.org/10. 1111/j. 1151-2916.1981.tb10320.x. [11] G.D. Quinn, R.C. Bradt, On the vickers indentation fracture toughness test, J. Am. Ceram. Soc. 90 (2007) 673–680, http://dx.doi.org/10.1111/j.1551-2916. 2006.01482.x. [12] M. Turon-Vinas, M. Anglada, Fracture toughness of zirconia from a shallow notch produced by ultra-short pulsed laser ablation, J. Eur. Ceram. Soc. 34 (2014) 3865–3870, http://dx.doi.org/10.1016/j.jeurceramsoc.2014.05.009. [13] L. Melk, J.J. Roa Rovira, F. García-Marro, M.-L. Antti, B. Milsom, M.J. Reece, et al., Nanoindentation and fracture toughness of nanostructured zirconia/multi-walled carbon nanotube composites, Ceram. Int. (2015), http://dx.doi.org/10.1016/j.ceramint.2014.10.060. [14] R.H.J. Hannink, P.M. Kelly, B.C. Muddle, Transformation toughening in zirconia containing ceramics, J. Am. Ceram. Soc. 83 (2004) 461–487, http://dx.doi.org/ 10.1111/j.1151-2916.2000.tb01221.x. [15] D. Casellas, J. Alcalá, L. Llanes, M. Anglada, Fracture variability and R-curve behavior in yttria-stabilized zirconia ceramics, J. Mater. Sci. 36 (2001) 3011–3025, http://dx.doi.org/10.1023/A:1017923008382. [16] J. Eichler, M. Hoffman, U. Eisele, J. Rödel, R-curve behaviour of 2Y-TZP with submicron grain size, J. Eur. Ceram. Soc. 26 (2006) 3575–3582, http://dx.doi. org/10.1016/j.jeurceramsoc.2005.11.012. [17] S. Deville, H. El Attaoui, J. Chevalier, Atomic force microscopy of transformation toughening in ceria-stabilized zirconia, J. Eur. Ceram. Soc. 25 (2005) 3089–3096, http://dx.doi.org/10.1016/j.jeurceramsoc.2004.07.029. [18] C.S. Yu, D.K. Shetty, Transformation zone shape, size, and crack-growth-resistance (R-curve) behavior of ceria-partially-stabilized zirconia polycrystais, J. Am. Ceram. Soc. 72 (1989) 921–928, http://dx.doi.org/ 10.1111/j.1151-2916.1989.tb06245.x. [19] M. Ortiz, Micro-crack coalescence and macroscopic crack growth initiation in brittle solids, Int. J. Solids Struct. 24 (1988) 231–250, http://dx.doi.org/10. 1016/0020-7683(88)90031-5. [20] T. Fett, D. Creek, S. Wagner, G. Rizzi, C.A. Volkert, Fracture toughness test with a sharp notch introduced by focussed ion beam, Int. J. Fract. 153 (2008) 85–92, http://dx.doi.org/10.1007/s10704-008-9288-1.

Paper IV Material removal mechanisms by EDM of zirconia reinforced MWCNT nanocomposites

Authors:

Latifa Melk, Marta-Lena Antti and Marc Anglada.

Paper Published in: Journal of Ceramic International, 42 (2016) 5792-5801.

103

104

Available online at www.sciencedirect.com

CERAMICS INTERNATIONAL

Ceramics International 42 (2016) 5792–5801 www.elsevier.com/locate/ceramint

Material removal mechanisms by EDM of zirconia reinforced MWCNT nanocomposites Latifa Melka,b,c,n, Marta-Lena Anttic, Marc Angladaa,b a

CIEFMA-Department of Materials Science and Metallurgical Engineering, ETSEIB, Universitat Politècnica de Catalunya, 08028 Barcelona, Spain b CRnE, Campus Diagonal Sud, Edifici C’, Universitat Politècnica de Catalunya, 08028 Barcelona, Spain c Department of Engineering Sciences and Mathematics, Luleå University of Technology, 97187 Luleå, Sweden Received 27 November 2015; received in revised form 20 December 2015; accepted 21 December 2015 Available online 6 January 2016

Abstract Several composites of tetragonal zirconia polycrystals doped with 3 mol% yttria (3Y-TZP) and multiwalled carbon nanotubes (MWCNT) with concentrations from 0.5 to 4 wt% CNT were processed, spark plasma sintered, and characterised for a wide range of mechanical, electrical and thermal properties. In particular, a strong increase in electrical conductivity at room temperature was found with only 0.5 wt% CNT. However, the thermal conductivity was decreasing with increasing CNT content. Electrical discharge machining (EDM) using die sinking was carried out using the composites of 1 and 2 wt% CNT as workpieces. It was shown that both compositions could be successfully machined by EDM. The surface integrity and the subsurface were studied by SEM/FIB in order to determine the material removal mechanisms, which were found to be associated to spalling and melting/evaporation. Raman Spectroscopy was used to evaluate the damage of CNTs after EDM. & 2015 Elsevier Ltd and Techna Group S.r.l. All rights reserved.

Keywords: C. Electrical conductivity; Zirconia; Carbon nanotubes; Electrical discharge machining; Surface integrity

1. Introduction Nowadays there is a growing interest in using zirconia based ceramics especially zirconia doped with 3 mol% yttria (3YTZP) in a wide range of applications due to their extraordinary properties: high strength, high toughness, good chemical resistance and biocompatibility compared to other oxide ceramics [1]. However, the high hardness and brittleness of ceramic materials limits their ability to be machined with conventional techniques such as cubic boron nitride (cBN) or diamond tools [2]. One successful and attractive machining technology is the use of electrical discharge machining (EDM) which enables machining complex shapes with high precision up to the microscale level being independent of the material hardness. In EDM, series of controlled electrical discharges are generated n Corresponding author at: Department of Materials Science and Engineering, ETSEIB, Diagonal 647, 08028 Barcelona, Spain. Tel.: þ34 934016701; fax: þ34 934016706. E-mail address: [email protected] (L. Melk).

http://dx.doi.org/10.1016/j.ceramint.2015.12.120 0272-8842/& 2015 Elsevier Ltd and Techna Group S.r.l. All rights reserved.

between a tool electrode and a workpiece immersed in dielectric fluid resulting in removing material from the workpiece mainly through melting, evaporation and spalling [3–5]. A crucial requirement in EDM, is the need of an electrically conductive workpiece, which limits its applications to nonconductive ceramics such as ZrO2, Al2O3, Si3N4 and SiC [6–9]. In some studies, a conductive layer referred as assisting electrode (AE) was used on the surface of the workpiece to make it conductive [10], and it was successfully applied to zirconia ceramics [11]. Another approach consists of adding relatively large amounts of a conductive phase to 3Y-TZP such as WC, TiC, TiN, TiCN or TiB2 [12–15] to make the composite electrically conductive and therefore, susceptible machining by EDM. Lauwers et al. [16] reported the use of wire EDM in zirconia where 40 vol% WC was used as conductive phase. One important result of their work was that the grain size of the second phase material significantly influences the performance of ZrO2 ceramic composites. Finer microstructure results in a lower thermal conductivity and hence a higher cutting speed for materials where the predominant material removing rate (MRR) is melting.

L. Melk et al. / Ceramics International 42 (2016) 5792–5801

By using carbon nanotubes (CNTs) as a second phase in 3Y-TZP matrix, it is expected that only low CNT weight fractions will be needed to be used as conductive phase to make the composite an electrical conductor due to the large electrical conductivity of CNT. Recent studies by Hvizdos et al. [17] and by Ukai et al. [18] clearly show a strong increase in the electrical conductivity in ZrO2 reinforced with carbon nanofiber (CNF) and ZrO2-CNT composites respectively. An increase of electrical conductivity of several orders of magnitude was observed in both materials by incorporating low concentrations of carbon second phases. The high electrical conductivity was attributed to the formation of a 3D network structure of carbon phase. It has also been reported that Si3N4 with 5.3 vol% CNTs can be machined by EDM inducing low surface roughness as compared to TiN/Si3N4 composites due to the excellent electrical conductivity of CNTs [19]. In this study we aim to investigate the machinability of 3Y-TZP reinforced CNTs composites using EDM. The damage produced is investigated and the material removal mechanisms are determined. 2. Experimental procedure 2.1. Powder processing Zirconia stabilized with 3 mol% of yttria (TZ-3YSB-E, Tosoh Co, Japan) was reinforced with 0.5, 1, 2, 3 and 4 wt% of Multiwalled Carbon Nanotubes (MWCNTs) with outer diameter 10–15 nm, inner diameter 2–6 nm and length 0.1–10 mm (Graphistrength C100, Arkema, France). The detailed preparation of the composite powders has been described elsewhere [20,21]. The samples were sintered using spark plasma sintering (SPS) (SPS FCT HP D25I, FCT System Gmβh, Germany) at 1350 1C for a dwell time of 5 min, a heating rate of 100 1C/min and a pressure of 50 MPa maintained during the sintering cycle. The final samples were ceramic discs (40 mm  3 mm), which were grinded and polished with decreasing diamond pastes from 30 mm down to 3 mm and finally with colloidal silica. The average grain size was determined using the line intercept method on scanning electron microscopy (SEM) images and the density was determined by the Archimedes method. 2.2. Electrical and thermal conductivity The electrical conductivities of all composites were measured with an accurate digital micro-ohmmeter (Keithley 580)

5793

using the two-point method on bar-shaped specimens (4 mm  4 mm  18 mm ) prepared using a diamond cutting machine. The thermal conductivity ðK Þ was calculated in the composites with 0.5, 1 and 4 wt% CNT by measuring the thermal diffusivity ðαÞ, the density ðρÞ and the specific heat capacity ðCpÞ according to the equation: K ¼ α  ρ  Cp: The thermal diffusivity was determined by laser-flash method (Laser Flash Apparatus, Netzsch LFA 457, Germany) where disk-shaped specimens were thin coated with graphite, ASTM E2585-09 (2009) [22]. The specific heat capacity was measured under argon atmosphere using a differential scanning calorimeter (Netzsch DSC 404C, Germany). 2.3. Electrical discharge machining Die sinking EDM (“þ GF þ AgieCharmilles”) was carried out with 3Y-TZP/ 1 wt% CNT and 3Y-TZP/ 2 wt% CNTs composites as workpiece. The tool used was a copper tool electrode with a surface of 0.12 cm2. The main properties of the workpieces are listed in Table 1. EDM machining test was performed using different parameters, see Table 2. The average observed current was 0.5 A and the overall machining distance was set to a value of z¼  0.2 mm without rotation or planetary movement of the tool. The dielectric used was IonoPlus IME-MH. To achieve a surface roughness of Ra ¼ 0.1, first impulse set 1 and afterwards impulse set 2 (see Table 2) were used to machine the material under the boundary condition “Cu – Carbide Metal”. The composites subjected to EDM were in the shape of one quadrant of the sintered discs, clamped by the sides and machined on one diametric plane. 3. Characterization techniques The roughness of the machined composite and Cu electrode surfaces was measured after EDM using an industrial size profilometer MahrSurf XR/XT 20. The damage produced after EDM on the surface and subsurface of the composites was investigated using a dual beam Focused Ion Beam (FIB)/SEM Microscope (Zeiss Neon 40). For the observation of the cross sections, a thin platinum layer was deposited on the sample prior to FIB machining in order to

Table 1 Density, grain size and mechanical properties of the studied composites SPSed at 1350 1C. Composites

Relative density* (TD %)

Grain size (nm)

HV1 (GPa)

Indentation KIC (MPa√m)

3Y-TZPþ 0 wt% CNT 3Y-TZPþ 0.5 wt% CNT 3Y-TZPþ 1 wt% CNT 3Y-TZPþ 2 wt% CNT 3Y-TZPþ 3 wt% CNT 3Y-TZPþ 4 wt% CNT

99.47 0.2 98.87 0.3 97.67 0.3 96.67 0.2 94.37 0.2 91.97 0.2

177 714 176 710 161 711 148 79  

14.2170.09 12.9870.08 11.3270.09 9.5270.05 5.670.02 5.570.02

3.5770.01 4.0270.01 4.5670.08 4.9770.06 3.6570.07 3.5470.08

*

Density of CNTs: 1.8 g/cm3.

5794

L. Melk et al. / Ceramics International 42 (2016) 5792–5801

Table 2 Parameter sets used for EDM. EDM Parameters

Name Unit

Impulse set 1 Impulse set 2

Peak Current On Time Polarity (Tool) Off Time Compression Gain Voltage Planetary Tolerance Planetary Feed Increment Removal Rate Relative Wear Impulse Undersize Impulse Undersize Finishing VDI 3400 Class Ra

I T POL P Comp Gain U TOL INC Wv Theta M 2GAP

A μs (þ  þ  ) μs % pos V mm mm mm3/min % mm mm

3.2 2.1  4.9 10.2 15 250 0.012 0.35 0.2 34 0.12 0.09

3.2 2.1  4.9 10.2 15 250 0.012 0.031 0.1 32 0.09 0.07

K Ra

pos μm

21 1.12

19 0.90

minimize ion-beam damage. A Gallium (Ga þ ) ion source was used to mill the surface at a voltage of 30 kV. The final polishing of the cross-sections was performed at 10 pA. Energy Dispersive X-ray Spectroscopy (EDX) analysis was performed on EDM surface in order to study the chemical composition of the recast layer. Raman spectroscopy (alpha300RA þ from WITec GmbH) was used to analyse the integrity and the damage of MWCNTs after machining using a laser power of 0.6 mV. The laser wavelength excitation was 532 nm and the acquisition was ranged between 100 to 3000 cm  1. Phase identification was conducted by X-ray diffraction (XRD) with a Bruker AXS D8 diffractometer and Cu-Kα radiation.

Fig. 1. Electrical conductivity of 3Y-TZP and 3Y-TZP/CNT composites at room temperature.

Table 3 Previous studies on the electrical conductivities of 3Y-TZP base composites reinforced with carbon phases.

Mazaheri et al.[23] Shi and Liang [24] Hvizdoš et al.[17] Our study

Reinforcement and wt fraction

Electrical conductivity (S/m)

3 wt% MWCNT 10 wt% MWCNT 3 wt% CNF 3 wt% MWCNT

 100 65 813 888

4. Results and discussion 4.1. Electrical conductivity In Fig. 1, a strong increase in electrical conductivity by adding 0.5 wt% of CNT content can be appreciated. The addition of CNTs to zirconia matrix drastically improves the electrical conductivity as it is well documented in the literature [17,23,24], see Table 3. It can be seen that the electrical conductivity for zirconia-CNTs increases many orders of magnitude with respect to the value quoted in reference material data for zirconia [25], even for relatively low second phase weight fractions. The increase in electrical conductivity for the composites could be due to the large aspect ratio of CNTs which results in entangled network of conductive pathways, see Fig. 2. A similar network of CNTs in the grain boundaries of alumina were observed by Inam et al. [26], but not in alumina-black carbon composites. They reported an electrical conductivity of alumina-CNTs composites four times higher than that of alumina-black carbon composite due to the fibrous nature and high aspect ratio of CNTs.

Fig. 2. STEM image of 3Y-TZP- 2 wt% CNTs composite showing CNTs percolating network location under a Vickers indentation.

High electrical conductivity of composites of 3Y-TZP and carbon nanofibers (CNF) was found by Hvizdos et al. [17] for CNF content less than 3 wt%, which was also related to percolation and formation of a continuous intergranular network of CNFs. This increase in electrical conductivity has been reported for smaller amounts of carbon content (1 wt% CNTs) by Ukai et al. [18]. This change was associated to the formation of a 3D network structure of MWCNTs in 3Y-TZP matrix which results in low electrical resistivity and allows percolation at small volume fraction of CNTs. The difference in the concentration and the large electrical conductivity reported by different authors in literature could be related to the different dimensions and to distribution of carbon phase in the composite. Furthermore, Inam et al. [27] showed

L. Melk et al. / Ceramics International 42 (2016) 5792–5801

Fig. 3. Thermal conductivity of 3Y-TZP and 3Y-TZP-CNT composites from 25 to 900 1C SPSed at 1350 1C.

by studying alumina-CNT composites that the temperature and the dwell time have an influence on the electrical conductivity during SPS. They found that increasing temperature during SPS is more effective in increasing the electrical conductivity than increasing dwell times. They reported an electrical conductivity of  390 W/Km at 1400 1C compared to  600 W/Km at 1800 1C using dwell times of 20 min and 3 min respectively. 4.2. Thermal conductivity Fig. 3 shows the effect of CNTs addition (0.5, 1 and 4 wt%) on the thermal conductivity of zirconia matrix from room temperature up to 900 1C. The thermal conductivity of the composite with 2 wt% CNT was not possible to measure due to too small dimensions of the sample. The thermal conductivity of the composites decreases by increasing CNT content and temperature in the range studied. Therefore, even if there is a percolation, it does not have the same effect as it has on the electrical conductivity. In Fig. 3, the thermal conductivity of 3Y-TZP with a grain size of 179 nm [28] is also presented and it is similar to that produced here by SPS. It can be noticed that the thermal conductivity decreases with temperature, with absolute values between 2.83 and 2.03 W mK  1 in the range 100–1000 1C [28]. Regarding the thermal conductivities of CNTs, there is substantial data scatter in the literature concerning thermal conductivity values at room temperature. Commonly quoted values for individual CNTs are  3000 W mK  1 for MWCNTs and  3500 W mK  1 for single-walled carbon nanotubes (SWCNTs) at room temperature [29]. To the best of our knowledge, there has not been any work on the thermal conductivity of 3Y-TZP-CNT in order to compare with the present work. However, the influence of CNTs on the thermal conductivity of alumina [30,31] as well as on Si3N4-based nanocomposites with the addition of 6 vol% and 8.9 vol% of SWCNTs [32] and MWCNT [33] respectively, are similar to the present results. In general, a decrease in thermal conductivity with CNT content and with temperature is found independently of the matrix studied. The decrease of thermal

5795

Fig. 4. Thermal conductivity of zirconia-CNT composites from 25 up to 900 1C SPSed at 1500 1C.

conductivity with CNT content may be explained by the increase in porosity at concentrations higher than 1 wt% CNT which is apparent in the density and the hardness (see Table 1) as well as in the drop of the elastic modulus [21]. Bakshi et al. [34] showed that the thermal conductivity of Al2O3 coatings using 4 wt% MWNTs increases with better CNT dispersion. According to Kumari et al. [35], the thermal properties of Al2O3-CNT composites are highly dependent on the CNT content, bulk density, and SPS conditions. However, in both studies, the maximum increase in thermal conductivity is far away from the values that could be expected considering the high thermal conductivity of CNT. In the present case, increasing the temperature of SPS to 1500 1C did not show any significant effect on the thermal conductivity of 0.5 and 1 wt% CNT compositions (see Fig. 4). Only a small decrease is observed at high temperature for 1 wt% CNT, but one would expect a higher thermal conductivity because of the increase in grain size when SPS is carried out at 1500 1C.

4.3. EDM After removing the composites from the clamping fixture, a successful EDM machined cavity was clearly visible on the surface (see Fig. 5). The machined cavity shows inhomogeneous removal of material which may be due to the inhomogeneity in CNT distribution. Examples of surface roughness of the machined composites with 1 and 2 wt% CNT and Cu surfaces are shown for representative tests in Tables 4 and 5. Fig. 6 shows the EDM surfaces of composites with 1 and 2 wt % CNT content where material was removed forming shallow and deep cavities. These cavities are produced by flakes that were detached from the surface of the composite workpiece by means of a spalling mechanism (see Fig. 6a). It is widely accepted that spalling is the main removal material mechanism for non-conductive fragile ceramics due to the generation of cracks that separate surface volumes of material during successive discharges [11]. It has been reported that thermal-shock induced spalling mechanism provides more material removal than through the mechanisms of melting, evaporation or dissociation [36].

5796

L. Melk et al. / Ceramics International 42 (2016) 5792–5801

Spalling was found to be more dominant in the composite with 1 wt% where the electrical conductivity and fracture toughness are lower compared to the composite with 2 wt% CNT content.

Fig. 5. Machined cavity on the composite with 1 wt% CNT and on the copper tool electrode.

Debris was observed in the form of particles with both rounded and sharp edges confirming the presence of spalling as removal material mechanism [12], see Fig. 7. SEM pictures in Fig. 8 reveal the presence of many droplets in clusters in both composites. Due to the high temperature achieved during EDM, the melting temperature can be eventually reached and this also contributes to material removal. Some of the molten material disappears and some fraction of the melt solidifies on the EDM surface forming the recast layer. Usually, melting has been observed as the main material removal mechanism in metals [37]. However, Bonny et al. [38] reported full melting as material removal mechanism in the EDM of ZrO2–WC composites. Melting was also reported in ZrO2–TiN and Al2O3–SiC–TiC using wire EDM [12]. In Fig. 8(b) a recast layer with a large network of cracks is displayed on the EDM surface when observed at higher magnification. These crack networks are probably related to thermal shock because of the low thermal conductivity of the composite as well as to the difference in thermal expansion between the recast layer and the composite substrate during dielectric cooling. In order to assess the thickness and the composition of the recast layer, a Focused Ion Beam (FIB) cross section was

Table 4 Surface roughness analysis of 3Y-TZP/1 wt% CNT composite. Surface roughness

Ra (mm)

Rz (mm)

Rmax (mm)

Surface of 3Y-TZP/1 wt% CNT Tool electrode prior machining Tool electrode after machining

2.44 0.2 1.06

16.34 1.13 5.15

26.37 1.28 6.50

Table 5 Surface roughness analysis of 3Y-TZP/2 wt% CNT composite. Surface roughness

Ra (mm)

Rz (mm)

Rmax (mm)

3Y-TZP/2 wt% CNT composite Tool electrode prior machining Tool electrode after machining

4.18 0.19 0.49

22.23 1.28 2.95

30.92 1.70 3.27

Fig. 7. SEM picture showing the presence of submicron debris on the EDM surface of the composite with 2 wt% CNT content.

Fig. 6. SEM pictures showing EDM surfaces of the composites with a) 1 wt% CNT content b) 2 wt% CNT content.

L. Melk et al. / Ceramics International 42 (2016) 5792–5801

5797

Fig. 8. Melting/evaporation in composite with a) 1 wt% CNT content and b) 2 wt% CNT content.

Fig. 9. FIB cross section in the recast layer of the composite with 2 wt% CNT content. b) High magnification of a).

performed below the EDM surface of the composites with 1 and 2 wt% CNT content, see Figs. 9 and 10. In Fig. 9, one can observe a crack below the recast layer in the composite with 2 wt% CNT which runs parallel to the surface and bridged by CNTs. More cracking is observed with cracks in different directions in Fig. 10. At high magnification some nanometric structure in the recast layer can be observed which may be the reaction product between zirconium and

carbon which results in ZrC, while the white phase may be the solidified zirconia in excess for the reaction. In addition, the presence of many grain boundary microcracks below the recast layer can be clearly observed. The heat affected zone (HAZ) has a thickness of 2.5 μm and 1 μm for the composites with 1 and 2 wt% CNT respectively. The sub-surface of both composites contains cracks that are parallel and perpendicular to the machined surface in addition to some pores which are

5798

L. Melk et al. / Ceramics International 42 (2016) 5792–5801

Fig. 10. FIB cross section in the recast layer of the composite with 1 wt% CNT content. b) high magnification of a).

Fig. 11. EDX analysis of the recast layer surface of the composite 3Y-TZP/2 wt% CNT where two different regions a) and b) were analysed.

related to evaporation of CNTs. The HAZ in the studied composites is relatively shallow and this can probably be due to the low thermal conductivity of the composites.

EDX analysis of two molten regions taken from the recast layer of the composite with 2 wt% CNT (Fig. 11) shows the presence within the molten region of zirconium, oxygen and

L. Melk et al. / Ceramics International 42 (2016) 5792–5801

5799

Fig. 12. Raman spectroscopy of the machined surface of the composite with 2 wt% CNT content showing different regions where Raman spectra was taken from each region.

carbon, but also of copper (Cu) due to the use of copper electrode during EDM. However, only a small amount of Cu was detected and in specific regions while in other regions Cu was not found. Raman spectroscopy was used to analyse both the MWCNTs damage and the recast layer of the machined surfaces of the composites (Fig. 12). All the Raman spectra present the three characteristic peaks of MWCNTs. The Dband at about 1350 cm  1 that corresponds to defects/disorder in MWCNTs, the G-band at about 1580 cm  1 indicating the MWCNTs crystallinity and the G’-band at about 2700 cm  1 that is attributed to the overtone of D band. The intensity ratio between the D and G bands (ID/IG) is commonly used as a tool to quantify the crystallinity of the CNTs [39]. The Raman spectra regions with blue and red colours show the typical D and G bands of MWCNTs indicating that MWCNTs have survived after EDM. However, in the regions

with green and yellow colours, broad Raman spectra are shown. It has been reported in [40] that the broadening of the bands is related to the amorphous character of CNTs. As the temperature of EDM reaches around 2000 1C [3], the heat could induce disorder and defects of MWCNTs. Similar study has been reported where Raman was used to study the damage of MWCNTs after machining Si3N4/ MWCNTs composites using EDM die sinking. It was shown that as the voltage increases the damage and degradation of MWCNTs become larger [41]. XRD measurements on the EDM surface of the composite with 2 wt% CNT content reveal that there has been an interaction between the composite phases, see Fig. 13. Formation of zirconium carbide (ZrC) was observed but no zirconium–copper ZrCu formation has been detected even if there was present some Cu from EDX results in Section 4.3. The formation of ZrC has been previously observed on the EDM surface of ZrO2 toughened

5800

L. Melk et al. / Ceramics International 42 (2016) 5792–5801

measurements. Dr. Andreas Rebschläger from Zentrum für Mechatronik und Automatisierungstechnik (ZeMA), Germany for EDM. Dr. Elena Bailo from WITec GmbH, Spain for the use of Raman spectroscopy and the financial support given by the “Ministerio de Ciencia e Innovación”, Spain through research grant MAT2011-23913. The authors acknowledge the UE for financial support to the e-Create-Network of the Rise program. L. Melk acknowledges the fellowship award received from the European Joint Doctoral Programme in Materials Science and Engineering (DocMASE) of the European Union. Finally, all authors thank Dr. Trifon Trifonov and Dr. Joan Josep Roa from UPC for their assistance in the FIB/SEM and STEM equipment. Fig. 13. XRD spectra of the EDM surface of the composite 3Y-TZP/ 2 wt% CNT.

WC composites and it was reported to be related to the high temperature achieved during spark erosion. The formation of ZrC also occurs during the sintering of zirconia/WC at temperatures higher than 1750 1C [42]. The surface finish of the EDM specimens should be improved because of relatively large roughness as can be seen in Tables 4 and 5. Furthermore, a detailed study is needed to evaluate the influence of the damage induced on the resistance to contact loading which could be severely affected because of the presence of surface microcracks. However, the defects that were induced by EDM are small compared with the critical defect size which is approximately an order of magnitude larger. Therefore, it is expected that the strength of EDM specimens will not suffer substantial loss. On the other hand, the influence of the molten recast layer and residual stresses could also affect the strength. Thus, more work is also needed in this direction. 5. Conclusions The addition of CNTs to zirconia matrix induces a strong increase in electrical conductivity and relatively slight changes in thermal conductivity. This makes the composites susceptible to machining by electrical discharge machining. The study of the surface integrity of the composites with 1 and 2 wt% CNT content after EDM reveals that the material removal mechanisms in both composites were melting/evaporation and spalling. A recast layer with a thickness of few microns in depth is formed in which the presence of ZrC is detected. Raman spectroscopy shows that some CNTs have survived while others were damaged and disordered due to the high temperature induced by spark erosion. Acknowledgments The authors gratefully acknowledge Prof. Michael Reece of Nanoforce Ltd for the use of SPS facilities. Prof. Andreas Kaiser and Dr. Nikolaos Bonanos from Technical University of Denmark (DTU) for thermal and electrical conductivity

References [1] I. Birkby, R. Stevens, Applications of zirconia ceramics, Key Eng. Mater. 122–124 (1996) 527–552, http://dx.doi.org/10.4028 /www.scientific.net/KEM.122-124.527. [2] O. Malek, J. Vleugels, Y. Perez, P. De Baets, J. Liu, S. Van den Berghe, et al., Electrical discharge machining of ZrO2 toughened WC composites, Mater. Chem. Phys. 123 (2010) 114–120, http://dx.doi.org/10.1016 /j.matchemphys.2010.03.069. [3] K. Ho, S. Newman, State of the art electrical discharge machining (EDM), Int. J. Mach. Tools Manuf. 43 (2003) 1287–1300, http://dx.doi.org/10.1016/S0890-6955(03)00162-7. [4] M. Kunieda, B. Lauwers, K.P. Rajurkar, B.M. Schumacher, E.D. M. Advancing, Advancing EDM through fundamental insight into the process, CIRP Ann.-Manuf. Technol. 54 (2005) 64–87, http://dx.doi.org/10.1016/S0007-8506(07)60020-1. [5] H.C. Tsai, B.H. Yan, F.Y. Huang, EDM performance of Cr/Cu-based composite electrodes, Int. J. Mach. Tools Manuf. 43 (2003) 245–252, http://dx.doi.org/10.1016/S0890-6955(02)00238-9. [6] A. Banu, M.Y. Ali, M.A. Rahman, Micro-electro discharge machining of non-conductive zirconia ceramic: investigation of MRR and recast layer hardness, Int. J. Adv. Manuf. Technol. 75 (2014) 257–267, http://dx.doi.org/10.1007/s00170-014-6124-9. [7] A. Muttamara, Y. Fukuzawa, N. Mohri, T. Tani, Effect of electrode material on electrical discharge machining of alumina, J. Mater. Process. Technol. 209 (2009) 2545–2552, http://dx.doi.org/10.1016 /j.jmatprotec.2008.06.018. [8] K. Liu, J. Peirs, E. Ferraris, Micro Electrical Discharge Machining of Si3N4-based ceramic composites, Proceedings of the 4th International Conference on Multi-Material Micro Manufacture (2008) 161–166. [9] Y. Fukuzawa, N. Mohri, T. Tani, a Muttamara, Electrical discharge machining properties of noble crystals, J. Mater. Process. Technol. 149 (2004) 393–397, http://dx.doi.org/10.1016/j.jmatprotec.2003.12.028. [10] N. Mohri, Y. Fukuzawa, T. Tani, N. Saito, K. Furutani, Assisting electrode method for machining insulating ceramics, CIRP Ann.-Manuf. Technol. 45 (1996) 201–204, http://dx.doi.org/10.1016/S0007-8506 (07)63047-9. [11] A. Sabur, M.Y. Ali, M.A. Maleque, A.A. Khan, Investigation of material removal characteristics in EDM of nonconductive ZrO2 ceramic, Procedia Eng. 56 (2013) 696–701, http://dx.doi.org/10.1016/j.proeng.2013.03.180. [12] B. Lauwers, J.P. Kruth, W. Liu, W. Eeraerts, B. Schacht, P. Bleys, Investigation of material removal mechanisms in EDM of composite ceramic materials, J. Mater. Process. Technol. 149 (2004) 347–352, http://dx.doi.org/10.1016/j.jmatprotec.2004.02.013. [13] K. Bonny, P. De Baets, J. Vleugels, A. Salehi, O. Van der Biest, B. Lauwers, et al., Influence of secondary electro-conductive phases on the electrical discharge machinability and frictional behavior of ZrO2based ceramic composites, J. Mater. Process. Technol. 208 (2008) 423–430, http://dx.doi.org/10.1016/j.jmatprotec.2008.01.020.

L. Melk et al. / Ceramics International 42 (2016) 5792–5801 [14] S. Salehi, O. Van der Biest, J. Vleugels, Electrically conductive ZrO2–TiN composites, J. Eur. Ceram. Soc. 26 (2006) 3173–3179, http://dx.doi.org/10.1016/j.jeurceramsoc.2005.10.010. [15] K. Bonny, P. De Baets, J. Vleugels, O. Van der Biest, A. Salehi, W. Liu, et al., Reciprocating sliding friction and wear behavior of electrical discharge machined zirconia-based composites against WC–Co cemented carbide, Int. J. Refract. Met. Hard Mater. 27 (2009) 449–457, http://dx.doi.org/10.1016/j.ijrmhm.2008.07.004. [16] B. Lauwers, K. Brans, W. Liu, J. Vleugels, S. Salehi, K. Vanmeensel, Influence of the type and grain size of the electro-conductive phase on the Wire-EDM performance of ZrO2 ceramic composites, CIRP Ann. -Manuf. Technol. 57 (2008) 191–194, http://dx.doi.org/10.1016 /j.cirp.2008.03.089. [17] P. Hvizdoš, V. Puchý, A. Duszová, J. Dusza, C. Balázsi, Tribological and electrical properties of ceramic matrix composites with carbon nanotubes, Ceram. Int. 38 (2012) 5669–5676, http://dx.doi.org/10.1016 /j.ceramint.2012.04.010. [18] T. Ukai, T. Sekino, A.T. Hirvonen, N. Tanaka, T. Kusunose, T. Nakayama, et al., Preparation and electrical properties of carbon nanotubes dispersed Zirconia nanocomposites, Key Eng. Mater. 317–318 (2006) 661–664. [19] C. Balázsi, B. Fényi, N. Hegman, Z. Kövér, F. Wéber, Z. Vértesy, et al., Development of CNT/Si3N4 composites with improved mechanical and electrical properties, Compos. Part B Eng. 37 (2006) 418–424, http://dx.doi.org/10.1016/j.compositesb.2006.02.006. [20] B. Milsom, G. Viola, Z. Gao, F. Inam, T. Peijs, M.J. Reece, The effect of carbon nanotubes on the sintering behaviour of zirconia, J. Eur. Ceram. Soc. 32 (2012) 4149–4156, http://dx.doi.org/10.1016/j.jeurceramsoc. 2012.07.028. [21] L. Melk, J.J. Roa Rovira, F. García-Marro, M.-L. Antti, B. Milsom, M. J. Reece, et al., Nanoindentation and fracture toughness of nanostructured zirconia/multi-walled carbon nanotube composites, Ceram. Int. 41 (2015) 2453–2461. http://dx.doi.org/10.1016/j.ceramint.2014.10.060. [22] ASTM Standard E2585-09, ASTM Standard E2585-09: Standard practice for thermal diffusivity by the flash method, ASTM, Int. 2009. [23] M. Mazaheri, D. Mari, Z.R. Hesabi, R. Schaller, G. Fantozzi, Multiwalled carbon nanotube/nanostructured zirconia composites: Outstanding mechanical properties in a wide range of temperature, Compos. Sci. Technol. 71 (2011) 939–945, http://dx.doi.org/10.1016 /j.compscitech.2011.01.017. [24] S.L. Shi, J. Liang, Effect of multiwall carbon nanotubes on electrical and dielectric properties of yttria-stabilized zirconia ceramic, J. Am. Ceram. Soc. 89 (2006) 3533–3535, http://dx.doi.org/10.1111/j.1551-2916 .2006.01232.x. [25] W. Martienssen, H. Warlimont (Eds.), Springer Handbook of Condensed Matter and Materials Data, 2006. [26] F. Inam, H. Yan, D.D. Jayaseelan, T. Peijs, M.J. Reece, Electrically conductive alumina–carbon nanocomposites prepared by Spark Plasma Sintering, J. Eur. Ceram. Soc. 30 (2010) 153–157, http://dx.doi.org/10.1016/j.jeurceramsoc.2009.05.045. [27] F. Inam, Development of Ceramic – Carbon Nanotube (CNT) Nanocomposites, School of Engineering and Materials Science, Queen Mary, University of London, 2009.

5801

[28] A.M. Limarga, D.R. Clarke, The grain size and temperature dependence of the thermal conductivity of polycrystalline tetragonal yttria-stabilised zirconia, Appl. Phys. Lett. 98 (2012) 211906, http://dx.doi.org/10.1063/1.3593383. [29] A. Balandin, Thermal properties of graphene and nanostructured carbon materials, Nat. Mater. 10 (2011) 569–581, http://dx.doi.org/ 10.1038/NMAT3064. [30] H.L. Zhang, J.F. Li, K.F. Yao, L.D. Chen, Spark plasma sintering and thermal conductivity of carbon nanotube bulk materials, J. Appl. Phys. 97 (2005) 10–15, http://dx.doi.org/10.1063/1.1927286. [31] O. Hanzel, J. Sedlá, E. Hadzimová, P. Sajgalík, Thermal properties of alumina–MWCNTs composites, J. Eur. Ceram. Soc. 35 (2015) 1559–1567, http://dx.doi.org/10.1016/j.jeurceramsoc.2014.11.032. [32] E.L. Corral, H. Wang, J. Garay, Z. Munir, E. V Barrera, Effect of singlewalled carbon nanotubes on thermal and electrical properties of silicon nitride processed using spark plasma sintering, J. Eur. Ceram. Soc. 31 (2011) 391–400, http://dx.doi.org/10.1016/j.jeurceramsoc.2010.10.020. [33] P. Miranzo, E. García, C. Ramírez, J. González-Julián, M. Belmonte, M. Isabel Osendi, Anisotropic thermal conductivity of silicon nitride ceramics containing carbon nanostructures, J. Eur. Ceram. Soc. 32 (2012) 1847–1854, http://dx.doi.org/10.1016/j.jeurceramsoc.2012.01.026. [34] S.R. Bakshi, K. Balani, A. Agarwal, Thermal conductivity of plasmasprayed aluminum oxide—multiwalled carbon nanotube composites, J. Am. Ceram. Soc. 91 (2008) 942–947, http://dx.doi.org/10.11 11/j.1551-2916.2007.02081.x. [35] L. Kumari, T. Zhang, G. Du, W. Li, Q. Wang, A. Datye, K.H. Wu, Thermal properties of CNT-Alumina nanocomposites, Compos. Sci. Technol. 68 (2008) 2178–2183, http://dx.doi.org/10.10 16/j.compscitech.2008.04.001. [36] C. Trueman, J. Huddleston, Material removal by spalling during EDM of ceramics, J. Eur. Ceram. Soc. 20 (2000) 1629–1635, http://dx.doi.org/10.1016/S0955-2219(00)00027-3. [37] T.H. Hocheng, National Tsing Hua University (Eds.), Machining Technology for Composite Materials.Principles and Practice, Elsevier Science, Taiwan, 2011. [38] K. Bonny, P. De Baets, J. Vleugels, a Salehi, O. Van Der Biest, B. Lauwers, et al., EDM machinability and frictional behavior of ZrO2– WC composites, Int. J. Adv. Manuf. Technol. 41 (2009) 1085–1093, http://dx.doi.org/10.1007/s00170-008-1551-0. [39] J. Schwan, S. Ulrich, V. Batori, H. Ehrhardt, S.R.P. Silva, Raman spectroscopy on amorphous carbon films, J. Appl. Phys. 80 (1996) 440, http://dx.doi.org/10.1063/1.362745. [40] B. Rebollo-Plata, R. Lozada-Morales, R. Palomino-Merino, J. A. DávilaPintLe, O. Portillo-Moreno, O. Zelaya-Angel and S. Jiménez- Sandoval, Amorphous carbon thin films prepared by electron-gun evaporation, aZo Journal Material Online, 2005. 〈〈http://www.azom.com/article.aspx? ArticleID ¼ 3097〉〉 (accessed May 23, 2014). [41] O. Malek, J. González-Julián, J. Vleugels, W. Vanderauwera, B. Lauwers, M. Belmonte, Carbon nanofillers for machining insulating ceramics, Mater. Today 14 (2011) 496–501, http://dx.doi.org/10.1016/S1369-7021(11)70214-0. [42] O. Malek, B. Lauwers, Y. Perez, P. De Baets, J. Vleugels, Processing of ultrafine ZrO2 toughened WC composites, J. Eur. Ceram. Soc. 29 (2009) 3371–3378, http://dx.doi.org/10.1016/j.jeurceramsoc.2009.07.013.

Paper V Surface microstructural changes of Spark Plasma Sintered zirconia after grinding and annealing

Authors:

Latifa Melk, Johanne Mouzon, Farid Akhtar, Miquel Turón-Viñas, Marta-Lena Antti and Marc Anglada.

To be submitted

115

116

Surface microstructural changes of Spark Plasma Sintered zirconia after grinding and annealing Latifa Melk a, b, c, e, Johanne Mouzond, Miquel Turona,b, Farid Akhtarc, Marta-Lena Anttic, Marc Angladaa,b a

b

CRnE, Campus Diagonal Sud, Edifici C’, Universitat Politècnica de Catalunya, 08028 Barcelona, Spain c

d

e

CIEFMA-Department of Materials Science and Metallurgical Engineering, ETSEIB, Universitat Politècnica de Catalunya, 08028 Barcelona, Spain

Department of Engineering Sciences and Mathematics, Luleå University of Technology, 97187 Luleå, Sweden

Department of Civil, Environmental and Natural Resources Engineering, Luleå University of Technology, 97187 Luleå, Sweden

Corresponding author at: Department of Materials Science and Engineering/ETSEIB Diagonal 647. 08028-Barcelona Tel. +34 934016701 Fax. +34 934016706 E-mail: [email protected]

Abstract Spark plasma sintered zirconia (3Y-TZP) specimens have been produced of 140 nm 372 nm and 753 nm grain sizes by sintering at 1250 C, 1450 C and 1600 C, respectively. The sintered zirconia specimens were grinded using a diamond grinding disc with an average diamond particle size of about 60 µm, under a pressure of 0.9 MPa. The influence of grinding and annealing on the grain size has been analysed. It was shown that thermal etching after a ruff grinding of specimens at 1100 C for one hour induced an irregular surface layer of about a few hundred nanometres in thickness of recrystallized nano-grains, independently of the initial grain size. However, if the ground specimens were exposed to higher temperature, e.g. annealing at 1575 °C for one hour, the nano-grain layer was not observed and the final grain size was similar to that achieved by the same heat treatments on carefully polished specimens. Therefore, by appropriate grinding and thermal etching treatments, nanograined surface layer can be obtained which increases the resistance to low temperature degradation.

Keywords: zirconia, grinding, annealing, low temperature degradation

1 Introduction Yttria stabilised tetragonal polycrystalline zirconia (3Y-TZP) has a wide range of applications, especially in the medical sector, because of its biocompatibility and very good mechanical properties, such as strength and toughness. The local and constrained phase transformation from tetragonal (t) to monoclinic (m) structure generates compressive stresses at the crack tip which enhances toughness. However, 3Y-TZP suffers from surface spontaneous t-m transformation in humid atmosphere, often referred to as hydrothermal degradation, aging or low temperature degradation (LTD), which is accompanied by formation of near surface microcracks and loss of surface mechanical properties [1–4]. During the processes of final shaping and surface finishing, 3Y-TZP may be subjected to different machining processes (cutting, polishing, grinding, and milling). The damage induced by machining affects structural integrity and reliability of the material. Therefore, machining zirconia is considered as a critical step in the manufacturing of long lasting and strong 3Y-TZP components. Previous investigations have shown that grinding influences the surface integrity and the flexural strength of 3Y-TZP materials [5]. Therefore, most of the studies on ground zirconia have focussed on characterizing surface microstructural changes that may affect the chemical and mechanical behaviour. The main changes frequently observed in the X-ray diffraction (XRD) spectrum are the following: (1) t–m phase transformation; (2) asymmetrical broadening of the (1 1 1) tetragonal peak at ~30° (2θ); (3) intensity reversal of the tetragonal doublet at 34.64° and 35.22° (2θ) corresponding to the (0 0 2) and (2 0 0) planes [6–10]. On the other hand, a TEM investigation of the ground surface by Munoz et al. [11] reported the existence of three different regions from the ground surface towards the bulk : (1) a recrystallized zone, exactly at the surface, where the grains have a diameter in the range 10–20 nm; (2) a plastically deformed zone; (3) a t-m transformed zone, which is mainly responsible for the formation of compressive residual stresses that usually increase the flexure strength and the apparent fracture toughness of ground specimens [11]. The near surface monoclinic phase formed during machining operations can be reversed to tetragonal by annealing. The operation of grinding and the time and temperature of annealing have an influence on the resistance to LTD, which can be inhibited or delayed [12]. The evolution of the resistance to LTD of specimens of initially 330 nm grain size subjected to grinding and annealing at 1200 °C for different times (1 min, 10 min and 1h) was analysed by Muñoz et al. 3

[13]. The results showed that LTD of ground 3Y-TZP was supressed due to the formation of a recrystallized nano-grain layer on the surface. Moreover, the resistance to LTD was decreasing during long time annealing at 1200 °C after that grain size reached grew beyond the initial surface grain size of 330 nm. The effect of different high annealing temperatures in the range 1200 °C - 1600 °C on the surface microstructure of ground zirconia was recently studied by Roa et al. [14]; this group also found the recrystallized surface nano-grain layer after annealing at 1200 °C by milling a small cross section of the near surface region by FIB/SEM, while at 1600 °C the near surface microstructure was composed of larger grain sizes than the grain size of the bulk material. To the best of our knowledge, previous studies of the effect of grinding on 3YTZP have been carried out only on conventionally sintered zirconia with grain size in the range of approximately 330 nm. The present work is focussed on the analysis of surface microstructural changes after grinding and annealing Spark Plasma Sintered (SPS) zirconia with different initial grain sizes of 140 nm, 370 nm and 750 nm achieved by sintering at 1250 °C, 1450 °C and 1600 °C, respectively.

2 Experimental 2.1 Material processing Zirconia powder stabilized with 3 mol % of yttria (TZ-3YSB-E, Tosoh, Tokyo, Japan) with a crystalline size of 36 nm was sintered using spark plasma sintering (SPS) at 1250 °C, 1450 °C and 1600 °C for 5 minutes. The pressure maintained during the sintering cycle was 55 MPa and the heating rate was 100 °C/min. The final samples were ceramic discs (50 mm x 3 mm). The average grain size was determined using the line intercept method on SEM images and the density was measured by the Archimedes method. The samples were ground using a new diamond grinding disc (MD-Piano 220 Struers) with an average particle size of about 60 μm, under a pressure of 0.9 MPa with a constant grinding speed of 3.6 m/s in one direction and water cooling. The selection of this grinding condition was based on a previous work of Juy et al. [15], who found that these particular parameters produce an increase in mechanical properties. The samples were then thermally etched at 1100 °C for 1 hour in standard furnace in air in order to observe grain size in the scanning electron microscope (SEM). 4

The samples studied will be referred to as SPS 1250, SPS 1450 and SPS 1600 according to the temperature used for sintering at 1250 °C, 1450 °C and 1600 °C, respectively. The samples were divided into two batches. The first batch corresponds to the as-ground samples which are referred in the current study as ASgr (AS ground). In order to reveal the grain size of the ASgr samples, they were maintained at 1100 °C for 1 hour for thermal etching, 1200 °C for annealing, and at 1575 °C for high temperature annealing. The second batch consists of specimens that were polished starting with 9 μm diamond down to 1 μm and finally polished with colloidal silica. The polished samples were subjected to the same heat treatment temperatures, and are herein referred to as ASpol (AS polished). In fact, each heat treatment on ground and polished specimens was carried out on the very same sample by polishing one face of the disc and grinding the other face in order to ensure that both faces were exactly subjected to the same temperature during annealing.

2.2 Mechanical testing Hardness was measured by Vickers indentation with a load of 98.1 N. The cracks emanating from the vertex of the residual impressions were used to measure indentation fracture toughness using Niihara equation [16] taking into account the observed Palmqvist configuration of the indentation cracks. Anstis et al. [17] equation was also applied for comparative purposes. However, as extensively reported in the literature [18], indentation fracture toughness does not really represent the "true" fracture toughness (KIc) of the material. Therefore, the fracture toughness measurements in the present work have been used only as an indication of t-m transformability under localised sharp contact of compressive loads.

2.3 Surface Analysis The crystallographic phases were identified by X-ray diffraction (XRD) with Bragg-Brentano symmetric-geometry, using PANalytical Empyrean equipment with PIXcel-3D detector and Cu-Kα (45kV and 40mA) radiation. The XRD spectra were obtained in a scan range of 20° ≤ 2θ ≤ 100°, using a step size of 0.013° and an anti-scatter slit of 1°. The monoclinic phase content was calculated by the equation proposed by Toraya et al. [19]. ̅

(

̅

)

5

)

) )

)

) )

Surface damage analysis of all the specimens was performed by extreme high resolution scanning electron microscopy (XHR-SEM) (Magellan 400, FEI Company) at an acceleration voltage of 3 kV. Microstructural changes below the surface induced during the grinding process were investigated by preparing thin cross-sections using focused ion beam (FIB). Cross sectioning observations were conducted using a dual beam workstation (Zeiss Neon 40). A thin platinum layer was deposited on the sample prior to FIB with the aim of reducing ion-beam damage. A Ga+ ion source was used to mill the surface at a voltage of 30 kV. Final polishing of the cross-section was performed at a current of 500 pA.

3 Results and discussion The present study shows clearly that the use of SPS results in dense zirconia samples of about 99 % of theoretical density, see Table 1. SPS is a highly efficient technique for the densification of zirconia ceramics compared with conventional sintering such as Hot Pressing (HP) for example. Self-heating from spark discharge between the particles could be the reason behind the low temperature and short time for sintering. It has been found that during SPS the residual gases in the powder will be efficiently removed as long as the system is open. As the pressure will not be applied until the isothermal temperature is reached, CO2(g) and H2O(g) can then escape before the densification starts [20]. Furthermore, by increasing the SPS temperature, the grain size increases from 140 nm to 750 nm for SPS 1250 and SPS 1600, respectively, see Table 2 and Fig. 1. It has been reported that the presence of an electric field could enhance the grain growth in yttria-stabilized cubic zirconia by increasing the grain boundary mobility [21]. Moreover, a dependency was shown between grain size and the heating-rate which promoted grain growth by increasing the defect concentration [22]. Fig. 2 shows the XRD spectra of ASgr in all SPS zirconia samples. It can be observed the presence of a broadening of the tetragonal peaks and that the tetragonal peak at 2θ≈30° that corresponds to (111)t has a speak shoulder that may correspond to either the rhombohedral phase or to distorted tetragonal phase as was reported in [14]. The presence of monoclinic phase at 2θ≈ 28° and 16° was also observed. It has been found that the monoclinic phase detected in ASgr is 13 %, 14 % and 12 % for SPS 1250, SPS 1450 and SPS 1600 respectively, see Fig. 3.

6

After annealing at 1200 °C and 1575 °C, the peak shoulder disappears and no monoclinic phase is detected due to the m-t transformation that actually starts at lower temperatures [23]. It could also be observed that the intensity of (002) and (200) tetragonal peaks at 2θ=34°, 64° and 35,22° is reversed compared to the )

ASpol. In ASpol, the intensity ratio of )

⁄

)

is equal to 0.45 while in

)

ASgr annealed at 1575 °C, ⁄ the ratio is equal to 1.63. This is attributed to the texture due stress induced reorientation from ferroelastic domain switching [7]. Indentation fracture toughness is similar to that for conventionally sintered 3YTZP with similar grain size, independently of the indentation equation used for calculating the fracture toughness. On the other hand, there is a small decrease in hardness with grain size which may be related to a higher transformability as the grain size increases, see Table 2. The specimens ASgr with subsequent thermal etching at 1100 °C for 1 hour show a nanometric grain size on the surface as can be seen on the images of left column of Fig. 3. The larger grain size observed in ASgr was always much smaller than the average grain size in ASpol. After thermal etching at 1100 °C, the grain size in the ASgr was reduced by a factor of 2 compared to the ASpol in SPS 1250 and by a factor of 10 in SPS 1575. However, if ASgr specimens are annealed at 1200 °C or 1545 °C, the surface grain size increases fast and it reaches dimensions roughly similar to as ASpol specimens subjected to the same high temperature treatment. This can be appreciated by comparing ASgr and ASpol specimens under the same heat treatment (see Fig. 3). The analysis of sections perpendicular to the surface obtained by FIB (Fig. 4) shows that there is a very thin layer of surface damage on ASgr specimens after grinding. The layer extends to depths of only a few hundred nanometers. The depth is not uniform and it changes from one place to another. It is deeper close to places where the material is piling up at the side of the grinding scratches. The same observation was carried out after annealing and they are shown in Fig. 5. Fig. 5 shows the presence of the nano-grain layer with a depth corresponding to the same depth of the nano-grain layer seen in the ground specimens before annealing. The existence of this surface nano-grain size layer on thermal etched ASgr specimens can also be detected on the fracture surface of specimens as shown in Fig. 6 where the fracture surface of ASpol and ASgr are compared. The presence of nano-grain layer in the etched ASgr specimens can be clearly observed. This 7

nano-grain layer is formed by recrystallization of a very thin highly deformed surface layer produced during grinding. Since the usual procedure to observe the grain size is by means of grinding with decreasing diamond particle size and careful polishing, the damage layer is finally removed so that no recrystallization is detected during standard specimen preparation and thermal etching for grain size determination. In one the earliest investigations on this topic, it was shown that this recrystallized nanometric layer can be useful for preventing hydrothermal degradation because the resistance to LTD increases as the zirconia grain size decreases [12]. The requirement of a smooth polished surface for many applications makes this procedure feasible when a specular smooth surface finish is not required. It may be of interest when a rough surface is beneficial as for example for implants since roughness favours osseointegration [24]. Regarding the influence of the surface damage induced by grinding on the nmnot affect the strength of ground specimens. strength, it was shown that 200 it does On the contrary, grinding induces the formation of a compressive surface layer which results in an increase of the strength [25,26]. However, after thermal etching and recrystallization, the compressive forces disappear as monoclinic phase is transformed back to tetragonal and then the strength may slightly decrease depending on the damage induced by grinding [11]. It is still unknown how much minimum plastic deformation by grinding is needed in order to form nanocrystals during etching. It will be interesting to find out which are the weakest grinding conditions for which recrystallization still takes place during thermal etching. Fig. 7 shows a shallow scratch left on a polished surface where recrystallization still takes place in and around the scratch. This shows that the recrystallization could still remain after polishing if a deep scratch is not fully removed by subsequent operations of grinding with smaller particle size followed by polishing.

4 Conclusions Spark plasma sintered 3Y_TZP specimens have been produced with different grain sizes of 140 nm, 372 nm and 753 nm by sintering at 1250 C, 1450 C and 1600 C. The influence of grinding and annealing has been analysed. Two main conclusions can be derived from the present work: a) the effect of the grinding conditions used in this study induces a few hundred nanometer surface layer which recrystallizes during thermal etching at 1100 C. The surface layer 8

contains recrystallized nano-grains with a grain size smaller and practically independent of the initial grain size depending on the SPS temperature. This behaviour is similar to that of conventionally sintered zirconia specimens; b) if the ground layer is exposed to higher annealing temperatures, the nano-grain layer disappears and the surface grains grow to a size which is similar to that achieved in polished specimens by heat treatment to the same temperature.

Acknowledgements The authors gratefully acknowledge the financial support given by the “Ministerio de Ciencia e Innovación”, Spain through research grant MAT201123913. The authors acknowledge the EU for financial support through the eCreate-Network of the Rise program. L. Melk acknowledges the fellowship award received from the European Joint Doctoral Programme in Materials Science and Engineering (DocMASE) of the European Union. Finally, all authors thank Dr. Trifon Trifonov and Dr. Joan Josep Roa from UPC for their assistance in the FIB/SEM equipment.

9

References [1]

K. Kobayashi, H. Kuwajima, T. Masaki, Phase change and mechanical properties of ZrO2-Y2O3 solid electrolyte after ageing, Solid State Ionics. 3 1981)489–493.

[2]

M. Yoshimura, T. Noma, K. Kawabata, S. Somiya, Role of H2O on the degradation process of Y-TZP, J. Mat. Sci Lett. 6 (2000) 465–467.

[3]

S. Lawson, Environmental Degradation of Zirconia Ceramics, J. Eur. Ceram. Sci. 15 (1995) 485–502.

[4]

J. Chevalier, L. Gremillard, A. V. Virkar, D.R. Clarke, The TetragonalMonoclinic Transformation in Zirconia: Lessons Learned and Future Trends, J. Am. Ceram. Soc. 92 (2009) 1901–1920.

[5]

T. Kosmač, Č. Oblak, P. Jevnikar, N. Funduk, L. Marion, Strength and reliability of surface treated Y-TZP dental ceramics, J. Biomed. Mater. Res. 53 (2000) 304–313.

[6]

H. Hasegawa, Rhombohedral phase produced in abraded surfaces of partially stabilized zirconia (PSZ), J. Mater. Sci. Lett. 2 (1983) 91–93.

[7]

A. V. Virkar, R.L.K. Matsumoto, Ferroelastic Domain Switching as a Toughening Mechanism in Tetragonal Zirconia, J. Am. Ceram. Soc. 69 (1986) C–224–C–226.

[8]

J. Kitano, Y. Mori, A. Ishitani, T. Masaki, A Study of Rhombohedral Phase in Y2 O3 -Partially Stabilized Zirconia, MRS Proc. 78 (1986) 17.

[9]

N. Mitra, K. Vijayan, B. Bai, S.K. Biswas, Phase Transformation Introduced by Mechanical and Chemical Surface Preparations of Tetragonal Zirconia Polycrystals, J. Am. Ceram. Soc. 76 (1993) 533–535.

[10] D.P. Burke, W.M. Rainforth, Intermediate rhombohedral (r-ZrO2) phase formation at the surface of sintered Y-TZP’s, J. Mater. Sci. Lett. 16 (1997) 883–885. [11] J.A. Munz-Tabares, E. Jiménez-Piqué, J. Reyes-Gasga, M. Anglada, Microstructural changes in ground 3Y-TZP and their effect on mechanical properties, Acta Mater. 59 (2011) 6670–6683. [12] P.J. Whalen, F. Reidinger, R.F. Antrim, Prevention of low-temperature surface transformation by surface recrystallization in yttria-doped tetragonal zirconia, J. Am. Ceram. Soc. 72 (1989) 319–321. [13] J.A. Muñoz-Tabares, M. Anglada, Hydrothermal degradation of ground 3Y-TZP, J. Eur. Ceram. Soc. 32 (2012) 325–333. [14] J.J. Roa, M. Turon-Vinas, M. Anglada, Surface grain size and texture after annealing ground zirconia, J. Eur. Ceram. Soc. 36 (2016) 1519–1525. 10

[15] A. Juy, M. Anglada, Surface phase transformation during grinding of YTZP, J. Am. Ceram. Soc. 90 (2007) 2618–2621. [16] K. Niihara, A fracture mechanics analysis of indentation-induced Palmqvist crack in ceramics, J. Mater. Sci. Lett. 2 (1983) 221–223. [17] G.R. Anstis, P. Chantikul, B.R. Lawn, D.B. Marshal, A critical evaluation of indentation techniques for measuring fracture toughness: I Direct crack measurements, J. Am. Ceram. Soc. 46 (1981) 533–538. [18] G.D. Quinn, R.C. Bradt, On the Vickers Indentation Fracture Toughness Test, J. Am. Ceram. Soc. 90 (2007) 673–680. [19] H. Toraya, M. Yoshimura, S. Somiya, Calibration Curve for Quantitative Analysis of the Monoclinic-Tetragonal ZrO2 System by X-Ray Diffraction, Commun. Am. Ceram. Soc. (1984) 119–121. [20] P. Dahl, I. Kaus, Z. Zhao, M. Johnsson, M. Nygren, K. Wiik, et al., Densification and properties of zirconia prepared by three different sintering techniques, Ceram. Int. 33 (2007) 1603–1610. [21] S.W. Kim, S.G. Kim, J. Il Jung, S.J.L. Kang, I.W. Chen, Enhanced grain boundary mobility in yttria-stabilized cubic zirconia under an electric current, J. Am. Ceram. Soc. 94 (2011) 4231–4238. [22] B.N. Kim, K. Hiraga, K. Morita, H. Yoshida, Effects of heating rate on microstructure and transparency of spark-plasma-sintered alumina, J. Eur. Ceram. Soc. 29 (2009) 323–327. [23] O. Fabrichnaya, F. Aldinger, Assessment of thermodynamic parameters in the system ZrO2-Y2O3-Al2O3, Zeitschrift Für Met. 95 (2004) 27–39. [24] Q. Flamant, F. García Marro, J.J. Roa Rovira, M. Anglada, Hydrofluoric acid etching of dental zirconia. Part 1: etching mechanism and surface characterization, J. Eur. Ceram. Soc. 36 (2016) 121–134. [25] A. Juy, L. Llanes, M. Anglada, Strength of Yttria-Stabilised Zirconia with Near-Surface Grinding Residual Stresses, ECF14, Cracow 2002. (2013). [26] T. K. Gupta, Strengthening by Surface Damage in Metastable Tetragonal Zirconia, J. Am. Ceram. Sci. 63 (1980) 117.

11

Tables Table 1. Density and grain size of the SPS zirconia specimens before and after annealing.

Specimen

Density (g.cm-3)

1100°C

Temperature/Grain size 1200°C 1200°C 1575°C

1100°C

1575°C

ASpol

ASgr

ASpol

ASgr

ASpol

ASgr

(nm)

(nm)

(nm)

(nm)

(nm)

(nm)

SPS1250

6.00 ±0.03

140±10

59±7

144±16

105±9

611±93

602±153

SPS1450

5.99± 0.09

372±50

66±15

386±29

149±22

699 ±79

896±167

SPS1600

6.05± 0.04

753±60

67±11

658±76

±

856±57

708±97

Table 2. Vickers hardness and indentation fracture toughness of the SPS zirconia samples. Vickers hardness

Indentation KIC

Indentation KIC

(HV10) (GPa)

(Niihara) (MPa·m1/2)

(Anstis) (MPa·m1/2)

SPS 1250

14.7±0.2

5.2±0.1

3.9±0.2

SPS 1450

13.8±0.1

5.1±0.1

3.8±0.1

SPS 1600

13.6±0.2

5.2±0.1

4.0±0.1

Specimen

12

Figures

Fig. 1 SEM images showing the microstructure of ASpol after thermal etching at 1100 °C for 1h: a) SPS 1250 b) SPS1450 and c) SPS1600.

13

Fig. 2 XRD spectra of SPS zirconia ASpol, ASgr, ASgr+annealed at 1200 °C and ASgr+annealed at 1575 °C.

14

Fig. 3 Grain size of ASgr SPS specimens after thermal etching for 1 hour at the temperatures indicated on the top row.

15

Fig. 4 Microstructure of ASgr below the surface a) SPS 1250; b) SPS1450 and c) SPS 1600 after grinding and before thermal etching.

16

Fig. 5 Microstructure of ASgr after thermal etching a) SPS 1250 top left; b) SPS1450 top right c) SPS 1600 bottom after grinding and thermal etching.

17

Fig. 6 Fracture surfaces of ASpol (top) and ASgr (bottom) SPS1250. General views on the left and high magnification details of the surfaces on the right.

18

Fig. 7 Surface of polished SPS 1450 after etching at 1100 °C with the initial grains and the recrystallized grains which appear in scratches still left on the surface after polishing.

19