Spectroscopic characterization and development of technique for high-pressure synthesis of carbon based nano-structural materials

Full text

(1)2007:184 CIV. EXAMENSARBETE. Spectroscopic Characterization and Development of Technique for High-Pressure Synthesis of Carbon Based Nano-Structural Materials. David Abrahamsson, Henrik Jonsson. Luleå tekniska universitet Civilingenjörsprogrammet Teknisk fysik Institutionen för Tillämpad fysik, maskin- och materialteknik Avdelningen för Fysik 2007:184 CIV - ISSN: 1402-1617 - ISRN: LTU-EX--07/184--SE.

(2) Spectroscopic characterization and development of technique for high-pressure synthesis of carbon based nano-structural materials. David Abrahamsson Henrik Jonsson. i.

(3) Abstract To be able to study high pressure properties and synthesis of carbon nanostructures, we built a pressure control box to regulate and fine tune pressure to our membrane diamond anvil cell (MDAC). Using ruby fluorescence method, we calibrate the pressure in the membrane against the generated pressure in the cell. This resulted in both a pre-indentation pressure curve for hardened stainless steel gaskets and a calibration curve for pressure in the membrane against generated pressure in the MDAC. Raman spectroscopy was used to characterize carbon nanostructures. Two kinds of carbon nanotubes (CNTs) were examined, one HiPCO and one Arc-discharged produced, with the purpose to see how side-wall functionalization affects their spectra. Spectra were acquired from pristine and functionalized samples with 532 nm (2,33 eV) and 632,8 nm (1,96 eV) excitations and compared. Both metallic and semiconducting tubes are being probed, and we see that metallic and small diameter semiconducting tubes are more affected compared to semiconducting tubes with larger diameters. Different types of polymeric fullerene samples were also characterized in order to determine their structure. Spectra from 1D and one kind of 2D polymeric fullerene samples were found. Multi wall CNT (MWNT) - epoxy composites, manufactured by SiCOMP, were examined with Raman spectroscopy to see how the MWNT interacted with the epoxy matrix and to get an estimation of how well dispersed the MWNTs were in the matrix. Spectra were acquired from epoxy and MWNTs separately, and then compared with spectra acquired from the composites. We see no sign of interaction between the epoxy and the MWNTs, and they do not seam to be that well dispersed.. ii.

(4) Contents 1 Thesis introduction ............................................................................................................4 1.1 Background ..................................................................................................................4 1.2 Motivation ....................................................................................................................4 1.3 Thesis Outline...............................................................................................................5 2 Theory ................................................................................................................................5 2.1 High pressure methods in materials synthesis................................................................5 2.1.1 Diamond anvil cell (DAC) .....................................................................................5 2.1.2 The gasket..............................................................................................................6 2.1.3 Pre-indentation of gasket........................................................................................6 2.2 Physical properties of carbon-based nanostructured materials .......................................8 2.2.1 Fullerenes ..............................................................................................................8 2.2.2 Carbon nanotubes...................................................................................................9 2.2.3 CNT-based composite materials ...........................................................................13 2.3 Synthesis and Characterization of carbon-based nanostructured materials ...................15 2.3.1 Polymerization of fullerenes at high pressure .......................................................16 2.3.2 Functionalization of carbon nanotubes .................................................................17 2.3.3. Vibrational properties of carbon nanostructures from Raman spectroscopy .........18 3 Experimental methods.....................................................................................................21 3.1 High-pressure method .................................................................................................21 3.1.1 Membrane DAC...................................................................................................21 3.1.2 Pressure control box.............................................................................................22 3.2 Sample preparation .....................................................................................................24 3.2.1 Preparation of sample holders and sample loading in a DAC................................24 3.2.2 Carbon nano materials..........................................................................................25 3.3 Spectroscopic characterization ....................................................................................27 3.3.1 CRM-200.............................................................................................................27 3.3.2 Ruby fluorescence................................................................................................27 3.3.3 CNT and fullerenes ..............................................................................................28 4 Results and Discussion.....................................................................................................29 4.1 Ruby fluorescence: calibration of DAC gas loading system.........................................29 4.1.1 Results from calibration experiment .....................................................................29 4.2 Spectroscopic characterization of CNTs......................................................................34 4.2.1 Pristine material ...................................................................................................34 4.2.2 Functionalized material ........................................................................................38 4.2.3 Composite material based on MWCNT (SiComp samples)...................................44 4.3 Raman study of polymerized fullerenes.......................................................................46 4.3.1 1D C60 polymers .................................................................................................46 4.3.2 2D C60 polymers .................................................................................................47 5 Summary..........................................................................................................................49 5.1 Summary ....................................................................................................................49 5.2 Conclusions ................................................................................................................49 5.3 Recommendations for future work ..............................................................................50 Extra................................................................................................................................50 Appendix A ..................................................................................................................55 Appendix B ..................................................................................................................58. iii.

(5) 1 Thesis introduction 1.1 Background The research on carbon based nanostructures has escalated in the last years. This is much due to the discovery of the single walled carbon nanotube (SWNT) by Iijima 1993, and the remarkable properties that the nanotubes have shown both in theory and in practice. The tubes can have either semiconducting or metallic electronic band structure, and they have outstanding mechanical properties. It has however shown to be difficult to construct materials using only nanotubes, due to their week van der Waal interactions. Work is being done trying to make nanotube composites, and by that transfer some of the remarkable properties of the SWNT to other materials. Today most of the research in on nanotube composites is done using polymeric matrixes because they are easier to handle than metals. Tests have shown that it is not possible just to mix raw nanotubes in the matrix and hope to get good results. The nanotubes interact badly with the matrix and have a tendency to agglomerate. To get the tubes to connect better with the polymer matrix, chemical groups that easier interact with the polymers are being attached to the nanotube sidewalls (functionalization). Fullerenes are another carbon nano material that has shown interesting properties. The fullerenes were discovered in 1985 by researchers at Rice University and named after Richard Buckminster Fuller and are sometimes called buckyballs. It has, among other things, shown possible to construct intrinsic materials from fullerenes with interesting properties. Of all methods used to characterize carbon nanostructures, Raman spectroscopy is the most powerful. The one dimensional structure of the CNTs create pikes in the electronic density of states, which gives rise to a resonance phenomenon that occurs when the tubes are lit up on with matching excitation energies, and results in a strong Raman signal. To investigate physical properties of carbon nanostructures, high pressure experiments have shown of interest. As an example, it has shown possible to polymerize fullerenes using combinations of high pressure and temperature.. 1.2 Motivation The projects were developed by Professor Alexander Soldatov at Luleå University of Technology. These projects needed to be started, and they are a part of a program aimed at development of new methods to produce novel nanostructured functional materials formed by polymerization of both fullerenes and nanotubes at high pressures and basic research on the physical and chemical properties of the products. An essential piece of the work needed to be dedicated to setting up a new High-pressure spectroscopy laboratory in the department of physics. Professor Soldatov are having a research cooperation with a research team at Nancy University in France, and as the laboratory on LTU has new state of the art equipment, the lab crew on LTU can provide nice complimentary work to the research at Nancy University.. 4.

(6) 1.3 Thesis Outline This thesis begins with a short introduction which is followed by brief theory, experimental methods, results and discussion, and finally we round up with a summary. 2 Theory 2.1 High pressure methods in materials synthesis High-pressure is an important parameter to probe physical properties and synthesis of new materials. To reach these pressures different techniques can be used. The major effects of high pressure on matter include decrease of volume, phase transitions, changes in electrical, optical, magnetic, and chemical properties, and increases in viscosity of liquids. In general solids are less compressible than liquids, and the compressibility of both solids and liquids decreases with increasing pressure [1]. There are two main techniques that are used in high pressure research, static and shock. Both techniques are capable of generating pressures in the megabar range, but in particular, static pressure yields continuous states on an isotherm (or isochore, when heating), with a slow loading rate, and sock compression can load samples to megabar pressures in a fraction of a nanosecond to microseconds. An example of a device used to generate high static pressure is a diamond anvil cell (see 2.1.1).. 2.1.1 Diamond anvil cell (DAC) A diamond anvil cell works much like a vice, squeezing the sample between two anvils, fig 1. To generate as high a pressure as possible, the anvils have to be made from the hardest material possible, diamonds. Another advantage in using diamonds as anvils is that they are transparent to a wide spectral range (visible, etc.), which makes it easier doing measurements on the sample. The sample is held at its place using a metal gasket as sample holder. The pressure that can be generated in a diamond anvil cell is limited to the size of the diamond through the force-area relation, and most of all to the strength of diamond as a material. One disadvantage of diamond anvil cells is that only small samples can be used. Another disadvantage in an ordinary dac is that the pressure is generated by manually tightening screws, making it hard to take small pressure steps. Pressures up to 3 Mbar can be reached with this method.. 5.

(7) Fig 1, Schematic diagram of a diamond anvil cell (DAC). (A) Support, (B) screws, (C) diamond anvils, (D) gasket, (E) sample chamber. The principle of DAC operation: a) Diamonds (small top surface = culet) b) Gasket – sample chamber c) Pressure generation force application ( ~100kg - 300kg). 2.1.2 The gasket The gasket is an important part of a high pressure experiment. The material and structure of the gasket may be freely chosen and might vary from one high-pressure experiment to another. A common choice of material that is used both in the range up to 10GPa is a hardrolled austenitic stainless steel, (in the multi-mega bar range Rhenium is used). Materials that show little or none work-hardening, such as mild steel, should be avoided. The gasket is preferable thicker then you want to use in your experiment. This means that you can produce the exact thickness you need for your experiment by doing a pre-indent and you only have to have one kind of thickness of your gaskets in stock [2].. 2.1.3 Pre-indentation of gasket Pre-indentation has many advantages. Two of them are already mentioned in 2.1.2, but it also makes the loading process easier. If you don’t have a pre-indent, you face the problem of centering the gasket hole on the culets and loading the sample at the same time. With a preindent the centering of the hole is already taken care of due to that the position of the hole is already registered and well centred in the indent. It also makes it easier to choose the correct thickness of the gasket as you can follow the hole during your experiments, and depending on what happens make the gasket thinner or thicker the next time. The part of the gasket under the diamonds undergoes plastic deformation (and is extruded) as the anvils advance with increasing pressure. In the material, the hydrostatic pressure decreases linearly in the direction of the extrusion. The gradient of the pressure increases as the gasket becomes thinner. The extrusion of the material can go either outwards (extending the hole), inwards (collapsing the hole), or both outwards and inwards. If the extrusion is booth inwards and outwards there will be a segment of metal that does not move at all and separates the two moving segments. The choice of gasket thickness and the diameter of the hole depend on the pressure medium used. In the case of using the relatively incompressible alcohol or argon as media, the hole should be one-third to two-fifths of the culet diameter, since the hole is expected to shrink just 6.

(8) slightly during the experiment. If for example helium is used, which is much more compressible, the gasket hole is expected to shrink at about a factor of 2 or more. In this case the hole should be made larger. The primary symptom of gasket failure is if the hole starts to grow [3].. 7.

(9) 2.2 Physical properties of carbon-based nanostructured materials Carbon is the only element in the periodic table that has isomers from 0 dimensions (0D) to three dimensions (3D), (the 0D C60 fullerene cluster, the 1D nanotube, the 2D graphite and the 3D diamond), see fig 2.. Fig 2, Isomers of carbon. Carbon is the sixth element in the periodic table and is listed in top of column 4. Each carbon atom has six electrons which occupy 1s2, 2s2, and 2p2 atomic orbitals. The ground state of carbon is 3P (S=1, L=1). The 1s2 orbital contains two strongly bound core electrons and the 2s22p2 orbitals contain four more weakly bound valence electrons. In the crystalline phase, the valence electrons give rise to 2s, 2px, 2py, and 2pz orbitals which are important in forming covalent bonds in carbon materials. The energy difference between the upper 2p energy levels and the lower 2s level is small compared with the binding energy of the chemical bonds, and therefore the electronic wave functions for these four electrons will easily mix with each other and thereby changing the occupation of the orbitals so that the binding energy of the carbon atom and its neighbours will get enhanced. This mixing of orbitals is called hybridization [4].. 2.2.1 Fullerenes The C60 fullerene was discovered 1985 by H.Kroto, R.Curl, and R.Smalley [5], and they were rewarded in 1996 with the Nobel price in chemistry. They use a technique that involves vaporization of graphite by laser irradiation. The C60 molecule consists of 60 carbon atoms and has the shape of a regular truncated icosahedron belonging to the point group Ih, fig 3a. A fullerene is a closed cage molecule containing only hexagonal and pentagonal faces. By the ‘isolated pentagon rule’ all fullerenes must have 12 pentagonal faces and an arbitrary number of hexagonal faces. The smallest fullerene to satisfy this rule is the C60 and the second smallest is the C70. Each carbon atom is trigonally bonded to three other carbon atoms and there are two different types of bonds in the molecule. Two single C-C bonds are located along a pentagonal edge at the fusion of a hexagon and a pentagon and a third bond is located at the fusion between two. 8.

(10) hexagons and is a double bond. The bond lengths for the single bonds are 1,46 Å and for the double bond is 1,40 Å [6]. Fig 3, (a) C60 (b) C70 (from [7]). The rugby ball shape of C70 can be envisioned by adding a ring of 10 carbon atoms around the equatorial plane of the C60 molecule and rotating the two hemispheres of C60 by 36°, fig 3b. The C70 molecule has five in equivalent sites and eight distinct bond lengths. Solid C60 forms two crystal phases at zero pressure. One face centred cubic (fcc) phase above the transition temperature T01 ≈ 261 K and a simple cubic (sc) phase below. At high pressuretemperature condition several different polymeric phases are formed. In fig 4a an orthorhombic one-dimensional crystal structure is shown and fig 4b and fig 4c shows two different two-dimensional structures, the orthorhombic and rhombohedral.. Fig 4. Different polymeric phases of fullerene. a) orthorhombic, b) tetragonal, c) rhombohedral (from [8]). Blank et al. has under high temperature and pressure obtained ultra hard fullerite with a hardness of at least 170 GPa, exceeding that of diamond [9]. Some of the promising fullerene applications suggested are a starting material for superhard materials and diamond, precursors for CVD diamond films and SiC, lithographic films, optical limiters, solar cells, lubricants, catalysts, fullerene-containing polymers, and medicines [10].. 2.2.2 Carbon nanotubes Research on vapour grown carbon fibres led to the discovery of very small diameter filaments. At the time no detailed studies were done, but later when the fullerenes were discovered speculations begun about whether there could be existing carbon nanotubes (CNT) of same dimensions as the C60 molecule. The first observations of nanotubes were reported by 9.

(11) Iijima 1991, and were done by transition electron microscopy [11]. The single-walled carbon nanotubes (SWNT) were discovered in 1993 [12]. The CNT has a small enough diameter compared to its length to be a one dimensional object. A single-wall carbon nanotube can be described as a 2D graphene sheets rolled up into a seamless cylinder in which sp2 hybridization is present. Multi-walled carbon nanotubes (MWCNT) are built up from many single-wall nanotubes arranged layer by layer inside each other. The distance between the layers are approximately equal to the graphite inter-layer distance. The diameter range of the SWNT is about 0.7-10.0 nm, though most of them have a diameter less than 2 nm [4]. The system that is used to classify the different single wall nanotubes are based on the orientation of the hexagons that the carbon atoms are arranged into, see fig 5. The primary symmetry classification of the nanotubes is to see if the tube is chiral or achiral. The achiral class is then divided into two groups, armchair, zigzag, making it three sub-classes of single walled nanotubes. A nanotube is achiral if its mirror image has an identical structure as the original, and chiral if it is not. The chirality is described by the chirality vector, C h , and it can be expressed by the real space unit vectors aˆ1 and aˆ 2 of the hexagonal lattice, see equation 1 below and fig 5. C h = naˆ1 + maˆ 2 ≡ (n, m) (n,m are integers, 0 ≤ m ≤ n ). (1). The chirality angle θ denotes the angle in which the hexagons are tilting with respect to the direction of the nanotube axis. Because of the hexagonal symmetry θ can wary between 0° and 30° and is given by . 3m    2n + m . θ = tan −1 . (2). The diameter, dt, of the CNT is given by dt =. Ch. π. =. 3a CC n 2 + nm + m 2. π. (3). where aCC is the nearest-neighbour C–C distance (1.421 Å in graphite). 10.

(12) Fig. 5, The atomic configuration of a single layer of graphite known as a graphene sheet. The vectors are used when describing the formation of SWNT’s by rolling the graphene sheet into a seamless cylinder. C is the chiral vector, θ the chiral angle and a1, a2 the unit vectors of the graphene. T describes the smallest translation along the tube axis. In the case shown C represents a (5, 2) tube.. In table 1 chiral index (n,m) and chiral angle (θ) of the different tube types are presented. Table 1. Chiral index and angle for different types of tubes Tube type chiral index (n,m) chiral angle (θ θ) Armchair n,n 30o o Zigzag 0,n 0 Chiral n,m 0<θ<30o. The unit cell for a carbon nanotube in real space is given by the rectangle generated by the chiral vector Ch and the translational vector T, fig 5. There are 2N carbon atoms per unit cell (where N is the number of hexagons in the unit cell), which means there will bee N pairs of bonding π and anti-bonding π* electronic energy bands. Similarly the phonon dispersion relations will consist of 6N branches, resulting from a vector displacement of each carbon atom in the unit cell. The phonon dispersion relations of CNTs can be understood by zone folding the phonon dispersion curves for a single 2D graphene sheet [4]. Phonons denote the quantized normal mode vibrations that strongly affect many processes in condensed matter systems, including thermal, transport and mechanical properties.. The electronic structure of a single wall carbon nanotube can to the first order be obtained from 2D graphite, but the quantum confinement of the 1D electronic states must be taken into account. The σ bands make up the covalent bonds within the 2D graphene sheets, while the π bands make up the weaker van der Waals interactions from one graphene sheet to another as in 3D graphite. The π bands are close to the Fermi level, giving the possibility for electrons to be exited from the valence band to the conduction band optically [13]. The electronic structure of most SWNTs can be obtained by using a nearest neighbour tight binding model [4], however if the diameter of the tubes is small, the curvature of the graphene sheet induces changes in the bound distance between the carbon atoms and causes mixing of the σ and π bounds. Thus more accurate methods must be used [14]. The 1D nature of the nanotube results in several spikes (van Hove singularities) in the electronic density of states (EDOS). The sub band peak positions of the EDOS are unique for 11.

(13) every diameter and chirality of the carbon nanotubes [15]. Plotting the energies between the van Hove singularities, fig 6, against the diameter of the tubes will result in a so called Kataura plot fig 7. From the Kataura plot it can be determined which tubes that are possible to detect with a specific excitation source. When the laser excitation energy matches the energies of the van Hove singularities, the tubes are in resonance. The SWNTs can show either metallic or semiconducting behaviour. Approximately one third of all nanotubes are metallic and two thirds are semiconducting. A general rule to determine if a tube is metallic or semiconducting is to look at its chiral index. If (2n + m), or equivalently (n - m), is a multiple of 3, then the tube is metallic and if not, the tube is semiconducting. The armchair nanotube, who has a chiral index of (n , n), is always metallic [4].. Fig 7. Kataura plot: transition energies of semiconducting (filled symbols) and metallic (open) nanotubes as a function of tube diameter. Calculated from the van-Hove singularities in the joint density of states within the third-order tight-binding approximation (Tomsen & Reich). Fig 6. Band structure and density of states of a semiconducting zigzag nanotube. The Fermi level is set to zero. (Thomsen & Reich). Carbon nanotubes exhibit outstanding physical properties: Single walled carbon nanotubes are predicted to show ballistic conductive behaviour with a resistance as low as 6454 Ω/tube, independent of the tube length. This would correspond to a bulk resistivity several times lower than copper [16]. Individual nanotubes has shown resistivities in the range of 5.1*10-6–5.8Ωcm. The carbon-carbon chemical bound in a graphene layer is probably the strongest chemical bound known in nature. Depending on the quality of the sample, the Young modulus of SWNTs has been measured to be over 1 TPa, which is larger that that of any other known material [17]. The thermal conductivity can be as high as 3500 Wm-1K-1 at room temperature for a single CNT, and they also have a very tiny expansion coefficient due to the strong in-plane C–C bonds [18]. For comparison, copper and silver, which have the highest electrical conductivities of any metal, have thermal conductivities of just around 400 Wm-1K−1 at room temperature. [19] 12.

(14) Due to their unique properties, the CNT have great prospective for applications in various disciplines. Application areas cover very widely for field emitters, hydrogen storage, transistor, secondary battery, supercapacitor, fuel cell, gas sensor, composites, and nanoprobes [15]. The MWCNT does not show quite as good physical properties as the SWCNT, but they are much cheaper to produce which makes them suitable for design of inexpensive CNT-based materials, for example composites. The smallest in the family multi-wall nanotubes is the double-wall nanotube which is built up from two single wall nanotubes. The walls of the nanotubes are rarely perfect. Defects in form of pentagons, heptagons or even vacancies can easily be created instead of hexagons. Local and global curvature, inter-tube and inter-shell interactions are dramatically modifying the electronic properties from those obtained from simple zone folding [4] of the graphene band structure. These imperfections give rise to the disorder induced D-band in Raman spectroscopy, se sec 2.3.3. However the possibility to control these defects will probably be important in future tailoring of carbon nanotube properties. Single-walled carbon nanotubes are rarely found as isolated specimens. They often assemble in bundles to minimize their energy through van der Waals interactions. The diameters of the nanotubes in a bundle are rather homogenous, but they often have a wide distribution of chiral angels. The present technologies for CNT synthesis always produce samples with mixing chiralities.. 2.2.3 CNT-based composite materials The electrical, mechanical and thermal properties of an individual single wall carbon nanotube make them very attractive in materials design. SWNTs interact with each other only through week van der Waal interactions, which makes a strong material that is built up only from the nanotubes themselves difficult to create. Many research groups are trying today to make composites containing nanotubes, and there by enhance the properties of the matrix material. As an example, estimations are done that show that in a copper-nanotube composite, the resistivity could be as low as 50% lower then in copper [16]. Research on CNT-reinforced nanocomposits is showing exiting and promising results. Polymers, ceramics and metals are being tested as matrix material, but the most studies has been done on the CNT-polymer composites due to their relatively convenient processibility. Both MWCNTs and SWCNTs have been used, but SWCNTs are usually favoured over MWCNTs as structural reinforcements due to their physical properties. Nevertheless the CNT composites still lag quite far behind in comparison to single (individual) nanotubes in terms of physical properties. The main problems in the making of nanotube composites are how to evenly disperse the tubes in the matrix and to get the nanotubes to interact with the matrix. Some remarkable results from experiments on CNT-nanocomposites:. 13.

(15) •. 80% improvement in the tensile modulus for polyvinyl alcohol (PVA) by adding 1 wt% MWCNTs [20]. •. an increase of about 140% in ductility for ultrahigh molecular weight polyethylene adding 1 wt% of MWCNTs [21]. •. By dispersing 2 wt% MWCNTs in Nylon-6, Liu et al showed that the elastic modulus and the yield strength of the composite were increased by 214% and 162%, respectively [22]. •. a three fold increase in youngs modulus is achieved adding 1 wt% SWCNT to polypropylene [23]. 14.

(16) 2.3 Synthesis and Characterization of carbon-based nanostructured materials When fullerenes were discovered a laser ablation method was used for their production. This method generated only small quantities of fullerenes but the discovery of an arc discharge method in 1990 made it possible to produce larger quantities [7]. These methods produce carbon “soot” from which the fullerenes are extracted by use of appropriate solvents [24]. The techniques for production of CNTs can be divided into three main categories: • Laser ablation • Arc-discharge • Chemical Vapour Decomposition (CVD) In all these synthesis methods, metals are used as process catalysts. Often used metals are iron, cobalt, nickel and yttrium. The metal catalysts favour the growth of SWNTs. In the case of the arc-discharge method, the metals are mixed with solid pure carbon electrodes. In the laser ablation process, the metal is placed on the substrate forming nanoclusters. In the CVD method the metal forms gaseous clusters acting as nuclei for the SWNT synthesis. One of the most commonly used CVD processes is the high-pressure CO conversion (HiPCO) process where SWNTs with up to 97% purity are produced by flowing CO and Fe(CO)5 under high pressure (30–50 atm). Nanotubes with diameters as low as 0,7 nm can be created with this process, which is the lowest diameter expected for SWNTs. The average diameter produced by this method is 1,1 nm. [25] There are numerous techniques used to characterize carbon based nanostructures: •. X-ray photoelectron spectroscopy (XPS) (allows determining the functionalization of the nanotubes).. •. Scanning tunnelling microscopy (STM) is a powerful technique used to obtain three dimensional images and electronic structure of the nanotubes. The helicities of the nanotubes can be determined with high resolution images.. •. Neutron diffraction is widely used for determination of structural features such as bond length and possible distortion of hexagonal network.. •. X-ray diffraction (XRD) is used to obtain some information on the interlayer spacing, the structural strain and the impurities.. •. Transmission electron microscopy (TEM) is used to investigate structural details such as intershell spacing, chiral indices, and helicity.. •. Infrared spectroscopy is often used to determine impurities remaining from synthesis or molecules capped on the nanotube surface.. 15.

(17) •. Photoluminescence technique can be used to determine geometries and diameters of CNTs. In order to observe the photoluminescence phenomenon, the bundles must be separated into individual tubes.. •. Raman spectroscopy is the most powerful technique to characterize carbon nanotubes. Without sample preparation, a fast and non-destructive analysis is possible. Moreover, Raman spectra simulations are performed on various nanotubes geometries.. For a correct characterization of nanotubes, all these techniques described here can not be used separately but must be used in complementary ways [26].. 2.3.1 Polymerization of fullerenes at high pressure C60 can be polymerized by submitting the material either to ultraviolet or visible light, photo polymerisation, or to high pressures and high temperatures. Photo polymerisation can only be carried out on very thin films. In high pressure- high temperature polymerisation larger volume samples of polymerized material can be produced. In the polymeric phases, the weak van der Waals binding between molecules is replaced by covalent bonds via a 2 + 2 cycloaddition reaction [27]. In this reaction the double bonds forming borders between hexagons open up to form the intermolecular bonds, fig 4 in section 2.2.1.. Fig 8. Pressure-temperature phase diagram of C60 (from [33]). Depending on the exact temperature-pressure conditions, one-dimensional (1D), twodimensional (2D) and tree-dimensional (3D) polymerisation have been found [28]. In the 1D phase the C60 molecules forms linear chains in an orthorhombic crystal structure. As an initial intermediate in the formation of the linear chains the formation of the dimer C120 and higher oligomers has been observed. There are two different structures in the 2D phase, tetragonal and rhombohedral see fig 8. In the 2D polymers the C60 molecules are linked via covalent bonds in the planes while interaction between the planes is van der Waals.. 16.

(18) 2.3.2 Functionalization of carbon nanotubes As already mentioned, carbon nanotubes exhibit extraordinary properties, and in order to transfer these properties in to composite materials functionalization has been proposed as a first step. The main reasons to use functionalized tubes are that they have shown to be easier to disperse, and they interact better within a composite matrix. Other areas in which functionalized nanotubes can be useful are in electronic applications and sensing applications [29]. There are three main strategies that can be used when functionalizing nanotubes. One strategy is to functionalize from created defects, another to functionalize the end caps of carbon nanotubes, and one is to functionalize the sidewalls of the tube, see fig 9 [30]. Of these strategies, the most interesting in order to get the nanotubes to interact with composite matrixes is the sidewall functionalization due to the increased solubility of the SWNTs, which allows for better manipulation and processing, and increased solubility leads to better dispersion in polymeric systems. There are two categories of side wall functionalization technique, depending on the type of bond binding the functional group to the nanotube. The two possible cases are either a covalent bond or a van der Waals bond. One big advantage of covalent functionalization is that no surfactant that can affect the properties of the composite has to be used, since such additives are difficult to remove and thus may have negative effects on the properties of the composites [31]. Covalent sidewall functionalization can even be performed selectively (i.e. metallic nanotube can be modified without affecting the semiconducting tubes) [32]. Another interesting property of functionalization is that the surface tension of the tubes can be modified. The pristine nanotubes are hydrophobic in nature due to their van der Waals interaction. Functionalization leads to transform the hydrophobic surface to hydrophilic one by ionizing the CNT surface [15]. (One of the most common van der Waal functionalizations is by use of Sodium dodecyl sulphate (SDS) in attempts to disperse nanotubes.). Fig. 9, Example on sidewall covalent functionalization. To analyse the functionalized nanotubes, the following techniques are the most commonly used: Absorption and resonance Raman spectroscopy are employed to ensure that the functionalization is covalent and occurs at the side-walls. Once covalent side-wall functionalization is confirmed, thermal gravimetric analysis (TGA) and x-ray photoelectron spectroscopy (XPS) are used to determine the degree of functionalization. (Imaging techniques such as atomic force microscopy (AFM) scanning electron microscope (SEM), and transmission electron microscopy (TEM) are also used as a compliment.) [30]. 17.

(19) 2.3.3. Vibrational properties of carbon nanostructures from Raman spectroscopy Fullerenes The vibrational modes of C60 fullerene can be subdivided into two classes: intermolecular vibrations (or lattice modes) and intramolecular vibrations (or simply ‘molecular’ modes). C60 intramolecular modes. The first-order Raman and infrared spectra of the C60 molecule are among the simplest compared to other fullerene molecules. This is because of the high symmetry of the molecule. Starting with 180 total degrees of freedom and subtracting tree corresponding to translation and tree corresponding to rotations, yields 174 vibrational degrees of freedom for an isolated C60 molecule. From group theory the 174 vibrational degrees of freedom corresponds to 46 distinct intramolecular modes. Of the 46 modes, ten are Raman-active (2Ag + 8Hg) in first order, four are infrared (IR)-active (4F1u) and the remaining are optically silent. The two Ag modes are nondegenerate and the eight Hg are fivefold degenerate. In the Ag(1) ‘breathing’ mode (492 cm-1) all the 60 carbon atoms are involved in identical radial displacement. The Ag(2) mode (1468 cm-1) is called the ‘pentagonal pinch mode’ and it involves tangential displacement with a contraction or expansion of the pentagonal rings. The eight Hg modes are more complex and span the wavenumber range from 269 Hg(1) to 1575 cm-1 Hg(8) [7]. There is a tendency of the lower frequency molecular modes (ω < 700 cm-1) to have displacements that are more radial in character and the high-wavenumber (ω > 700 cm-1) modes tend to exhibit approximately tangential displacement. The C60 solid has vibrational frequencies similar to those for a free molecule because of the very weak van der Waals bonding between the molecules. The expected number of second-order Raman lines are consider to be very large because it include both overtones and combination modes of the first order modes. From group theoretical considerations we get a total of 151 modes with Ag symmetry and 661 modes with Hg symmetry. The study of the second-order Raman spectra of C60 is a powerful technique for the determination of the frequencies of the 32 silent first order modes [7].. C60 intermolecular modes. In the solid phase of C60 there is an orientational ordering temperature at T01 ≈ 261 K. At temperatures well below the orientional ordering temperature there are five additional modes that are Raman-active, but observation by Raman spectroscopy is difficult because of their low wavenumbers ( < 60 cm-1). Above the orientational ordering temperature the C60 molecules rotate rapidly about their equilibrium positions in the fcc lattice and thus no rotational intermolecular modes exist.. 18.

(20) Polymerisation of C60. After polymerisation of C60 most of the Hg modes split into several components and new modes appear which were optically silent earlier. This is due to change in symmetry. A useful method to identify the polymeric structure of C60 by Raman spectroscopy is to follow the evolution of the pentagonal pinch mode. This is because the number of double bonds forming borders between hexagons will decrease as the bonds break up to form the intermolecular bonds which leads to lower intramolecular average bond stiffness. The pentagonal pinch mode has been found to shift to 1464 cm-1 for dimers, to 1459 cm-1 for linear chains, and to 1447 cm-1 for tetragonal polymers [33]. For the rhombohedral it shifts to 1410 cm-1[34]. Other values of the shift of the pp mode were found to be 1462 cm-1 for dimers, 1457 cm-1 for linear chains, 1449 cm-1 for tetragonal and 1406 cm-1 for rhombohedral polymers [35]. Another sign of polymerisation is the appearance new peaks in the intermolecular vibration region between 50 and 200 cm-1. In ref [36] three well-resolved Raman peaks at 97, 119 and 173 cm-1 was found and in ref [37] four peaks was found at 98, 121, 157, and 174 cm-1.. Single-wall carbon nanotubes First-order Raman scattering in the RBM and G-band of SWNTs. In the Raman spectra of SWNTs the strongest features of the first-order scattering processes are the radial breathing mode RBM and the tangential G-band, see fig 20 and 22 in section 4.2.1. The RBM correspond to vibration of the carbon atoms in the radial direction, like the tube were breathing. Typically it is observed in a frequency range of about 100-500 cm-1 and its appearance is a strong evidence of SWNT presence in the sample [13]. In CNT samples containing isolated SWNT it is possible to find only one RBM. In samples containing bundles, all the tubes in resonance will be contributing to the spectra. The RBM frequency (ωRBM) is inversely proportional to the tube diameter and for characterization of the diameter (dt) distribution in a sample the following expression can be used:. ω RBM =. C1 + C2 dt. (4). where C1 and C2 are parameters to be determined experimentally. Their dependence comes from the fact that the mass of all the carbon atoms along the circumferential direction is proportional to the diameter. Several values for the parameters have been proposed depending on CNT environment and tube-tube interactions. For SWNT in bundles values of C1=224 cm-1 and C2=14 [38] and for isolated SWNTs on a SiO2 substrate C1=248 and C2=0 [39]. For small diameter tubes dt < 1 nm the simple relation in equation 4 is not expected to hold exactly, due to nanotubes lattice distortions. In order to get a good characterization of the diameter distribution several laser excitation energies are needed. This is due to the different resonance energies of nanotubes with different diameters. From RBM measurements using several laser energies the ratio of metallic to semiconducting tubes in the sample can be estimated.. 19.

(21) In graphite the peak at 1582 cm-1 is called the G-band, and is related to the tangential mode vibrations of the carbon atoms in a graphite plane. The corresponding mode in SWNTs is composed of several peaks. The two main peaks are the G+ at 1590 cm-1, which is associated with vibrations along the nanotubes axis, and G- at 1570 cm-1, which associated with vibrations along the circumferential of the tube. The frequency of the G+ is sensitive to charge transfer from dopant addition [13]. The shape of the G- is dependent on whether the SWNTs are metallic or semiconducting. For metallic tubes it has a Breit-Wigner-Fano (BWF) lineshape and for semiconducting it has a Lorentzian shape. The BWF is described by 2 [ 1 + (ω − ω BWF ) / qΓ ] I (ω ) = I 0 2 1 + [(ω − ω BWF ) / Γ]. (5). where the 1/q represents the asymmetry of the peak shape, ω BWF , I 0 and Γ are fitting parameters of the central frequency, the intensity and the broadening factor, respectively. The big difference in the G- profile for metallic and semiconducting nanotubes is useful for determine the specific type of CNT that is present in the sample. Charge transfer to SWNTs can lead to an intensity increase or decrease of the BWF feature.. Second-order Raman scattering, D- and G’-band of SWNTs. In difference to the first-order features the second-order phonon mode frequencies show dispersive behaviour, i.e. change with laser excitation wavelength. The two main peaks that shows this behaviour are the D-band at 1350 cm-1 and the G’-band occurring at 2700 cm-1 (~2 ωD) for a laser excitation of 2,41 eV. The D-band stemming from the disorder-induced mode in graphite with the same name and the G’ is its second harmonic. The D-band frequency changes by 53 cm-1 when changing the laser energy by 1 eV [13]. See fig 20 and fig 22 in section 4.2.1 for examples of the dispersive behaviour of the D-band. Analysis of the intensity of the D-band can be used to identify structural modifications or defects on the nanotube sidewalls and the intensity ratio D/G can determine sample purity. With covalent sidewall functionalization an increased in D-band intensity are expected [29]. Raman spectra of MWNTs. Because of the large distribution of diameters from small to very large in the MWNTs most features from the SWNT spectra are missing. The RBM signal from large tubes is usually too weak and the RBM can only be observed in some cases for small diameter inner tube (less than 2 nm) when good resonance condition is established because of the broadening of the signal. The splitting of the G-band is small in MWNTs and appears as one weakly asymmetric peak close to the graphite frequency of 1582 cm-1.. 20.

(22) 3 Experimental methods 3.1 High-pressure method 3.1.1 Membrane DAC One alternative to generate the pressure in the DAC by manually tightening of the screws is to use a gas driven membrane to generate the pressure in a DAC. With manual loading it is difficult to change the pressure with increments less than 1 GPa (10 kbar). In our experiments we used a Membrane DAC (MDAC) purchased from Diacell (M7G), fig 10. It allows more precise pressure control and pressure increase by increments as small as 0,1 GPa. Increasing pressure in the MDAC is done by injecting a metered amount of inert gas inside the flexible membrane causing it to expand. As the membrane expands it applies force to the DAC piston increasing sample pressure between the diamonds. To control the gas pressure in the membrane, and by that the pressure generated by the anvils, equipment that made it possible to control gas flow very exact had to be manufactured.. Fig10, The M7G MDAC. 1: Anvil in support ring, 2: Piston, 3: Membrane pressure ring, 4: Gas membrane assembly, 5: Membrane retaining ring, 6: Piston holding shrews. 21.

(23) 3.1.2 Pressure control box In order to be able to do fine tuning and control of the gas pressure in the membrane we designed a gas loading system (pressure control box), fig 11.. Fig 11, Pressure Control box scheme. List of components: 1: Fine tuning needle valve, 2: Bleed Valve, 3: Regular ball valves, 4: Quarter turn valve, 5: 0,5 µm sintered filter, 6: Pressure indicator.. The work principle of the box is easy. First gas is let in to the ballast volume from the small bottle. Now the fine tuning valve (1) is opened slightly to pressurize the system. When a desired pressure is reached, the quarter turn valve can be shut, the pressure released in the box and the capillary tube disconnected from the box while pressure is maintained in the DAC. This system has many special features and allows for: a: Precise monitoring and control of gas pressure on the membrane The digital manometer “LEO2” manufactured by Keller is used to measure the pressure in the box with a resolution of 100 mbar. The accuracy of LEO2 is 0,1%. b: Fine tuning of pressure in the loading system The fine tuning needle valve (1) allows us to control the gas flow from the ballast volume very precise. A Cv at 0,0005 corresponds to a gas flow of 20 l/min at a pressure difference of 200 bar, our design manages this easily c: Cleaning of the gas before filling up the capillary A filter (5) with a pore size of 0,5 mm cleans the gas before the gas reaches the capillary tube. d: Evacuation for use of special gases. 22.

(24) The system can be evacuated and the grade of the vacuum can be measured close to the capillary tube. The small bottle makes the box easy to handle, and there is a possibility to refill the bottle with other gases if needed. e: Disconnection of the DAC with maintained pressure The use of miniature quick-connectors and a quarter turn valve on the capillary tube, makes it possible to seal of the DAC from the rest of the system and disconnect the DAC with maintained pressure. f: High safety of diamonds The small ballast volume makes it impossible to raise the pressure in the system to a higher pressure than 40 bars in one step. The possibility to seal the DAC out from the high pressure box with a quarter turn valve if something looks suspicious. The long capillary tube prevents the pressure to rise quickly on the membrane.. 23.

(25) 3.2 Sample preparation 3.2.1 Preparation of sample holders and sample loading in a DAC Sample holders (gaskets) we made out of hardened stainless steel and were about 260 µm thick. Preparation of a gasket for high-pressure experiment includes two stages: squeezing the gasket to desired thickness (pre-indentation) and drilling the sample hole in the gasket. In order to do pre-indentation using the pressure control system (Fig 11) we have carried out a calibration experiment. Five identical gaskets were squeezed at different pressures defined by the pressure control system. Further each gasket was drilled to determine its thickness after deformation. This was achieved using Betsa MH20M, a motorized Electronic Discharge Machine (EDM). The MH20M is a high precision instrument that can make holes centered with a precision down to a few micrometers. It uses electrodes of different diameters made out of tungsten wire to erode holes in metal gaskets. The calibration curve obtained in our experiment is shown in Fig 12. The gasket thickness decreases with applied pressure at a rate of 6,2 µm/bar. In our experiment the gasket was pre-indented to 64 µm. The size of the sample (the hole in the gasket) is determined by the size of the culets of the diamonds. It should be between 1/3 to 2/5 of culet diameter to ensure a safe high pressure experiment. In our case we are using 500 µm culets which corresponds to a hole size of 166-200 µm. We choose the hole to be 200 µm for our experiment.. 220. Total deformation Linear Fit of Drill_B y=6,23396x+0,76966. 200 180. total deformation, um. 160 140 120 100 80 60 40 20 0 0. 2. 4. 6. 8. 10. 12. 14. 16. 18. 20. 22. 24. 26. 28. 30. 32. 34. pgas, bar. Fig 12, the result from the pre-indent experiment. Preparation of high-pressure experiment is completed by placing the sample in the gasket hole and further filling up the hole with pressure transmitting medium. The pressure transmitting media we used was 4:1 mixture of methanol-ethanol. Closing the DAC (bringing the top. 24.

(26) diamond in contact with the gasket) was monitored using material research microscope Olympus BX51. We performed 2 experiments with similar experimental methods. The first one with a 77 mm gasket, and the second done with a 63 mm gasket. The dac was placed under CRM-200 and connected to the pressure box through the capillary tube. A fast scan was made of the area to se precisely where the ruby crystals were located, (fig 14b). Spectra were acquired from all crystals once again to have as reference in methanol-ethanol mixture, see section 4.1.. 3.2.2 Carbon nano materials a) CNT functionalization. The fuctionalization of the arc-discharge and HiPCO CNT sample were performed by our collaborates at Nancy University using the following protocol. The CNTs was placed along with toluene in a bottle and was stirred. 4-methoxyphenylhydrazine hydrochloride was added and the mixture was sonicated for 30 min. The temperature was then raised to 130° C for 72 h and then the mixture was cooled down to room temperature. Ethanol was added to dissolve unreacted 4-methoxyphenylhydrazine hydrochloride and the mixture was then filtered. The nanotubes were finally dried in vacuum at 80° C for 12 h. For further details see ref 31 and 40.. Fig 13 Free radical functionalization of SWNTs through 4-methoxyphenylhydrazine hydrochloride. Before Raman measurements the samples were dissolved in DMF with a concentration of about 1 mg/ml , ultasonicated for 30 min and were dried onto a glass slide at 50° C.. b) Polymerization of fullerenes. Polymerization of C60 was done at high temperature-pressure in a piston and cylinder device. The C60 powder sample is encapsulated in a thin-walled stainless steel cylinder which was loaded in a Teflon pressure cell along with NaCl as a pressure medium. A common protocol is to first apply a small pressure to compact the pressure medium and the sample and then heat to a desired temperature. When the temperature is reached the pressure is increased after. 25.

(27) which temperature and pressure is held constant for a certain time. Then the heat is switched of and the sample is cooled to room temperature before pressure is lowered to atmospheric. In the polymerization of fullerenes to 1D state NaCl was used as a semi hydrostatic pressure medium. The sample was heated to 550- 585 K and with a pressure of 1, 1 GPa [36]. To obtain the 2D polymerized sample the pristine material was subjected to a temperature of 820 K and pressure of 2,3-2,5 GPa [41].. 26.

(28) 3.3 Spectroscopic characterization 3.3.1 CRM-200 The Confocal Raman Microscope (CRM-200) combines a triple grating Raman spectrometer with a confocal optical microscope. With this combination it is possible to obtain a Raman spectrum of points of the sample with a lateral resolution in the sub-micrometer regime. For example using green excitation light, a resolution down to 220 nm is possible. The microscope, as well as the spectrometer and detectors, are optimized for the highest throughput and efficiency which gives the CRM-200 an unrivalled sensitivity. The system is equipped with two detectors: CCD and ADP The charge-coupled device (CCD) used is a 1D enhanced- back illuminated CCD detector that is cooled down to about -94 degrees Celsius. The SPCM-AQR Avalanche Photodiode detector (APD) is a self-contained module which has a single photon sensitivity over a spectral range from 400 nm to 1060 nm. With the CRM-200, a variety of Raman modes are possible: Single spectrum (Collection of Raman spectrum at selected sample area). Raman fast imaging (The ADP counts photons with a certain wavelength, creating a map of the intensities in the scan area) Time Spectrum (Collection of time series of Raman spectra at selected sample area) Line spectrum (Collection of Raman spectra along a selected line) Raman spectral imaging (Spectra collected from each point in a scan area) The most powerful mode of the CRM-200 is the Raman spectral imaging mode. Complete spectra are obtained at every image pixel. The number of pixels in the image is only limited by the computer memory. As an example, an image size of 512*512 would render in 262144 spectra acquired. Images can then be calculated from the spectra by applying large numbers of analyzing modes, like integrating over certain areas, calculating the peak position or peak width in the spectra. The distribution of different materials or properties of the sample can be analyzed in 3D and with spatial resolution down to 200 nm.. 3.3.2 Ruby fluorescence Ruby is a Al2O3 crystal containing about 5% Cr3+ ions. It can be found in nature (gems) and can be grown in a lab (synthetic ruby). The red colour comes from the presence of theCr3+ ions. Among the natural gems only diamond is harder. When lit upon with visible light, ruby emits light with certain wavelengths (fluorescence). Two prominent fluorescence peaks (R1 and R2) are centred around 694,2 nm and 692,8 nm respectively. These peaks are due to the 2 E 4A2 transitions of Cr3+. The R1 and R2 lines exhibit a “red” shift when the ruby crystal is subjected to high pressure. This is due to the decrease of the energy gap between these two. 27.

(29) states [42]. Pressure dependence of R1 and R2 wavelength is used to determine pressure in high-pressure experiments. During the experiments the ruby crystals, as earlier mentioned, acts as pressure sensors. To make it easier to acquire spectra from all crystals in every pressure step a fast imaging scan was made of to se exactly where the ruby crystals where located, see fig14b, and from the scan crystals with strong signal and different sizes were chosen. At every pressure increase, the hard ware spectrum from one of the crystals was monitored. b). a). Fig 14 a) Picture of the hole taken through the bottom diamond. b) Fast imaging scan of the hole in fig 14a. The yellow areas are detected signals from ruby crystals. Scanning for 694,7 nm wavelength. The scan area is 200*200 µm. 3.3.3 CNT and fullerenes In experiment study of carbon nano based materials the following factors are important: to ensure low background fluorescence, because Raman signals have generally low intensity. to have access to more than one excitation source when characterizing carbon nanotubes to ensure that nanotubes with different chiral vectors are in resonance with different excitation wavelength. to ensure that we do not introduce depolymerization of polymerized sample due to high laser irradiation. In our setup we have two different lasers, one with wavelength 532 nm (2,33 eV) and one with 632,8 nm (1,96 eV). The combination of these two lasers allowed us to take into account the factors listed above.. 28.

(30) 4 Results and Discussion 4.1 Ruby fluorescence: calibration of DAC gas loading system 4.1.1 Results from calibration experiment As seen in fig 15, the points in the graph can be divided in to three areas. The first part between 0 and 31,5 bar, which has exponential-like fit fig 15, the second part, between 31,5 and 40 bar, is linear fig 16, and a third part from 40 bar up to 52 bar. The experiment was aborted at 52 bar because the hole had grown significantly and asymmetrically, and the estimated thickness of the gasket was under 25 µm. At this moment the pressure distribution in the sample was very inhomogeneous. It reached almost 15 GPa at points B and C, and exceeded 17 GPa at point D.. P o in t P o in t P o in t P o in t. 18 16. A B C D. 14. 10 8. p. DAC. , GPa. 12. 6 4 2 0 -2 0. 10. 20. 30. 40. 50. 60. p gas, b a r. Fig 15, Pressure in the DAC (pDAC) vs gas pressure in the membrane (pgas) for different pressure probes (ruby crystals). 29.

(31) 6. 5. y = 0 ,0 0 5 6 8 1 * e x p (x /6 ,6 8 3 9 7 ). B. , GPa. 2. DAC. 3. p. 4. 1. 0 0. 5. 10. 15. 20. 25. 30. p gas, b a r. Fig 16, Data from pressure range 0-32 bar. Analyzing the first part of the pressure curve in fig 15, the data fit is clearly exponential, fig 16. The exponential growth of pressure in the DAC up to the pre-indentation pressure will be very useful in further high pressure experiments. Up to about 2 GPa we see that a change in membrane pressure renders in a rather small change in anvil pressure, fig 16. This will allow for careful studies of material properties in this pressure range. It will also be possible to change the pre-indent pressure and through that change this pressure range. 16. Y = - 7 ,9 8 2 7 2 + 0 ,4 3 9 9 9 * X 14. 10. p. DAC. , GPa. 12. 8. 6. 4 30. 35. 40. 45. 50. 55. p gas, b a r. Fig 17, Data from pressure range 30- 52 bar, linear fit, point D excluded. The most important result in this calibration experiment is when the pressure on the membrane exceeds the pre-indent pressure, fig 17. We can clearly see a linear dependence here, with a increase of about 0,44 GPa in the cell for one bar increased in the membrane. This is important to know when planning the next high-pressure experiment.. 30.

(32) In the third part, between 40 and 52 bar, the pressure in the DAC exceeds 10 GPa. At this pressure, the methanol-ethanol mixture starts to transfer in to solid state, and the pressure will become non-hydrostatic. This explains why crystal D behaves differently from the other ruby crystals in this pressure area. Thus we can assume that crystallization of the pressure medium began in the middle of the sample as crystal D is the one that lie in the very middle of the sample hole. Generally over the experiment, as the pressure went up, the ruby peaks become less intense. Exceeding 10 GPa, the peaks start to broaden and get more defuse. This is due to the nonhydrostatic pressure as the solid state transfer of the methanol-ethanol mixture occurs, fig 18.. 0 GPa 9 GPa 16 GPa. 1600 1400 1200. CCD cts. 1000 800 600 400 200 0 -2 0 0 680. 685. 690. 695. 700. 705. 710. W a v e le n g th ( n m ). Fig 18, Ruby fluorescence lines from crystal D, (left = 0 GPa, middle = 9 GPa, right = 16 GPa). In the beginning of the experiment the hole had a diameter of 198 µm, fig 19a. Up to 31,5 bar the hole shrunk down to a diameter of 179 µm. Going above 31,5 bar in pressure on the membrane, the hole started to grow again, and more alarming it started to deform un evenly and move to the left. At 52 bar, fig 18b, the hole had deformed un evenly, and moved about 6 microns to the right. The deformation was almost entirely in one direction, down right in fig 18a. The most probable reason for this sample hole behaviour is the fact that the hole was slightly off- centred on the diamond culet after the gasket drilling. Another possible reason could be the vacuum grease on the top diamond.. 31.

(33) b). a). Fig 19 a) The hole before increasing pressure in the cell. 19b) The hole at 15GPa. On pressure release in the membrane, the pressure in the DAC exhibits hysterisis fig 20. This is because the stress pattern in the gasket must be modified. When the new stress pattern is established, the pressure starts to drop more quickly [43].. P o in t P o in t P o in t P o in t. 20 18. A B C D. 16 14. p, GPa. 12 10 8 6 4 2 0 -2 0. 10. 20. 30. 40. 50. 60. p, bar. Fig 20, Pressure increase and decrease in same graph. The upper lines representing downshift.. Conclusions: • Several regions on PDAC vs PGAS calibration curve:. Hydrostatic (below 10 GPa). 1) Below pre-indent pressure 2) Above pre-indent pressure Non-hydrostatic pressure (above 10 GPa). 32.

(34) •. We calibrated (linear dependence Pdac vs Pgas) the gas loading DAS system. •. We established recommendations for further experiments different “sensitivity” in exponent vs linear regions. •. Centring of the gasket hole is very important. (Off centre 10 µm is already unacceptable). 33.

(35) 4.2 Spectroscopic characterization of CNTs Raman spectroscopy is a powerful tool when characterizing carbon nanotubes because it makes it possible to identify whether there are semiconducting or metallic tubes in the sample. Raman spectroscopy is non-destructive for individual nanotubes. It has hove ever been reported that nanotubes in bundles can be damaged by high laser flux. For more information about the vibrational properties of CNTs see chapter 2.4.3. We have done a detailed characterization of the pristine sample, section 4.2.1, followed by a comparison between pristine and functionalized material, section 4.2.2.. 4.2.1 Pristine material Two different SWNT samples, one arc-discharged and one HiPCO produced, were characterized using Raman spectroscopy. Each sample was investigated using two different laser excitations in order to probe different chiralities in the samples. For the 632,8 nm laser the effect was 1,0 mW and for 532 nm laser the effects was 0, 35 mW for the HiPCO and 0,78 mW for the arc-discharge produced sample. The spectra were acquired using a 20X objective, and the spot size was measured to be ~50 µm2. A 600 grooves/mm grating was first used to acquire a over view spectra, fig 20 and 22, and to be able to characterise the samples more precise a 1800 grooves/mm grating was used, fig 25. Typical integration time for the spectra was 1-2 s with 200-300 software integration giving a totally time of 10-20 min. We used a two step characterization process which involved first determining if the tubes in resonance were metallic or semiconducting, and then determining the chirality of each peak. To determine if the tubes are metallic or semiconducting we compare the wavenumber of each peak combined with the excitation energy used with an empirical Kataura plot. This tells us to which energy band the tubes belong. From the high resolution (1800gr/mm) spectra of the RBMs, a peak deconvolution was done using Peakfit software, and from this the chiralites of the tubes were assigned. In the chirality assignment we compare our measured RBM frequency with frequencies from literature.. Arc-discharged produced sample. The Raman spectra of the pristine arc-discharged sample are showed in fig 21, a) 632,8 nm and b) 532 nm excitation wavelength. In the spectra the RBM, D-band, G- and G+ are pointed out (see chapter 2.4.3 for theory about the Raman features). The inset spectrums in the top left corners shows the RBMs enlarged. From the wavenumbers of the RBMs we can estimate the diameter distribution in the sample by using equation 4 from section 2.3.3 with parameters for bundle sample [38]. For the sample we get diameters 1,22-1,65 nm which gives a average of ≈ 1,4 nm in agreement with ref 40.. 34.

(36) By looking in the Kataura plot fig 22, we could assign the RBMs in fig 21 to metallic or semiconducting energy bands, see inset in fig 21. From the enlarged spectra of the RBM in the insets of fig 21 a and b, we see that different types of nanotubes are in resonance from the different laser excitations. In 21a there are almost only metallic tubes belonging to the E11M band in resonance and in 21b there are only semiconducting tubes from E33S. a). Intensity (a.u.). 100. Intensity (a.u.). In the 632,8 nm excitation spectra the G- shows an asymmetric Breigt-Wigner-Fano peak shape which is typically for metallic nanotubes in bundles. This is in agreement with our assignment of the RBM to the metallic band. With the 532 nm excitation the G- peak shows a more symmetric Lorentian peak shape which is expected for samples containing mostly semiconducting tubes.. RBM. 50 40. Semiconducting tubes. 30 20. Arc-discharge produced SWNT Excitation 632,8 nm (1,96 eV). Metallic tubes. +. G -. G. 10. 50. D. 0 120. 140. 160. 180. 200. 220. -1. Wavenumber (cm ). 0 200. 400. 600. 800. 1000. 1200. 1400. 1600. 1800. -1. Wavenumber (cm ). 150. 100. Intensity (a.u.). b) Intensity (a.u.). +. G. 20 Semiconducting tubes. 15 10. Arc-discharge produced SWNT Excitation 532 nm (2,33 eV) -. 5. 50. G. 0 120. RBM. 140. 160. 180. 200. 220. D. -1. Wavenumber (cm ). 0. 200. 400. 600. 800. 1000. 1200. 1400. 1600. 1800. -1. Wavenumber (cm ). Fig 21 Raman spectra of pristine arc-discharge produced SWNT with different laser excitations, a) 632,8 nm b) 532 nm. Inset: the RBMs enlarged. 3,0. E33. S. 2,8. E11. Energy (eV). 2,6. M. 2,4 2,2. E22. 2,0. S. 1,8 1,6 1,4 100. 120. 140. 160. 180. 200. 220. 240. 260. 280. 300. -1. Wavenumber (cm ). 35. 320. 340. Fig 24 Kataura plot showing the possible tubes in resonance from fig 21 marked by two shadowed areas. The circles represent experimentally measured values from ref 44 and the triangles are from ref 45. Filled circles and triangles represent metallic tubes and none filled represent semiconducting.

(37) HiPCO produced sample. For the HiPCO sample we calculate a diameter distribution of 0,8-1,27 nm which gives an average of ≈ 1,0 nm which is in agreement with ref [31]. The RBM peak positions derived from PeakFit through the Kataura plot, fig 24, tells us that the tubes in resonance from the 632,8 nm laser, are mostly semiconducting tubes from the E 22S band but also a few metallic from the E11M band at low frequencies. With 532 nm laser we probe both metallic E11M and semiconducting E 22S tubes, but a majority of the peaks are assigned to the metallic band, fig 23b. The G-band of fig 23 shown a typical semiconducting profile in fig 23a, and in fig 23b the profile looks like we probe more metallic tubes than in 23a, and it seams to be consistent with the RBM analyse, see inset of fig 23a and 23b. HiPCO produced SWNT Excitation 632,8 nm (1,96 eV). 30. a). Intensity (a.u.). Intensity (a.u.). 200 150. 25 Semiconducting tubes. 20 Metallic tubes. 15. +. G. 10 5. 100. 0 180. 200. 220. 240. 260. 280. 300. 320. -. G. -1. Wavenumber (cm ). 50. RBM. D. 0 200. 400. 600. 800. 1000. 1200. 1400. 1600. 1800. -1. Wavenumber (cm ). +. 35. Intensity (a.u.). Intensity (a.u.). b). 200 150. G. 30 25. Metallic tubes. 20. HiPCO produced SWNT Excitation 532 nm (2,33 eV). Semiconducing tubes. 15. G. 10. -. 5. 100. 0 180. 200. RBM. 50. 220. 240. 260. 280. 300. 320. D. -1. Wavenumber (cm ). 0 200. 400. 600. 800. 1000. 1200. 1400. 1600. 1800. -1. Wavenumber (cm ). Fig 23 Raman spectra of pristine HiPCO produced SWNT with different laser excitations, a) 632,8 nm b) 532 nm. Inset: the RBMs enlarged S E33. 3.0 2.8. E11. M. Energy (eV). 2.6. Fig 24 Kataura plot showing the possible tubes in resonance from fig 23 marked by two shadowed areas. The circles represent experimentally measured values from ref 44 and the triangles are from ref 45. Filled circles and triangles represent. 2.4 2.2 S. E22. 2.0 1.8 1.6 1.4 100. 120. 140. 160. 180. 200. 220. 240. 260 -1. Wavenumber (cm ). 280. 36 300. 320. 340.

(38) metallic tubes and none filled represent semiconducting.. The analyse of the pristine samples based on their Raman spectra and the Kataura plot is summarized in table 2. Table 2 Summary of analysis for the two different produced nanotubes Excitation Sample 632.8 nm (1.96 eV) 532 nm (2.33 eV) Arc-discharge. ω < 160 Semiconducting ω > 160 Metallic. Semiconducting. HiPCO. ω < 230 Metallic ω > 230 Semiconducting. ω < 285 Metallic ω > 285 Semiconducting. Intensity (a.u.). From the Peakfit deconvolution of the radial breathing modes we see that the peaks are composed of several smaller ones when fitted with Lorentzian peak shape see fig 25. A typical full with half maximum (FWHM) of around 7,5 cm-1 were chosen because the nanotubes were considered to be in bundles [38], and a residual r2 value better than 0,99 and a good visual agreement was achieved. The background and a linear baseline were subtracted before the deconvolution. The result from the deconvolution of the arc-discharge produced sample with 1,96 eV is represented in table 3, and data for the other samples with different laser excitation can be found in appendix A. The chirality assignment (n,m) is presented in table 3, and they are assigned from the measured RBMs compared with literature. According to Maultsch et al. ref 45 the RBM frequency alone will never be sufficient for assigning the chirality, because it depends on the environment of the tubes. Since none of the literature uses DMF to disperse their samples, a small error is being introduced. We estimate the error to be no more than ±0.1 eV. 7. 7. 6. 6. 5. 5. 4. 4. 194 197.55 189.4. 165.72. 3. 170.23. 2. 3. 183.46. 2. 175.46. 1. 1. 150.12 157.29 211.32. 0 -1 125. 0. 150. 175. 200. Wavenumber (cm^(-1)). 37. 225. -1 250.

(39) Fig 25. Peakfit deconvolution of the RBM:s of arc-discharged sample. Excitation 632,8 nm Table 3. Data from Peakfit deconvolution for the RBM compared with data from other groups to assign the chirality index. M and S in last column stands for metallic and semiconducting respectively Peak Center1 Center2 Energy2 Center3 Energy3 n m M or S 1 150,1 148,8 2,05 18 5 S 151,8 2,07 20 1 S 2 157,3 155,7 2,17 16 6 S 3 165,7 164 1,69 15 6 M 4 170,2 170 1,70 17 2 M 5 175,5 175,7 1,89 10 10 M 6 183,5 184 1,90 11 8 M 7 189,4 191 1,93 12 6 M 8 194 195,3 2,02 9 9 M 9 197,6 196 1,93 13 4 M 10 211,3 212,3 2,06 11 5 M 1 Wavenumber from peakfit deconvolution 2 Experimental values from [44] 3 Experimental values from [45]. Conclusion: The Raman spectra together with the Kataura plot made it possible to determine if the tubes in the samples were metallic or semiconducting and an index assignment could be done The intensity of the D-band of both samples is relative small compared to the intensity of the G-band so we can conclude that both pristine samples have a small amount of defects on the tube sidewalls.. 4.2.2 Functionalized material To find out how the different nanotubes were affected by functionalization, Raman spectra from functionalized samples were compared with spectra from pristine samples. Two different laser excitation wavelengths were used to cover both metallic and semiconducting energy bands and to get a more complete picture of the samples. Two types of nanotubes were investigated, one arc-discharge produced and one HiPCO produced. There are special features in a Raman spectrum that one can look for to see if a sample is affected by the functionalization process. The intensity of the D-band should be higher compared to the Gband in a Raman spectrum acquired from a functionalized sample then for a spectrum from a pristine one, also the RBM modes are expected to be less intense [29]. The RBM analysis to se if the nanotubes in resonance are semiconducting or metallic are made using the Kataura plot, see section 4.2.1. The integrated intensities of the peaks are calculated using PeakFit software, see appendix B. Arc-discharge produced SWNT with laser excitation 632,8 nm (1,96 eV) and 532 nm (2,33 eV). 38.

(40) Using the 632,8 nm excitation on the arc-discharge produced samples, almost only metallic tubes from the E11M energy are in resonance. The profile of the G- band is of BWF character, which also is a sign that the tubes probed are mostly metallic. Comparing the Raman spectrum acquired from the functionalized sample with the pristine, some differences in the intensities of the peaks can be observed (fig26). The RBMs are less intense and the D-band is more intense for the functionalized sample compared to the pristine. In the inset an enlarged picture of the RBM area can be seen, and the profile of the RBM show a small tendency of the higher wavenumbers to have lost more in intensity then the lower once. As integrated intensities are compared, the D/G ratio increases with a factor 2,08 and the RBM/G quota decreases to about half its value from the pristine sample. Table 4. Calculation of the integrated intensity for the different peaks relative the G-band (both G- and G+). Arcdischarged produced, 632,8 nm excitation Integrated intensity Sample RBM/G D/G Pristine. 0,096. 0,156. Functionalized. 0,049. 0,326. Change, Funct./Pristine. 0,52. 2,08. 60. Intensity (a.u.). 75. Intensity (a.u.). 50 100. Arc-discharge produced SWNT Excitation 632,8 nm (1,96 eV) Pristine Functionalized. Metallic tubes. 40 30 20. Semiconducting tubes. 10 0 100. 50. 120. 140. 160. 180. 200. 220. 240. -1. Wavenumber (rel cm ) 25. 0 200. 400. 600. 800. 1000. 1200. 1400. 1600. 1800. -1. Wavenumber (rel cm ). Fig 26. Raman spectra of arc-discharge produced SWNT. Excitation 632,8 nm (1,96 eV). Red line is for pristine sample and blue is functionalized. The large spectrum is normalized after the G-band intensity. The inset show the RBMs enlarged. The RBM peaks on the left hand side of the line belongs to tubes in resonance from semiconducting energy band. E33S in the Kataura plot, and on the right hand side the tubes are metallic, belonging to the. 39. E11M band..

No results found