Pulsed laser deposition system

The deposition of titanium oxynitride and titanium oxycarbide films was carried out in a PLD system. The base pressure in this PLD system can reach 1x10^-6 Torr. Figure 2.2 presents a schematic view of the PLD system. The basic structure of the PLD system consists of the following parts:

(i) The substrate stage can be heated up to 700^oC.

(ii) A 2-inch target is placed opposite to a substrate stage at a distance of 14 cm. The target can be rotated to avoid pitting during deposition.

(iii) KrF (λ = 248 nm) laser beam is incident at an angle of 45^o with respect to the target.

(iv) The PLD reactor chamber is made of stainless steel to sustain high temperature and pressure.

Figure 2.2: A schematic view of the PLD system.

2.2.2. Experimental flowcharts and parameters and material analysis methods

22

Figure 2.3 shows the experimental flowcharts of deposition and characterization of epitaxial of titanium oxynitride and titanium oxycarbide films on MgO (001) substrates.

Before titanium oxynitride and titanium oxycarbide films deposition, MgO substrates were heat-treated at 700^oC for 30 minutes to obtain a smooth and clean surface. To obtain TiN_xO_y films with different chemical composition, a TiNO_0.064 target was used, and the deposition process was carried out under base pressure of 1 × 10^-6 Torr and under nitrogen ambient gas of 10^-3-10^-5 Torr. Detailed deposition parameters are shown in Table 2.1. A TiCO_0.5 target was used to deposit TiC_xO_y films, and deposition parameters for the TiCxOy films are showed in Table 2.2. After the deposition process had been completed, the substrate was cooled down to room temperature in 90 minutes.

Figure 2.3: Experimental flowcharts of deposition and characterization of epitaxial of titanium oxynitride and titanium oxycarbide films on MgO (001) substrate.

23

Table 2.1: Deposition parameters for titanium oxynitride films by the PLD method.

Sample A B C D

Table 2.2: Deposition parameters for titanium oxycarbide films by the PLD method.

Sample 1

24

The deposited films were then analyzed by atomic force microscopy (AFM), x-ray photoelectron spectroscopy (XPS), x-ray diffraction pattern (XRD), transmission electron microscopy, Hall measurements, and nanoindenter. The results of nanoindentation as well as data analysis will be presented in Chapter 3.

2.2.3. Instruments:

2.2.3.1. X-ray photoelectron spectroscopy

X-ray photoelectron spectroscopy (XPS) is one of the most powerful techniques used in the surface, interface and thin film analysis. Of all the presently available instrumental techniques for surface analysis, XPS can generally do quantitative analysis with readily interpretable and the informative results of chemical analysis.

In an XPS experiment, the sample is irradiated by low energy X-rays in an ultra high vacuum environment. This causes photo-ionisation of the atoms at the specimen's surface:

photoelectrons are emitted from the atomic energy levels with very specific Binding Energies and, consequently, with a very accurate spectral signature/fingerprint for all the elements from the Periodic Table and their chemical compounds. Quantitative data can be obtained from peak heights or peak areas. The quantitative sensitivity is in the range of (10^-2 - 10^-4) of a monolayer and the surface sensitivity is in the range of (2-100) monolayers (<0.5 - 20nm). From the results of this analysis, it is possible to infer which elements are present on the specimen, what their chemical states are (due to chemical shifts of the binding energy of the electron shells), and in what quantities they are present.

The following quantitative results are obtained with errors <10% (and <5% for using well known standards): element relative concentrations, oxidation states relative concentrations, and chemical states relative concentrations.

In general, the basis advantages of XPS are:

(i) Nondestructive

(ii) Surface sensitive (100 Å) (iii) Elemental sensitive

(iv) All elements (except for hydrogen and helium) (v) Quantitative

(vi) Chemical bonding information (vii) High information content

25

In this study, XPS was used to analyze the chemical composition of the deposited films and chemical states of titanium, nitrogen, carbon, and oxygen. XPS analyses were formed on a PHI Quantera SXM (ULVAC-PHI) system with monochromatic Al Kα radiation source. Argon ion with ion energy of 5keV was used for sputter profiling. For XPS quantitative analysis, the peak area was corrected with relative sensitivity factors from manufacture’s program and database. The spectra were deconvoluted into components using Voigt curve fitting.

2.2.3.2. X-ray diffraction (XRD)

X-ray diffraction (XRD) is a very important experimental technique that has long been used to address all issues related to the crystal structure of solids, including lattice constants, identification of unknown materials, orientation of single crystals, and ect. The Bede D1 system is a versatile high resolution X-ray diffractometer for the characterization of advanced materials. The system is most suitable for characterization of thin films, superlattices, and single crystal wafers, although it can also characterize other forms of materials. A range of parameters can be measured including thickness, composition, relaxation, strain, area uniformity, density, roughness, phase, crystalline texture, crystallinity, pore size and grain size.

In this study, a Bede D1 high-resolution x-ray diffractometer, equipped with two two-bounce Si 220 channel-cut collimator crystals (CCC), a dual channel Si 220 analyser crystal, and CuKα1 radiation (λ = 1.5406 Å), was used to investigate the crystallinity, microstructure, and to calculate the strain/stress tensors of the films. Since CuKβ (λ = 1.39 Å) is also radiated from x-ray tube with Cu target, a Ni filter is used to eliminate the XRD peaks caused by the CuKβ. A symmetric 2θ-θ scan is used to determine the d-spacing of the planes parallel to the sample surface [see Fig. 2.4a]. In this type of scan, the angle θ of the incoming beam with respect to the sample surface is varied, while simultaneously keeping the detector at an angle of 2θ with respect to the incoming beam. The angle θ at which a diffraction peak is observed, can then used to give the interplanar distance by using Bragg law. In order to determine the d-spacing of a set of planes that are tilted by an angle with respect to the sample surface, an asymmetric θ-2θ scan can be performed [see Fig. 2.4b]. As in the case of a symmetric scan, the detector is placed at an angle of 2θ with respect to the incoming beam. The incoming beam, however, makes an angle of ω with respect to the sample surface. Note that ω = θ – ψ, where is ψ the angle between

26

sample surface and the measured plane. An alternative to the asymmetric scan method is the skew-symmetric measuring geometry [Fig. 2.4c)]. As in the case of the symmetric scan, the incoming beam forms an angle θ with respect to the sample surface, while the detector is put at 2θ. The difference is that the sample is tilted over a fixed angle ψ around the axis that is parallel to the sample surface and the plane of the incoming and outgoing beam.

Figure 2.4: X-ray a) symmetric scan, b) asymmetric scan, and c) skew-symmetric scan techniques.

The XRD 2θ-θ scans, -scans, and asymmetric reciprocal space mapping (RSM) were performed under high-resolution mode set up. The set up of high-resolution mode consists of two Si 220 channel-cut crystals as the beam conditioner to provide four-bounce reflection and collimate the incident beam for 25 arcsec angular divergence, and a

27

dual channel Si 220 analyser (DCA) crystal. When RSM mode is applied, the loop scanning is used and the reciprocal space mapping is constructed by adding all the loop scan result. The X-ray reflectivity (XRR) was performed using only one two Si 220 channel-cut crystal and without DCA. The film thickness was then derived from analysis of XRR data using Bede software.

2.2.3.3. Transmission electron microscopy (TEM)

Transmission electron microscopy (TEM) is the most important instrument to obtain the atomic high-resolution images and to analyze defects in detail. There are two basic modes of TEM operation: diffraction patterns and imaging modes. As the beam of electrons passes through a crystalline specimen, it is scattered according to the Bragg’s law. The beams that are scattered at small angles to the transmitted beam are focused by the objective lens to form a diffraction pattern at its back focal plane. The scattered beams are recombined to form an image in the image plane. The diffraction patterns are obtained by adjusting the imaging system lenses so that the back focal plane of the objective lens acts as the object plane for the intermediate lens. Then the diffraction pattern is projected onto the viewing screen. At the imaging mode, the intermediate lens is adjusted so that its object plane is the image plane of the objective lens. The image is then projected onto the viewing screen.

In this thesis, cross-sectional TEM specimens were prepared by tripod polishing method and focused ion beam (FIB) technique (FEI Nova 200). The tripod polished specimens were Ar-ion milled at angle of 4-6^o and acceleration voltage of 4-5 kV. TEM specimens were then examined in a JEOL 2010F microscope at 200 kV accelerating voltage. The high-resolution TEM images were taken at high magnification up to 800k, and the lattice image of specimens can be obtained at magnification of 600k. Dislocation analyses were performed by tilting the specimen to the so called two-beam conditions, where only one diffracted beam is strong and the direct beam is the other strong spot. The steps to obtain the appropriate tilt conditions as follows:

(i) Turn to diffraction patterns mode.

(ii) Tilt the specimen into an appropriate zone axis (in this case [100])

(iii)Tilt the specimen to the two-beam condition: only transmitted beam and the desired diffracted beam g (in this case g = 002 and g = 022) and get rid of almost other diffraction spots (Fig. 2.5a). To get best diffraction contrast

28

from defects, the specimen is advisory to be tilted so that we have two-beam condition with slight deviation from the Bragg conditions. This is measured by the vector s (K = g + s, where K is a diffraction vector) which is called the excitation error or deviation parameter. The best possible strong-beam image contrast conditions for a defect imaging is achieved when the excess Kikuchi line lies slightly outside its corresponding diffraction spot g or small and positive deviation parameter s > 0 (Fig. 2.5b).

(iv) To get two-beam bright-field image: insert the objective aperture on transmitted beam at optic axis. To get two-beam dark-field image: tilt the incident beam so that the strong diffraction spot g moves onto the optic axis. If one does so, the g diffraction will become weaker and the so called weak-beam dark-field (WBDF) image conditions or 3g condition (Fig. 2.6) will be set, and then insert the objective aperture on the diffraction g.

(v) Turn to imaging mode and take the pictures.

Figure 2.5: Ewald sphere constructions and the diffraction patterns for one intense diffraction spot with a) Kikuchi line runs exactl y through its corresponding spot g (s = 0), and b) the excess Kikuchi line lies outside its corresponding diffraction spot g (s > 0).

29

Figure 2.6: 3g condition for the WBDF image. The g refection is in the optical axis with a large excitation error.

2.2.3.4. Atomic Force Microscopy (AFM)

Atomic Force Microscopy (AFM) is a common tool for the use of the surface imaging and analytical studies of roughness. Both contact and tapping mode are well suited for topographical imaging of surfaces, with vertical resolution ranging from one micron down to sub nanometer scales. In all techniques (contact, tapping, scanning tunneling, and lateral force mode) share a common approach where a motor controlling a mechanical tip is placed in a feedback loop as the tip is scanned across a surface. Tapping mode in D3100 is the most common imaging technique we have used. This mode operates by scanning a tip attached to the end of an oscillating cantilever across the sample surface.

The amplitude of oscillation ranges from 20 nm to 100 nm, with the frequency near the resonant peak of the cantilever. The tip lightly taps the surface, altering the oscillatory motion as the scanner moves across the surface. By adjusting the vertical position os the scanner to maintain a constant RMS signal of oscillation, a surface is imaged. The oscillation is measured by a laser positioned by the user to reflect signal into a photodiode detector.

30

In this study, we have used Digital Instrument Nanoscope, D3100 AFM at a scan size of 5 × 5 µm² and the scan rate of 1 Hz to know surface imaging and roughness of the films.

2.2.3.5. Electrical conductivity measurement

The Hall Measurement System is a complete system for measuring the resistivity, carrier concentration, and mobility of semiconductors and compound semiconductors.

The Hall Measurement System includes software with I-V curve capability for checking the ohmic integrity of the user made sample contacts.

In this thesis, a HMS-3000 Hall Measurement System was used to measure the resistivity of TiNxOy and TiCxOy films. Three measurements were performed for each deposited films, and the average results of the group are presented here.

2.3. Structure and properties of epitaxial TiNxOy films on MgO (001) substrates 2.3.1. Chemical composition and chemical state

Figure 2.7 shows XPS depth profiles of samples A, B, C and D with different oxygen content. As seen in Fig. 2.7, the chemical composition of samples A, B, C, and D is uniform. The chemical composition of the four samples is presented in Table 2.3. The results suggest that the addition of oxygen atoms occurred in the overstoichiometric TiNx

(x > 1). The chemical states of the four TiNxOy films were identified by examining Ti-2p, O-1s, and N-1s XPS spectra in high-resolution mode after Ar sputtering for one minute (~

2 nm). As shown in Fig. 2.8, the Ti-2p3/2 peak in the spectra can be deconvoluted into three components of Ti-N bonding (454.9 eV corresponding to titanium nitride), N-Ti-O bonding (~ 456.7 eV corresponding to titanium oxynitride), and Ti-O bonding (458.4 eV corresponding to titanium dioxide) [2.5-7]. The N-1s spectra (Fig. 2.9) reveal a small amount of chemisorbed molecular nitrogen (398.7 eV) and two main components of titanium nitride (397 eV) and titanium oxynitride (396.2 eV) [2.6] in agreement with the results determined from Ti peaks. As shown Fig. 2.10, samples with higher oxygen content show stronger O-1s signals.

31

Figure 2.7: XPS depth profiles for sample a) A, b) B, c) C and d) D with different oxygen content.

32

Figure 2.8: XPS spectra for Ti-2p of samples A, B, C and D after Ar sputtering for one minute. The spectra are deconvoluted into components of titanium nitride, titanium oxynitride, and titanium dioxide.

33

Figure 2.9: XPS spectra for N-1s of samples A, B, C and D after Ar sputtering for one minute. The spectra are deconvoluted into components of titanium nitride and titanium oxynitride.

34

Figure 2.10: XPS spectra for O-1s of samples A, B, C and D after Ar sputtering for one minute. The spectra are deconvoluted into com ponents of titanium

oxynitride and titanium dioxide

2.3.2. Microstructure

X-ray diffraction (XRD) patterns of samples A, B, C, and D in Fig. 2.11 show only TiN_xO_y (002) and TiN_xO_y (004) reflections in addition to MgO ones, suggesting that (001) oriented single-phase titanium oxynitride has been deposited on MgO(001) substrates in our experimental conditions. As seen in Fig. 2.12 of high-resolution XRD 2θ-θ (HR-XRD) scans, TiN_xO_y(002) peaks shift to higher 2θ values with increasing oxygen content due to a change in the lattice parameter and to different residual strain. The finite-thickness interference fringes in HR-XRD 2θ-θ scans indicate that TiN_xO_y surfaces and TiN_xO_y

35

/MgO interfaces are smooth. The full width at half maximum (FWHM) of TiN_xO_y(002) determined by x-ray rocking curve scans are about 58-62 arcsec (shown in Table 2.3), indicates that all as-deposited films have a good quality (the FWHM of MgO substrate is 42 arcsec). The x-ray -scan was done on those films to verify the orientation relationship between TiN_xO_y and MgO substrates. As seen in Fig. 2.13, four {022} peaks of MgO and TiNxOy appear at the same  angles with separation of 90^o. This result suggests that the TiN_xO_y films have epitaxially grown on MgO substrates with the cube-on-cube relationship of TiNxOy(001)//MgO(001) and TiNxOy[100]//MgO[100].

Table 2.3: Chemical composition, thickness, in-plane a, out-of plane c, and relaxed ao lattice parameters, FWHM of (002) TiNxOy, in-plane residual strain ε|| and

36

Figure 2.11: XRD 2θ-θ scans for four TiN_xO_y films.

Figure 2.12: High-resolution XRD 2θ-θ scans for epitaxial TiNxOy films with different chemical composition deposited on MgO substrates

37

Figure 2.13: XRD -scan of {022} reflections for TiNxOy films grown on MgO substrate, showing epitaxial relationship between the film and the substrate is TiN_xO_y(001)//MgO(001) and TiN_xO_y[100]//MgO[100].

To investigate the effect of oxygen content on the lattice parameters of TiNxOy films, XRD reciprocal space maps (RSM) of asymmetric (113) MgO and (113) TiNxOy

reflections were acquired. As shown in Fig. 2.14, asymmetric (113) MgO and TiNxOy

reflections for both samples are almost vertically aligned, implying that TiNxOy lattices are in coherency with MgO one. Hence, all TiNxOy films may be under fully compressive strain as a result from lattice mismatch and thermal mismatch with MgO. The thermal strain was induced due to a large difference in coefficient of thermal expansion (CTE) of MgO and TiNxOy (αMgO = 13 x 10^-6 K^-1 [2.8] and assume αTiNO ~ αTiN = 9.35 x 10^-6 K^-1 [2.8]). When substrate temperature dropped from 700^o C down to room temperature, MgO substrate contracted more than the TiN_xO_y layer, resulting in the generation of compressive strain of -0.54% in the TiNxOy layer. The out-of-plane, c, and in-plane, a, lattice parameters of TiN_xO_y can be determined as follows: c = 3/Q_z, and a = 2/Q_x, where Q_z and Q_x arevertical and horizontal vectors that lie along

38

Figure 2.14: Reciprocal space maps of the asymmetric (113) MgO and (113) TiN_xO_y reflections with different composition of a) TiN_1.1O_0.10 , b) TiN_0.97O_0.23, c) TiN0.81O0.38 and d) TiN0.63O0.55.

MgO [001] and MgO [110] directions, respectively [2.8]. The relaxed lattice parameter ao

can be calculated from c and a by using equation: ao = c[1 - 2ν(c - a)/c(1 + ν)] [2.8], where ν is the Poisson ratio of the deposited films. Due to the slight difference in Poisson ratio between TiN and TiO (νTiN = 0.22 [2.8] and νTiO = 0.232 [2.9]) and the large ratio of N/O of the deposited TiNxOy films, we can assume νTiNO ~ νTiN. Using MgO (113) peak as reference (aMgO = 4.2112 Å [2.8]), the lattice parameters (c, a, and ao) and in-plane residual strain ε|| and stress σ||, and theoretical critical thickness hc of samples A, B, C, and D can be calculated and are listed in Table 2.3. Theoretical critical thickness hc of TiNxOy

films is calculated by using Matthews and Blakeslee model [2.10]. The in-plane residual stress σ_|| is calculated using Young’s modulus value obtained from nanoindentation data

39

in Chapter 3. The results show that lattice parameters of TiN_xO_y films decrease with the increase of oxygen concentration but still lie in the lattice parameter range between bulk TiN (a_TiN = 4.2417 Å from powder diffraction file PDF 38-1420) and bulk TiO (a_TiO = 4.1770 Å, PDF 8-117). Those results are also in good agreement with the values reported in ref. [2.11]. Due to the fact that the radius of oxygen anion is smaller than that of nitrogen anion, the substitution of oxygen for nitrogen enables lattice parameter to decrease with increasing oxygen concentration [2.12]. Additional to the effect of anion radii on the lattice parameter, the effect of electrostatic repulsion between N and O anions around Ti vacancies has also been taken into consideration [2.13]. Indeed, the XPS results above suggest that the substitution of oxygen for nitrogen occurs in the over-stoichiometric TiNx, and such a non-stoichiometric structure has been reported to contain many Ti vacancies [2.14]. Therefore, there may have an electrostatic repulsion between anions around Ti vacancies. Besides, N atoms need three electrons to close its shell in order to achieve the most stable configuration (N^-3) while O atoms only need two (O^-2) [2.13]. Hence, the replacement of N^3- by O^-2 induces a decrease in the electrostatic repulsion between the anions around Ti vacancies and consequently lattice parameter decreases.

The second source of compressive strain can be generated from the film/substrate lattice mismatch that is determined as δ = [aMgO - afilm]/aMgO . δ is 0.73% for pure TiN, -0.716% for sample A, -0.655% for sample B, -0.466% for sample C, and -0.185% for sample D. The result also shows that more oxygen content incorporated into the deposited films can reduce the lattice mismatch between MgO and TiN_xO_y. In other words, the addition of oxygen can result in the high-quality titanium oxynitride with a composition which can make the deposited films to be excellently coherent with MgO, contrasting with the case of pure TiN on MgO. Indeed, author in ref. [2.8] have shown that stoichiometric TiN film is semi-coherent with MgO that resulted from the generation of misfit dislocations at TiN/MgO interface.

The x-ray reflectivity (XRR) curves of samples are shown in Fig. 2.15. From curve fitting, the thickness of the TiNxOy films can be determined and shown in Table 2.3.

40

Figure 2.15: X-ray reflectivity curve for the four TiNxOy films.

The epitaxial growth of TiNxOy on MgO is also confirmed by cross-sectional

The epitaxial growth of TiNxOy on MgO is also confirmed by cross-sectional