Polycrystalline Diamond Characterisations for High End Technologies

Characterisations of polycrystalline diamond (PCD) coatings have routinely been done over the past three decades of diamond research, but there is less num-ber of reports on some of its very unique properties. For example, diamond is the hardest known material and, in probing such hard surfaces with any indenter tip, it may lead to damage of the instrument. Due to such chances of experimental accidents, researchers have performed very few attempts in evaluating the mechanical properties of PCDs. In the present work, some of these very special properties of diamond that are less reported in the literature are being re-investigated. PCDs were characterised by photoluminescence (PL), Fourier transform infrared (FTIR) spectroscopy, transmission electron microscope (TEM), and X-ray diffraction (XRD) techniques. The diamond surface was also polished to bring the as-grown micron level of surface roughness (detrimental for wear application) down to few hundreds of nanometer. The tribological properties of such polished and smooth surfaces were found to be appropriate for wear protective coating application. This chapter revisits some of the unreported issues in the synthesis and characterisation of PCD coatings grown on Si wafer by the innovative 915 MHz microwave plasma chemical vapour deposition (MPCVD) technique.

But the as grown diamond surface is very rough in nature which is required to be planarised before putting into real environments. It is well understood nowadays that the diamond nucleation and growth takes place first by coalescence of the seeded layer on the substrate surface and then on top of which diamond crystals grown in columnar fashion. Such vertical growth of crystals causes the top surface to be very rough and simultaneously there may be some inherent porosity present due to rise of such vertical columns. On the other hand, the nucleation side of the freestanding diamond may have some porosity due to random coalescence of the islands during CVD processing. There is no such report of studying the surface area and porosity of CVD grown polycrystalline diamond (PCD). Here, it has been attempted for the first time to evaluate the pore size that may be present in the diamond coatings grown by CVD, using Brunauer-Emmett-Teller (BET) technique.
Moreover, tribological action may cause rise of temperature under humid atmospheric condition for diamond material to degrade. So it is necessary for the CVD grown diamond to be well characterised for the wear and friction applications [30][31][32][33][34]. Thermal stability has been studied so far by many authors [35][36][37] but it lacks proper assessment of oxidation temperature [38][39][40] since only standard furnaces have been used so far. Here, thermo-gravimetric analysis (TGA) and differential scanning calorimeter (DSC) techniques have been used for the first time ever to pinpoint the augmentation of oxidation.
Characterisation [103][104][105][106][107] of the CVD grown PCDs is very essential besides its synthesis. Some techniques which are frequently used for powder samples like, BET surface area analysis, TGA-DSC, Zetasizer particle size measurements are being reported here for the first time, for CVD polycrystalline diamond coatings. Moreover, black and white grades of diamond have been grown [108][109][110] and characterised so far, but their corresponding growth defects have not frequently been reported [111][112][113]. Here, authors have tried to elucidate the CVD growth defects that are present in their CVD grown diamonds. Afterwards, the PCD samples were polished to make them effective for tribological applications. Later on, such polished diamond surfaces have been mechanically characterised for studying tribological action [114] against silicon nitride balls under machine oil lubricants for the first time.

Processing of polycrystalline diamond (PCD)
Polycrystalline diamond samples were fabricated with CSIR-CGCRI based DT1800 microwave plasma enhanced CVD (MPCVD) reactor on single crystal Si

Physical characterisations of PCDs
The detonation nanodiamond (DND) particle size of the seeding slurry was measured by Malvern make Zetasizer instrument. Brunauer-Emmett-Teller (BET) surface area analyser (Quantachrome Instruments) was used to measure the surface area; and nitrogen adsorption isotherm (Barrett-Joyner-Halenda or BJH method) was used to measure the pore size and volume in the freestanding diamond coatings. The samples were degassed under vacuum at 100°C for 1 h prior to measurement. Thermogravimetric (TGA) and differential scanning calorimetry (DSC) analysis of the freestanding diamond sample was done at a heating rate of 10°C min −1 on a simultaneous thermal analyser (STA 449F, Netzsch, Germany). X-ray diffractograms (XRD) were recorded in the 2θ range 30°−100° at a slow scanning rate of 1° min −1 by an X-ray diffractometer (Philips X'Pert, The Netherlands) with Cu Kα radiation (at 40 kV and 40 mA). Transmission electron microscopy (TEM-Tecnai G2 30ST, FEI Company, USA) was used to evaluate the defects present inside the diamond lattice. FTIR/FIR spectrophotometer (Model: Frontier IRL 1280119, Perkin Elmer) attached with reflectance measurement arrangement with incident and reflectance angles set at 22.5°, was used for studying opaque and translucent PCD samples. The photoluminescence (PL) studies including emission and excitation measurements were carried out on a fluorescence spectrophotometer (Model: Quantum Master, enhanced NIR, from Photon Technologies International) using a xenon arc lamp of 60 W as a pump source. Raman spectra were obtained using a STR500, Cornes Technologies, (formerly known as Seki Technotron) micro-Raman spectrometer, with excitation by argon ion (514.5 nm) laser. The resolution was about 1 cm −1 with 1200 grating size. In all the Raman experiments, the laser spot size was 1-2 μm and an exposure time of 20 s was used which was repeated thrice to acquire the Raman signals during each measurement.

Mechanical and tribological characterisations of PCDs
Polished surfaces of the PCD samples were seen under a ZEISS Supra 35 (Germany) field emission scanning electron microscope (FESEM) with EDAX attachment. Roughness of the PCD surfaces were evaluated by a contact profilometer (Talysurf PGI 2000S, Taylor Hobson). The nanohardness and Young's modulus were measured by nanoindentation technique applied to the plan section Diamonds in Scientific Research and High Technology of the CVD grown polycrystalline diamond polished and nucleation side surfaces. A typical load of 1000 mN was used for the nanoindentation experiments. A standard nanoindenter (Fischerscope H100XY P , Fischer, Switzerland) was used for this purpose. The nanoindenter machine was operated according to the DIN 50359-1 standard. It offered a load range of 0.4-1000 mN. It was equipped with a Berkovich indenter. The indenter possessed a tip radius of 150 nm. The machine was used in ambient laboratory conditions (23 ± 4°C, 70 ± 5% relative humidity). The depth sensing resolution of the machine was 1 nm. The force-sensing resolution of the machine was 0.2 μN. The machine was calibrated with nanoindentation based independent evaluation of H ≈ 4.14 ± 0.1 GPa and E ≈ 84.6 ± 3.5 GPa of a BK7 Glass (Schott, Germany). This material was provided by the supplier as a standard reference block. A ball-on-disc tribometer assembly (NANOVEA) was used for highly accurate and repeatable wear friction testing in rotational modes. The counter-face was a Si 3 N 4 ball which was attached to a load arm and the load arm was directly in contact with experimental PCD disc surface. 3 mm diameter Si 3 N 4 ball was in circular motion with 5 N applied normal load against PCD surface using commercially available machine oil lubricant in ambient laboratory condition. The speed of the ball was 0.04 m/s. Optical microscope (Olympus BX 51, country) study was done after each tribological experiment to know the radius of worn out silicon nitride ball.

Results and discussion
3.1 Physical characterisations-DND seed particle size, PCD crystallinity, porosity, growth defects and thermal stability Detonation nanodiamond (DND) suspension in dimethyl sulfoxide (DMSO) by 0.5 wt% is used as seeding slurry for CVD growth of polycrystalline diamond material [118]. Nucleation density determines the faster grain coalescence phenomenon during chemical vapour deposition. It has been observed that DND seeds are much superior in enhancing nucleation of diamond than conventional micron size grits for seeding diamond nuclei on the foreign substrates [119]. The particle size of seeds is thus important in determining nucleation efficiency. One commercially available DND in DMSO was used for seeding. Before using to make seeding suspension by mixing it with 3 parts of methanol, it was essential to know the exact particle size distribution of such suspensions. Dynamic light scattering (DLS), is a non-invasive, well-established technique for measuring the size and size distribution of molecules and particles. The Brownian motion of particles or molecules in suspension causes laser light to be scattered at different intensities. Analysis of these intensity fluctuations yields the velocity of the Brownian motion and hence the particle size using the Stokes-Einstein relationship. The DND seeding suspension was procured from a commercial supplier and it was being used from time to time over a period of 1 year. When the particle size was calculated from DLS technique using Malvern make Zetasizer, it was found that the original suspension has become agglomerated considerably. The fresh stock of commercial suspension came with 25-30 nm DND particle size on its label but to our surprise after one long year in the shelf, the suspension only has 7% with original size of 25.56 nm whereas, 93% of the DND particles agglomerated to give an average size of 123.4 nm, as shown in the following Figure 1a. But such agglomeration does not appear to hinder its seeding efficiency.
The as-grown surfaces of the PCD coatings are very rough in nature which is labelled in Figure 1b. Cross-sectional microscopy image reveals columnar growth of diamond crystals from the substrate side. The interface between the Si wafer and diamond columns consists of a very thin layer of diamond microcrystals [116] which were initially coalesced with each other to form a continuous coating before diamond could grow vertically on top of them. Such thin crust layer on the nucleation side gives very poor XRD signals (top of Figure 1c) corresponding to diamond reflection planes, whereas, the peak intensities from the opposite as-grown surface (bottom one of Figure 1c) is very strong and characteristics of diamond cubic crystals. DND seeds initially formed islands which grow in a single layer until touching upon each other to form a continuous film. The diamond columns also do not grow exactly at 90° in the z direction. The adatoms from the plasma may add in a random way to produce some misorientation between the adjacent diamond columns. Moreover the top surface of diamond PCDs are very rough and sometimes appear porous. Such, (i) random coalescence of thin crust layer on the nucleation side, (ii) misorientation in between diamond columns and (iii) uneven as-grown surface of the diamond films, suggest that the CVD grown diamond may have some pores present in the freestanding coating. It is also true that in practice we assume that the CVD grown films are having 100% theoretical density but it is understood that that is not actually the case. So in order to know the pores that might present, BET surface area was calculated for the PCDs. Gas molecules are dosed into the sample chamber after cooling the solid down to a constant temperature in order to partially accumulate the gas molecules onto the solid surface which can be applied for the characterisation of surfaces as well as for the pore characterisation. The nitrogen adsorption at the temperature of liquid nitrogen (77 K) is standard method for such surface characterisation [120]. It is assumed that the nitrogen condenses onto the surface in a monolayer and so once the size of the gas atoms/molecules are known, the surface area can be estimated from the amount of adsorbed gas.
It was found that white PCD is having less surface area compare to the black PCD ( Table 1). Nitrogen adsorption and desorption isotherms measured at 77 K allow the determination of pore volume and sizes. It was similarly observed that both the values are higher for black colour PCD. So it is inferred that black PCDs are having more CVD growth defects which is giving bigger pores and higher amount of porosity-i.e. addition of carbon atoms into the diamond sp3 lattice is more incoherent. PCDs grown using 3% CH 4 in H 2 resulted in black colour (Figure 2a), whereas, the coatings were white translucent in nature (Figure 2b) when 1% CH 4 was used. Colour of the freestanding diamond coatings vary due to the presence of defects inside the diamond lattice. For example when boron or nitrogen is substituting the carbon atoms, it results in blue or yellow colour of the diamond crystals. The PCD could become completely opaque or black in colour due to the presence growth defects like dislocations, twining, grain boundaries, stacking faults etc. [121]. Figure 2c and 2d shows transmission electron microscopy images of the line defects pinned at grain boundary in one such poorer variety of black diamonds grown using 3% methane in hydrogen. On the other hand when 1% CH 4 was used to deposit diamond, the crystals were less defective and could synthesise white transparent freestanding coatings as shown in Figure 2e.
The quality of white PCD sample was found to be very high by Raman Spectroscopy. It has been found that under atmospheric conditions, the CVD diamond sample could sustain up to 700°C furnace temperature, but 750°C was not tolerable for the diamond material to resist oxidation. As per available literature [38], the procedure to study the thermal stability is to heat the diamond sample in an environment controlled furnace, but such procedure does not pinpoint the oxidation temperature of diamond material. In order to know the exact oxidation temperature of the CVD diamond prepared by DT1800 reactor, it was decided to conduct TGA-DSC experiments. Figure 3 shows the TGA curve for white PCD sample when heated from room temperature at constant heating rate of 10 K/min in air with alumina crucible up   to 900°C. There was no change in weight until 700°C temperature was crossed, after which there was continuous drop in weight, losing 29.09% of the initial value. The corresponding DSC curve shows that the exact onset temperature for oxidation reaction is 732°C, after drawing tangents. The weight loss can be attributed to the loss of carbon from diamond as gaseous by product of oxidation reaction. So, further to study the behaviour of CVD grown defects in PCD samples, they were heated in an air furnace at 600°C for 1 h. Such TGA-DSC curves confirm that annealed PCD samples do not undergo degradation on annealing and are good for studying the effect of heating on their internal defects.

FTIR studies
The FTIR reflectance spectra of as grown PCD samples, white PCD and black PCD, together with white PCD sample after being annealed at 600°C for 1 h are presented in Figure 4. The spectra reveal the presence of a broad absorption peak in the range 1745-2840 cm −1 in white PCD sample which may be attributed to the characteristic two phonon absorption of diamond [122,123] and it is seen to be almost absent in black PCD sample. However, this peak disappears completely in annealed white PCD sample with simultaneous appearance of a new peak at 1090 cm −1 . Interestingly, the peak at 1090 cm −1 is observed to be more intense in as prepared black PCD sample. This peak marks the presence of N-defects in diamond lattice [123]. Thus, it can be stated that the as grown white PCD sample contains less nitrogen defects as compared to black PCD and is of better quality in terms of non-diamond inclusions. As a result characteristic carbon-carbon bond absorption vibrational modes are more distinct in white PCD as compared to black PCD, which has many internal planar and volume defects as already observed in Figure 2. But, annealing at 600°C results in the growth of atomic defect centres in white PCD sample. As the nitrogen impurity centres grow at the expense of carbon replacement in the lattice, intrinsic absorption band due to diamond is weakened while the vibrational band due to N-defects intensifies.
In addition, a third distinct peak is noticed at around 1490 cm −1 in both as grown white PCD and annealed sample which may be assigned to absorption due to H-centres trapped in the lattice [124]. In black PCD sample this peak is seen to be extremely weak in nature. This may be explained on the basis of the processing condition of the two samples. White PCD was grown with 1% methane and 99% hydrogen whereas black PCD was grown with 3% methane and 97% hydrogen. Thus use of more hydrogen precursor during the CVD process may have resulted in trapping of H-centres in the lattice of white PCD. For further confirmation of the presence of impurity centres, photoluminescence study of these samples was performed.

Photoluminescence studies
Photoluminescence is a characteristic property of several defect centres in lattice and is thus a confirmatory tool for the presence of defects especially in diamonds.
In an attempt to have detailed information regarding non-diamond inclusions, the photoluminescence spectra of all the samples were recorded under excitation of 270 nm wavelength and the spectra is presented in Figure 5.
The spectra reveal the presence of seven distinct peaks at 470, 496, 567, 625, 689, 734 and 767 nm respectively. Pezzagna et al. [125] suggests that small PL signal for neutral nitrogen vacancy centres is present at 575.67 nm, but for the negatively charged nitrogen vacancy centre, the PL peaks are observed at 637 nm with multiple phonon replica thereafter. It is also well known that silicon vacancy centre emits strong PL signal at 738 nm and a weak replica at 758 nm [121,126]. Although nitrogen vacancy centres are best photo emitters for diamond but it has strong photon-phonon interactions which is responsible for broad emission spectrum from 600 to 850 nm. That may be the reason for us not getting the nitrogen vacancy signal exactly at the theoretical values in Figure 5. The peaks at the 567, 625 and 689 nm can be attributed to nitrogen vacancy centres, but not definitively. Similarly the peaks at 734 and 767 nm in Figure 5 can be assigned to silicon vacancy centres but again, it is a complex defect centre with an impurity atom and vacancy-so the peaks are not always reproducible at identical positions. Surprisingly there are two more peaks in Figure 5 at 470 and 496 nm which are typically found for Ni-related defect centres-but usually reported for HPHT synthetic diamonds [127]. It is not yet known about the origin of such PL peaks in CVD grown diamonds. Although the presence of the peaks can be identified in all the three samples, variation in the peak intensity indicates the difference in  quality of the samples. It is observed that black PCD sample exhibits the stronger peak intensity as compared to that of white PCD. This indicates white PCD contains lesser defect centres and is of better quality as compared to black PCD which is in accordance with the FTIR results. However, upon annealing white PCD sample is seen to exhibit strongest peak as compared to as prepared white PCD and black PCD samples. This signifies that the concentration of defects in white PCD is enhanced upon annealing which may be explained as due to possible aggregation of defect centres. At elevated temperature, the vacancy in a lattice becomes mobile and moves through the crystal and aggregates [128]. Vacancies upon encounter with isolated impurity atoms (nitrogen, silicon), forms the aggregation of defect centres thus resulting in enhanced luminescence.

Raman spectroscopy
The diamond peak positions of both the white and black annealed PCD samples are at 1328.45 cm −1 which is a downshift from the theoretical peak position for single crystal diamond as shown in Figure 6. Down shift of spectra is due to the tensile stress inside the lattice (thermal stress due to expansion mismatch with the Si substrate does not exist as the PCDs are freestanding) [129]. Now on the other hand, a-C (amorphous carbon) hump centre is around 1486 cm −1 for white annealed PCD but such diamond like carbon (DLC) phase centre shifts to 1560 cm −1 for the black annealed PCD sample. So this a-C hump central position change may be an indication of the nature of predominant amorphous carbon phase that is present in the samples. It is essentially trans-polyacetylene (TPA) for white annealed sample; whereas, the non-diamond carbon inside black annealed PCD is primarily graphitic carbon. The ratio of peak intensities of diamond to DLC gives estimation about the quality of the deposited diamond coating. Now to calculate the area under the curve, corresponding to the sp3 phase, the integration has been carried out from 1300 to 1360 cm −1 , which therefore would take into account of any upshift/downshift of the diamond peaks and disordered graphite that might also contaminate the coating. The integrated area for DLC phase is calculated from 1450 to 1600 cm −1 . As expected the quality of white annealed diamond (41%) has found to be better than the quality of the black annealed PCD (33%) sample. The FWHM value of diamond peak is also better for the white annealed PCD (4.62 cm −1 ) than the black annealed PCD (7.38 cm −1 ), as found from the analysis of the curves in Figure 6.

Polished PCD surfaces
Annealed samples were different from the PCD samples that were polished for mechanical property and tribology studies. Polishing was done to reduce the as grown surface roughness (Figures 2b and 7a) which were in the order of 3-4 μm. Before starting polishing experiments, the white as grown PCDs were characterised by Raman spectroscopy (Figure 7b). It was found that the as-grown diamond samples were having tensile stress of about 0.8 GPa (calculated from the peak shift data). It also had some non-diamond impurity in the form of TPA which gives very little undulations around 1508 cm −1 . The FWHM was calculated to be 5.5 cm −1 at the diamond peak position. But when the Raman spectra were taken from the same white as-grown PCD sample after polishing, there is considerable enhancement of diamond peak intensity with appearance of graphitic D peak-giving shoulder to the sp3 peak position, centred around 1329 cm −1 . Tensile stress appears to increase on the surface of the same white PCD sample up to 1.34 GPa after polishing action. Moreover, the a-C hump becomes apparent in the polished sample which was not prominently present in the un-polished white PCD sample. The centre of a-C hump is around 1492 cm −1 , which indicates that the non-diamond composition was essentially trans-polyacetylene (TPA). FWHM of the diamond peak increases for the polished sample (7.65 cm −1 ) which also confirms the degradation of the quality of the original (55%) white PCD surfaces after polishing (26%). Figure 7c shows the EDAX signals received from such polished surfaces under electron microscope. It can be seen that the carbon peak is not predominant due its low atomic weight. There are prominent peaks corresponding to metal elements like Fe, Bi, which might came from the metal bonded polishing discs. Moreover there are oxygen and sulphur elemental peaks also present, which are essentially from the oxidising chemicals used during chemo-mechanical polishing.
In order to reveal the morphology of polished diamond surfaces, Figure 8 is provided with successive higher magnification images up to 25kX. It can be seen that there are longitudinal polishing marks, as shown by double headed arrows. Essentially the area under observation is covered with wear debris generated during both mechanical and chemo-mechanical polishing. Due to abrasive action of the polishing materials, fine particles were generated, which, when reacted with the polishing chemicals, were oxidised to dislodge diamond particles from the as-grown rough surfaces. Thus the final microstructures were like wear debris compressed under polishing load with intermediate pores and cracks. Some regions were very smooth with grooves present inside, whereas, their adjacent areas were very rough with scattered particles spread all over. It appears that such unclean top surface after polishing has to be annealed/chemically cleaned in order to remove non-diamond impurities and wear debris, before putting into use for technological application.

Hardness and elastic modulus
Nanoindentation was carried out on black and white PCD samples, both on the polished surfaces and on the mirror smooth nucleation sides (NS). Altogether four PCD surfaces were indented, as shown in Figure 9. The roughness of each surfaces are labelled in each Figure 9.
At least 10 nanoindentations were made along a line for the plan sections of each CVD grown polycrystalline diamond surfaces, at an applied load of 1000 mN. The time to reach the peak load was kept at 30 s for all the nanoindentation experiments. The unloading time was also kept same as the loading time. The experimental conditions for all the nanoindentation measurements were kept identical. From the experimentally measured load (P) versus depth of penetration (h) data, the nanohardness and Young's modulus values are evaluated for all PCD samples. Oliver and Pharr method [130] was utilised to evaluate the nanohardness and Young's modulus values.
The experimentally measured load (P) versus depth (h) plots for the plan sections of four different types of CVD grown polycrystalline diamond surface samples are shown  in Figure 10. In general, all the P-h plots were smooth in nature. An important observation from these data was that at comparable loads, the P-h plots of Black PCD NS, White PCD polished and Black PCD polished were much steeper than those of the White PCD NS sample. Moreover, at comparable loads the depth of penetration data recorded for the Black PCD NS, White PCD polished and Black PCD polished samples was much smaller than those of the White PCD NS sample. These data (Figure 10) indicated that nanohardness of Black PCD NS, White PCD polished and Black PCD polished would be higher than that of White PCD NS samples. The P-h plots obtained for White PCD NS samples showed more of the plastic deformation behaviour (Figure 10) compared to Black PCD NS, White PCD polished and Black PCD polished. However, the P-h plots obtained for Black PCD NS, White PCD polished and Black PCD polished samples showed more of the elasto-plastic deformation behaviour.
The nanohardness (H) of the PCD samples is shown in Table 2. Results show that the polished PCD surfaces have much higher hardness compare to their corresponding nucleation side, as expected. Black PCD has almost 5 GPa less hardness than white PCD polished surface, may be due to the presence of softer carbon phases other than diamond. But on the other hand, their corresponding nucleation sides are showing opposite trend-black PCD NS being harder than White PCD NS. It is already understood from our previous results and discussion that the black PCDs are having more CVD growth defects than white PCDs. Presence of such defects make the poorer variety diamond coatings black or grey in colour; their Raman signals become different as well as their other spectral data. Figure 1b and c shows that the as grown side is different from the nucleation side. Nucleation side has thin crust like layer whereas the as-grown side is having extended ends of the elongated diamond columnar crystals. From nanoindentation data, it is being revealed that the nucleation side of black quality diamond is harder which may be because of the presence of harder diamond like carbon phases than their white variety. Corresponding average Young's modulus (E) values showed the similar trend as nanohardness. The samples can be ranked in the increasing order of elastic modulus as follows: White PCD NS > Black PCD NS > Black PCD polished> White PCD polished. The hardness and elastic modulus results found in the present work are in agreement with the literature [93,105] values.

Tribological properties
In order to understand the efficacy and necessity of the polishing method adopted in the present work, the white PCD sample was chosen for further tribological studies. Both the polished and the corresponding opposite nucleation side were made to rotate against 3 mm diameter Si 3 N 4 counter-face ball under machine oil lubrication. The coefficient of friction was found to be very low in compare to our earlier study of the similar polished/unpolished diamond surfaces against alumina balls [131]. Here also the diamond surfaces do not wear, but the counter body Si 3 N 4 wears out, although not as severe as alumina ball which worn out under dry and SBF lubricating conditions by travelling half of the distance travelled by Si 3 N 4 balls in the present case. The polished side of white PCD sample is rougher than their corresponding nucleation side (Figure 9a and  b). So, as expected the COF values were found to be higher (~0.08) for polished white PCD surface than the nucleation side of white PCD sample (~0.03). The Hertzian contact pressure was calculated to be 26.2 GPa when 3 mm diameter ball was made to rotate against PCD disc surfaces under 5 N normal load as shown in schematic Figure 11a. The worn out ball was investigated under optical microscope (Figure 12) and the wear data was calculated as described in authors previous work [131]. It was found that the wear rate for the polished white PCD surface is very small 4.08 × 10 −7 mm 3 /N m and the wear rate for its corresponding nucleation side is much higher 3.95 × 10 −5 mm 3 /N m. The hardness of the polished side was found ~28 GPa, whereas its nucleation side was much softer with ~10 GPa nanohardness. It is counter-intuitive that the counterbody Si 3 N 4 wears out more against the softer diamond nucleation surface than while rubbing against the hard polished surface. In order to explain this apparent anomaly, it was necessary to know the hardness of Si 3 N 4 from commercial supplier's database and it was found to be 10-12 GPa. Such comparable hardness values of diamond nucleation surface and Si 3 N 4 counterbody might have caused nanodiamonds that are present in loosely held colonies [118] on the nucleation side to come out under abrasive wear action of silicon nitride. Such nanodiamonds that come out from the soft nucleation side under abrasive wear action may cause the Si 3 N 4 counterbody to further wear out severely, since nanodiamonds are very hard particles.
CVD diamond surfaces sliding against silicon nitride balls reportedly give high COF values under dry lubricating conditions [91]. Researchers have tried commercial soybean-derived biodiesel lubricant [132] to improve the performance of the combustion engine. They could achieve 0.07 as the minimum value of COF when trying self-mated diamond coated Si 3 N 4 material. They used 86 m of sliding distance under 2.5-4.5 GPa initial Hertzian contact pressures. Recently a grapheme coating has also been applied on top of microcrystalline diamond surfaces to improve the COF values against Si 3 N 4 balls under reciprocating actions [133]. 0.04 was the minimum achievable COF value. The researchers could notice wear track on such grapheme coated diamond flat surfaces. The calculated wear rate of the counterface Si 3 N 4 ball from the measurement of the ball scar diameter was found to be in the order of 10 −5 mm 3 /N m. This same group of researchers also conducted a tribo-map of diamond films [134] against silicon nitride ceramics sliding in air, but under reciprocating action (present paper reports circular motion). The normal load was 2-6 N, sliding speed was 0.13-0.3 m/s. They reported 0.07 as the best possible COF value with nanocrytsalline diamond, which was increased to 0.11, when the diamond film became microcrystalline. They also could observe worn out diamond film surface; and profilometer scan wear volume data revealed wear rate less than 10 −7 mm 3 /N m. Bogatov et al. [135] also reported >10 −7 mm 3 /N m wear rate of diamond film sliding against silicon nitride ball (ball characteristics were same as used in the present work) with 0.12 COF after 24,000 sliding cycles. But in the our present experiments with diamond disc, no wear track could be seen on the diamond disc surfaces sliding against silicon nitride balls-as also it was observed against alumina balls in authors' previous paper [131]. Only the balls worn out and the presented wear data of balls are in tandem with the reported literatures.  Moreover, the lowest COF of ~0.03 obtained using machine oil lubricant in the present paper appears to be better than liquid bio-diesel or solid grapheme lubricant used for diamond coatings previously.

Conclusions
As-grown, annealed, polished CVD grown diamonds have been characterised adopting some new techniques which are conventionally used for powder bulk specimens. Use of TGA-DSC analysis could pinpoint the diamond oxidation temperature for the first time. Attempt has been made to quantify the porosity that might be present in the freestanding coating, which reduces the theoretical density values of CVD grown diamond. It has been found that the internal defects that are present inside diamond lattice determine the black or white nature of diamond coatings. Qualitative analysis of such defects that are present in different grades of PCDs (line, plane or volume defects) like dislocations, grain boundary, twining has been revealed by transmission electron microscope. Whereas, the other atomic defects, like nitrogen vacancy, atomic hydrogen, are revealed by FTIR and PL studies. Raman signals quantify the internal stress, crystallinity and the quality of the CVD grown diamond coatings. It has been found that after polishing the surface of diamond becomes very unclean due to contamination from the polishing consumables and generation of non-diamond carbon phases. Use of machine oil lubricant for tribological application of such polished diamond surfaces has found to be very effective in reducing the COF values considerably. However, the CVD grown diamond nucleation side produced best COF value of about 0.03 with 3.95 × 10 −5 mm 3 /N m wear rate for the silicon nitride counterface.