1. Introduction
Diamond coatings on WC-Co hard alloy tools, due to their high hardness and thermal conductivity, low coefficient of friction, strength, and wear resistance, play an essential role in mastering the production of promising materials [
1,
2,
3,
4]. The development of new methods for applying polycrystalline diamond coatings (DC) directly to the working surface of base materials enables the manufacture of axial tools using the best properties of diamond. Axial tools (cutters, drills, mills) with chemical vapors deposition (CVD) diamond coating are mainly used for processing carbon-fiber-reinforced plastic (CFRP), metal matrix composites (MMC), ceramics, graphite, titanium, printed circuit boards (PCB) and other hard-to-process materials [
2]. The diamond-coated cutting tools have higher wear resistance and longer service life compared to uncoated ones. However, the characteristics of diamond films, both tribological and physical–mechanical, may have apparent differences in the perfection of the crystallite structure and the morphology of the surface of the diamond film. Nanocrystalline diamond (NCD) films, due to their record high hardness and reduced roughness, are particularly suitable for mechanical and tribological applications [
5]. The primary purpose of diamond coatings applied to WC-Co tools is to ensure wear resistance and heat removal from the cutting edge; for this purpose, the films must be resistant to delamination and to chipping [
6].
The technology of applying polycrystalline diamond coatings is quite complex, multi-stage, and has some technological features [
7,
8,
9,
10]. The most common method of applying DC to tungsten carbide-based hard alloys (WC-Co) is gas phase CVD deposition [
11], more often using the hot filament chemical vapors deposition method (HF CVD) [
12] or, less frequently, the microwave plasma chemical vapors deposition method (MP CVD); however, this provides a higher growth rate and film quality compared to HF CVD. The expansion of the possibilities of widespread use of MP CVD is hindered by a violation of the stability of the process and the uniformity of the diamond film, which depends on the shape and ratio of the 3D dimensions of the substrates and their aspect number [
13,
14], as well as the types of unique means to ensure uniform heating associated with the shape of the plasma [
15]. The formation of the microwave plasma is sensitive to the geometry of the substrate due to the edge effect: the plasma is concentrated in areas of electric field distortion, especially at the edge of a strongly protruding substrate, which causes uneven deposition of the diamond film. The detrimental effect of the local microwave field on the temperature increase is aggravated with an increase in the substrate length [
16,
17], for example, for a drill or a mill. Therefore, temperature uniformity becomes a critical condition in determining the possibility of the successful growth of the diamond film.
In the work cycle [
18,
19,
20], DC deposition on WC-Co was carried out at a substrate temperature of 700–750 °C and a pressure in the chamber of 80 Torr. The proportion of methane in hydrogen was 1%, the deposition time was six hours, and the deposition rate of DC was 1.5 µm/h. In Refs. [
21,
22,
23], when nanocrystalline DC was deposited in microwave plasma, nitrogen was added to the gas mixture, and smooth and solid (80 GPa) DC ~10 µm thick was obtained. In Refs. [
24,
25,
26], the samples of a new type of super hard tool with significantly higher wear resistance were created based on DC, intended for use in producing parts and structures made of difficult-to-process composite materials. To a large extent, the root problem of DC deposition in microwave plasma has been solved, due to the so-called edge effect, which consists of perturbation of the electric microwave field due to the heterogeneity of its concentration in the plasma volume and temperature along the perimeter of a metal substrate with a low aspect number when the substrate lies on the surface with the largest linear size on the central conducting platform. Therefore, the grain size inhomogeneities in the coating area, uneven coating thickness, and different degrees of inclusions of the non-diamond phase are possible, which is unacceptable for cutting tools used for high-precision surface treatment when an essential characteristic of which is the homogeneous fine-grained structure of DC.
In Refs. [
24,
25], an original design of the substrate holder was proposed, which enables to implementation of the deposition mode of a homogeneous DC and eliminates direct heating of low-aspect (height-to-diameter ratio) substrates in a microwave plasma reactor. The transition to indirect heating in the ethereal region of the plasma enabled the reduction in the edge effect sharply and protected the edges ether the substrates from overheating, as a result of which coatings with uniform morphology were obtained both in the center region and on the periphery of the DC. It was found that the temperature affecting the growth rate of diamond and the grain size of polycrystalline DC depends on the distance between the plasma-forming surface of the substrate holder and the growth surface of the substrate. The temperature range was determined with an accuracy of 20 °C. It was possible to deposit DC on plates made of WC-6% Co with an aspect ratio of 0.45. The plate’s roughness in the center and periphery of the coating differed by no more than 10% [
27]. It is shown that an effective method of regulating the temperature of the growth surface and controlling the structure of polycrystalline DC is an operational change in the distance between the substrate and the substrate holder during synthesis in microwave plasma with the position of the cooled substrate holder unchanged.
This paper presents a method for applying a conformal multilayer MCD/NCD coating in MPCVD plasma to an axial tool model (TM) with a high aspect number. The microwave source of the Ardis-100 reactor operates at a frequency of 2.45 GHz (wavelength λ ≈ 122 mm). Therefore, an excessive ring (ER) with two types of holes ~1/2 and 3/4 λ was used to weaken the density of the microwave electric field acting directly on the TM volume inside the ring. The outer diameter of the ER is determined by taking into account the length of the TM to accommodate another part of the TM volume that acts as a holder and protects it from heating by microwave plasma. The undesirable effect of the edge is minimized. It is assumed that the primary heating source is indirect heating from the thermal radiation of high-temperature plasma generated in the region of the microwave field beam symmetrical to the inner hole of the ER. The TM heating during diamond deposition is modeled using a surface heater above the model with a lower surface temperature of 1400 °C. Modeling was performed using the finite element method (ANSYS) for TM with a diameter of 12 mm and a length of 75 mm inserted into a molybdenum sector of a 15 mm high ER with a hole for TM of 13.5 mm. It can be seen that this thermal system reaches stationary heat exchange conditions in about 20 s, and heating/cooling to T = 400 °C takes less than 5 min of rotation around its axis. Hence, the rotation speed is determined to be at least 12 rpm. The heating/cooling cycles in building up the diamond film should be shorter than 2.5 s with such a small thermal inertia of the model. The structure of the diamond film deposited with TM rotation is a 12-layer MCD/NCD conformal coating with a total length of 40 mm with an average NCD grain size of 51 nm.
2. Materials and Methods
The DC growth was carried out in the ARDIS-100 microwave reactor (Opticsystems, Moscow, Russia). The simulation was carried out in the ANSYS Multiphysics (Canonsburg, PA, USA) program (the steady-state thermal module for calculating thermal fields and heat transfer optimization options for leveling the tool model (TM) temperature, the transient thermal module for calculating the thermal inertia of the system and the minimum required rotation speed of the TM). The model is a cylindrical imitation of a tool with a diameter of 12 mm inserted into a molybdenum holder with a height of 15 mm with a hole for TM of 13.5 mm. The internal diameter of the molybdenum ER measured 52 and 75 mm. TM heating during diamond deposition is modeled using a surface heater with a surface temperature of 1400 °C.
A calibrated rod of KFM-39 tungsten carbide manufactured by Konrad Micro Drill (Kulmbach, Germany) hard alloy with a size of Ø12 × 75 mm was used as a model for coating, representing the equivalent of an axial cutting tool. A hole in one of the four sectors of the ER made of molybdenum was used as a TM holder. The region in the diametral and lateral sectors of the ER served for visual and temperature control of the TM heating. The rotation mode was carried out using a microwave transparent shaft from a stepper drive mechanism hermetically installed in the flange of the reactor.
The controlled TM surface was chemically etched with Murakami and Karo reagents, and to prevent cobalt diffusion at the deposition temperature of the coating, a 600 nm thick layer of tungsten was applied via magnetron sputtering based on the schematic diagram of obtaining diamond films on WC-Co alloy [
7].
The TM surface was pre-coated with diamond nanoparticles about 5 nm in size by centrifugation (2000×
g rpm) from a water-based suspension. The growth of polycrystalline diamond films occurred in a methane/hydrogen gas mixture at a concentration of CH4/H2 = 4%, a pressure of 65 Torr, a microwave radiation power of 5 kW, and a substrate temperature of 800 °C ± 15 °C, which was controlled using a two-beam pyrometer METIS M322 (SensorTherm GmbH, Sulzbach, Germany). Additionally, 12-layer DC with reduced roughness was obtained according to the method described in [
25]. The stimulation of secondary diamond nucleation was carried out via short-term nitrogen additions to the gas mixture during synthesis. The DC consists of alternating layers grown in microcrystalline diamond (H
2/CH
4 = 480/20 cm
3/min) modes for 25 min and nanocrystalline diamond (N
2/CH
4/H
2 = 20/20/460 cm
3/min) modes for 5 min. The total DC growth time was 6 h.
Microstructural studies and energy-dispersive X-ray spectroscopy were performed using a JEOL JSM-7001F microscope (Tokyo, Japan) at an electron energy of 30 keV to probe the material to a depth of approximately 3 µm.
The surface topology was investigated using the ZYGO NewView 5000 white light profilometer (Middlefield, CT, USA).
The uniformity of the coating on the surface of a TM was estimated through variations from the roundness of the profile and alignment by measuring deviations of the radius vector in the Cartesian coordinate system, followed by the calculation of approximating circles using the least squares method after applying the 50 irregularities/revolution filter. For that purpose, MarForm MMQ 400 (Göttingen, Germany) was used with a probe with a ball tip diameter of 3.0000 mm, and a contact force of 0.2 N. The profiler Hommel Tester T8000 (Hommelwerke Gmbh, Villingen-Schwenningen, Germany) was also used. The diameters of the workpiece were measured using the NORGAU NVMII-4030Di video measuring system (Moscow, Russia).
The structural phase analysis of the coating was performed via Raman spectroscopy using a LabRam HR 800 spectrometer (HORIBA Jobin Yvon, Tokyo, Japan) with the following parameters: laser radiation wavelength 473 nm, power 70 mW (the power on the sample was halved by the filter). The beam was focused using a long-focus lens Olympus ×100 (NA = 0.9) into a spot measuring Ø1 µm. A diffraction grating with a stroke density of 1800 mm−1 was used to measure the lines of the Raman spectra (RAMAN).
3. Results and Discussion
3.1. Simulation of a Cylindrical Tool Model Heating in Microwave Plasma (2.45 GHz)
Heat transfer between objects included radiation heating of surfaces from the lower surface of the heater with a temperature of 1400 °C, and gas convection in the internal volume of the model, as well as thermal conductivity in gas and solids. The cooling of the model was carried out by maintaining a constant temperature of the lower surface of the base at 25 C. Three calculations were carried out: one with the inner diameter of the molybdenum ring 1/2 λ and with the length of the TM part inside the molybdenum ring 40 mm, and two with the diameter of the ring 3/4 λ and the size of the protruding TM part 40 and 50 mm. The result of the thermal field calculation for the case of two (3/4 λ and 40 mm) is shown in
Figure 1, from which it can be seen that for a good heat sink, the temperature of the molybdenum ring does not exceed 250 °C. The temperature of the TM varies significantly in different parts of the model depending on the efficiency of heat removal in the holder and in the free part of the TM. The surface temperature at the end and in the upper part of the TM is shown in the graphs in
Figure 2.
From the graphs in
Figure 2, it can be seen that an increase in the diameter of the molybdenum ER, as well as an increase in the length of the free end of the TM, contributes to the alignment of the thermal field both on the cylindrical surface and on the end of the TM. This phenomenon is due to the heat removal from the TM to the ring holder being 3–5 times higher than on the free surface of the cutter. This can be seen both through the large temperature gradient at the junction of the TM and the holder, and through the value of the heat flux on the surface of the TM. Additionally, in the conditions of non-stationary heat exchange, the thermal inertia of the model was evaluated with instantaneous switching on and off of the heater. The temperature graphs of the upper and lower parts of the cylindrical TM are shown in
Figure 3.
It can be seen that this thermal system reaches stationary heat exchange conditions in about 20 s, and heating/cooling to T = 400 °C takes less than 5 s, that is, to smooth out the temperature difference on the TM surface by rotating around its axis. The rotation speed should be at least 12 rpm. With such a small the thermal inertia of the model, the heating/cooling cycles in building the diamond film should be shorter than 2.5 s.
The following design options of the ER and the placement of the TM were considered to reduce the longitudinal temperature gradient in the TM (
Figure 4).
The thermal fields were calculated for the considered configurations. It is concluded that it is necessary to minimize the heat sink through the molybdenum ER, in the version with a quartz gasket, and to reduce the thermal contact between the tool model body and the molybdenum ring since these places are the heat removal bridges in this thermal model. The best result is shown by moving the TM into the hole by 10 mm. Ideally, only the TM holder should remain inside the hole, and the heat sink through the holder should be excluded.
The results of the TM heating modeling showed that the temperature along the TM axis decreases from the free end to the end fixed in the holder inside the molybdenum ring. The TM body temperature drops sharply in the area near the holder. Rotation above 12 rpm during deposition makes it possible to compensate for the radial temperature unevenness on the TM surface. It is possible to balance the temperature difference along the axis only by increasing the length of the TM free part since heat dissipation in the holder is much higher, and changing its design does not lead to a fundamental change in the conditions of heat exchange. The departure inside the outer ring of the working part of the TM with a temperature of 700–800 °C, in which the growth of the diamond coating is possible, is 20–25 mm.
For experimental verification of the simulation results, the following scheme of a growth plant with an Ardis-100 microwave reactor was adopted (
Figure 5).
The results of modeling the temperature distribution over TM agree with the experimental data related to pyrometric temperature measurement on the workpiece (
Figure 6).
3.2. Results of Linear Measurements of a Cylindrical Workpiece Coated with Polycrystalline Diamond
The measurement results are shown in
Figure 7 and
Table 1. The measurement positions of the diameters and deviations from the circles coincide. The eccentricity value was 2.09 µm.
It can be seen from
Figure 7 that the single high values of deviations from the circumference and the profiles along the lines along the forming cylinder, as well as the presence of eccentricity, are caused by the chaotic distribution over the surface of individual elements of the growth of the diamond coating up to 5 µm high. Without them, the deviation does not exceed 1 µm. The coating thickness is uniform. According to
Table 1, the variation from the roundness of the surface of the uncoated TM at a distance of 50 mm is 0.56 µm. In the coated areas, two maxima are distinguished at a distance of 1 and 20 mm from the end of 4.34 and 4.09 µm, respectively, and between them, a decrease to 2.86 µm.
The average surface roughness values Ra ≈ 5 µm and Rz up to 27 µm were measured with an optical profilometer on an area of 0.9 × 0.7 mm for the initial samples. The surface reliefs of the films for the lines along the forming cylinders of the TM in four positions through 90° with a step of 10 mm are compared in
Figure 8. At the end and at a distance of 10 mm from the rear, areas with increased roughness values were observed.
The analysis shows that the regularity of the increase in roughness along one of the measured lines at any distance from the end is not observed. The most significant values are marked on the 180° degree line at a distance of 1 mm from the rear and on the 0°, and 90° lines at a distance of 10 mm; the size of the image area was chosen in so as to obtain sufficient statistics.
A more pronounced smoothing from Ra ≈ 186, 222, and up to Ra ≈ 287 nm was detected after measuring the end face on the line 0°,90°, and 270° (
Figure 8a), and from Ra ≈ 222, 238 and 1270 nm at a distance of 10 mm on the line 90°, 180°, and 270° (
Figure 8b). Since the deposition of diamond material occurs via rotation at the calculated speed, the final roughness cannot be equalized in a narrower interval. Improved surface roughness can be achieved using alternative rates found experimentally in further studies.
3.3. Results of RAMAN Spectra
The spectra for Raman analysis were recorded along the lines starting from the end every 10 mm. The lines are located on the cylinder on opposite sides to each other. That is, each line is 90° away from the previous one. Thus, we obtained four lines of four spectra in each. The spectra (
Figure 9) are typical for a nanocrystalline CVD diamond film with a highly disordered graphite sp
2 phase content and CH-compounds. The presence of graphite is indicated by D and G bands located at 1350 and 1550 cm
−1, respectively. These bands have a significantly higher intensity than the diamond peak about 1337 cm
−1. The high CH compounds’ content at the intercrystalline boundaries is indicated by trans polyacetylene (TPA) bands at 1140 and 1490 cm
−1 due to the uneven heating of the milling cutter. Significant differences in the shape of the spectra were observed at points recorded at the same distance from the end face but on different lines due to the uneven heating of the milling cutter.
The parameters of these bands such as intensity, position, and width at half-height were determined by approximating the spectra with Lorentz curves (for the diamond peak and the trans polyacetylene band) and Gauss curves (for the G band).
Figure 10a shows graphs of the dependence of the width of the diamond peak on the coordinate. The peak width is directly related to the size of diamond crystallites. It was more than 12 cm
−1, which is typical for nanocrystalline diamonds, at all points of the milling cutter. The maximum width of the diamond peak (~15.5 cm
−1) is reached at a distance of 10 mm from the end and the minimum (~13.0 cm
−1)—at a length of 30 mm. A similar pattern is observed for the width of the TPA strip at 11.4 cm
−1 (
Figure 10b), which is strictly reduced from ~50 cm
−1 at the end of the cutter to ~40 cm
−1 at a distance of 30 mm from the rear. Thus, the lower the substrate temperature, the larger, on average, the crystallites of the nanocrystalline diamond film.
The diamond film experiences elastic compression stresses at all points judging by the position of the diamond peak
Figure 10c. During the growth of the film, the milling cutter was heated, and when cooled after synthesis, it decreased in size; thus, the diamond coating was compressed. Judging by the dependence graph, the value of elastic stresses reaches a maximum at the end of the milling cutter since this area was the most heated during the synthesis process.
Figure 10d shows the ratio between the diamond peak areas and the G-band of disordered carbon. There is a tendency to decrease the proportion of the disordered phase compared to the diamond when moving away from the end of the cutter. It is especially clearly manifested in line 4. At a distance from the end, the temperature of the substrate was lower. As a result, the film growth was slower but with a higher percentage of diamond.
3.4. Morphology of the Diamond Film Surface
The morphology of the diamond film surface on a cylindrical MV sample with rotation is shown in
Figure 11. The average diamond grains sizes are presented in
Table 2. It was determined from SEM photographs using the Thixomet Pro (St Petersburg, Russia) automatic image analysis system for the microstructure of metals and alloys via quantitative metallography.
According to
Table 2, a graph of the dependence of the average grain size on DC film on the lines 0°–90°–180°–270° along the forming TM with a step of 10 mm was constructed (
Figure 12).
In all cases, the maximum grain sizes are observed on the cylinder sample with rotation (
Table 2) at a distance of 30 mm from the end (the average grain size at points 1–4, 2–4, 3–4, 4–4 d = 73 ± 4 nm); the minimum are observed at the end of the cutting edge (d = 51 ± 7 nm at points 1–1, 2–1, 3–1, 4–1); and between them we observed intermediate values: d = 58± 7 nm at a distance of 10 mm from the end at points 1–2, 2–2, 3–2, 4–2, and d = 62 ± 7 nm at a distance of 20 mm from the end at points 1–3, 2–3, 3–3, 4–3. The average grain size at a distance of 10 mm from the end is 13% larger than at the end; similarly, for 20 mm, it is 20%, and for 30 mm, it is 41%. The average grain sizes grow slower as they move away from the end face, which correlates with the local temperature of the sample surface or, in another way, with the distance from the center of the plasma cloud during diamond growth, where the maximum temperature is observed. In the case of a cylinder sample without rotation at the end of the cutting edge, both the average grain size and the size spread are smaller: d = 45 ± 3 nm (at similar points 1–1, 2–1, 3–1, 4–1), while there is no diamond in volume 3–3 on the cylinder without rotation. The phase is a nanocrystalline graphite film; in SEM, it is indistinguishable from a nanocrystalline diamond film.
Figure 13 shows the morphology of a diamond film formed at the end of a cylindrical tool model under conditions without rotation. The dependence of the average grain size on the angle of rotation when measuring TM at the end without rotation is presented in
Table 3.
According to
Table 3, a graph of the average grain size of the TM at the no-rotation mode is shown in
Figure 14. Here, the dimensions of the diamond grains at distances of 6 and 12 mm from the center of the TM end depending on the angles of location (0°, 90°, 180°, 270°) relevant to the points indicated are presented, according to
Table 3. The maximum value of the diamond grain size is 49 nm, and the face is located 12 mm from the center. The minimum value of the diamond grain is 32 nm and it is at a distance of 6 mm from the center of the end face. In the center of the rake face, the size of the diamond grain is 37 nm.
4. Conclusions
Mathematical modeling has determined that the temperature along the cylindrical tool model (TM) axis decreases from the free end to the one fixed in the holder inside the molybdenum ring. The TM body cools down especially sharply in the area near the holder.
It is possible to reduce the temperature difference along the axis by increasing the length of the free part of the TM since heat dissipation in the holder is much higher, and changing its design does not lead to a fundamental change in the heat exchange conditions. The departure of the TM part with a temperature of 700–800 °C, in which the growth of the diamond coating without rotation is possible in the ARDIS-100 type reactor, is 20–25 mm.
Rotation above 12 rpm during deposition makes it possible to compensate for the radial temperature unevenness on the TM surface and increase the growth of the diamond coating to 40 mm.
In all cases, in the carbide cylinder sample with rotation, the maximum grain sizes are observed at a distance of 30 mm from the end (the average grain size is 73 ± 4 nm); the minimum are observed at the back at the cutting edge, 51 ± 7 nm; between them, intermediate values are observed: 58 ± 7 nm at a distance of 10 mm from the rear, and 62 ± 7 nm at a distance of 20 mm from the rear.
The average grain size at a distance of 10 mm from the end is 13% larger than at the rear; similarly, at a distance of 20 mm, it is 20%, and a distance of 30 mm, it is 41%.
The average grain sizes grow slower as they move away from the end face, which correlates with the local temperature of the sample surface or with the distance from the center of the plasma cloud, where the maximum temperature is observed during diamond growth.
In the case of a sample of the carbide cylinder without rotation at the end, both the average grain size and the size spread are smaller—45 ± 3 nm. At the same time, there is no diamond phase on the underside of the cylinder. The phase is a nanocrystalline graphite film; however, in SEM, it is indistinguishable from a nanocrystalline diamond film.
The observed variation in the values of roughness and grain size can be attributed to the TM rotation speed 12 min−1, which was determined via mathematical modeling and required experimental adjustment.