Influence of Boriding Process in Adhesion of CVD Diamond Films on Tungsten Carbide Substrates

Sintered tungsten carbide (WC-Co) manufactured by powder metallurgy is widely used to produce cutting tools for the machining industry. CVD diamond coating is a great alternative to increase cutting tools lifetime for machining nonferrous metals. However, there are still fundamental technological issues to overcome concerning adhesion of the coating to substrate surface. The cobalt (Co) presence in WC-Co is of fundamental importance for the final mechanical properties of the tool, namely toughness, compressive strength and transverse rupture strength, which are fundamental requirements for cutting tools applications. However, cobalt chemically interacts with carbon atoms during chemical vapor deposition of diamond films due to its 3d incomplete sublevel, producing a graphitic phase1-3 on the substrate surface, inhibiting the direct deposition of CVD diamond films on WC-Co. Several pretreatments to avoid Co harmful effects have been tried, as shown in literature4-6. Boriding technique7 forms an intermediate barrier that blocks Co diffusion to the substrate surface, minimizing the Co binder negative effects for diamond growth. Boriding is one of the pretreatments used in this work. It consists on boron thermodiffusion from heated powders mixed according to pre-established mass proportions. As previously published8 the addition of CF4 in the gas phase promotes the formation of solid carbons, including diamond. Halogens, as atomic hydrogen, chemically react with graphitic phases, preferentially to react with diamond. The CF4 addition in the gas phase promotes selective etching of W from WC-Co matrix, forming a C rich surface, from where diamond nucleation starts. This “in situ” WC etching enhance diamond film adhesion. Previous studies suggest that the CF4 is effective to etch the cobalt binder phase from the WC-Co surface9. In this paper, we show clearly that brittlening is due to cobalt diffusion out of the substrate during the heating stage of boriding salt to process temperature. Additionally, our diamond growth experiments take advantage of the already known CF4 beneficial effects for growth on WC-Co substrates.


Introduction
Sintered tungsten carbide (WC-Co) manufactured by powder metallurgy is widely used to produce cutting tools for the machining industry.CVD diamond coating is a great alternative to increase cutting tools lifetime for machining nonferrous metals.However, there are still fundamental technological issues to overcome concerning adhesion of the coating to substrate surface.The cobalt (Co) presence in WC-Co is of fundamental importance for the final mechanical properties of the tool, namely toughness, compressive strength and transverse rupture strength, which are fundamental requirements for cutting tools applications.However, cobalt chemically interacts with carbon atoms during chemical vapor deposition of diamond films due to its 3d incomplete sublevel, producing a graphitic phase [1][2][3] on the substrate surface, inhibiting the direct deposition of CVD diamond films on WC-Co.Several pretreatments to avoid Co harmful effects have been tried, as shown in literature [4][5][6] .Boriding technique 7 forms an intermediate barrier that blocks Co diffusion to the substrate surface, minimizing the Co binder negative effects for diamond growth.Boriding is one of the pretreatments used in this work.It consists on boron thermodiffusion from heated powders mixed according to pre-established mass proportions.
As previously published 8 the addition of CF 4 in the gas phase promotes the formation of solid carbons, including diamond.Halogens, as atomic hydrogen, chemically react with graphitic phases, preferentially to react with diamond.The CF 4 addition in the gas phase promotes selective etching of W from WC-Co matrix, forming a C rich surface, from where diamond nucleation starts.This "in situ" WC etching enhance diamond film adhesion.Previous studies suggest that the CF 4 is effective to etch the cobalt binder phase from the WC-Co surface 9 .
In this paper, we show clearly that brittlening is due to cobalt diffusion out of the substrate during the heating stage of boriding salt to process temperature.Additionally, our diamond growth experiments take advantage of the already known CF 4 beneficial effects for growth on WC-Co substrates.reduce salt bath melting temperature.Graphite powder it added to balance the reaction.The boriding process was carefully controlled in all of the stages.Even though CoW 2 B 2 and CoWB stable Co phases work as an efficient barrier to Co diffusion, they also embrittles the substrate surface, mainly when using fine grain WC-Co substrates.In order to control surface brittleness, by controlling the thickness of those commonly formed brittle regions, we modified the traditional boriding process.Initially, we heated the powder mixture, in a stainless steel crucible, up to 1000 °C and kept at this temperature for 15 min.The sample was then placed into the crucible for the thermodiffusion process, with temperature kept constant at 1000 °C, for 5h.After this, the sample was removed at high temperature, with salt mixture still melted, to simplify handlings and excess powders cleaning.Sample preparation was completed with chemical etching 10 , according to Table 1.

Experimental
Chemical etching starts with Caro's acid followed by Murakami's alkaline etching.The acid etching removes cobalt residues 11,12 from the surface after heat treatment.The Murakami's solution specifically etches WC grains 13,14 to increase surface roughness.
Previous to the CVD diamond growth, the samples were pre-treated in an ultrasonic bath with a 0.25μm diamond powder suspension in n-hexane for 1h to increase diamond nucleation rate 15 .Diamond growth was performed in a hot filament reactor with six straight tungsten filaments, 125μm diameter.Filament temperature was kept at 2200 °C.All films were deposited for 5h in a gaseous atmosphere of methane (2%), carbon tetrafluoride CF 4 (2%) and hydrogen (96%).The total flow-rate was 100sscm and reactor pressure was 6.5 × 10 3 Pa [16] .
A tungsten resistance placed in the substrate holder was used to control substrates temperature at chosen level, in the range of 700 °C to 900 °C.Adhesion tests of diamond films were performed in a Rockwell C indenter [17][18][19] .Applied loads ranged from 300 N to 700 N. Gradual variation of loads allowed measurement of critical loads supported by each diamond film 20 .
Samples were characterized by Scanning Electron Microscopy (SEM), Energy Dispersive X-ray (EDX), X-ray Diffraction (XRD) and Raman Scattering Spectroscopy (RSS).Residual stresses in diamond films were evaluated considering the deviation from the Raman shift of stress-free diamond, of 1332 cm -1 .Raman spectroscopy was performed with Renishaw System 2000, using argon laser excitation (514.5nm).Microscopic evaluation of surface morphology and interface were performed with Jeol (JSM-5310) Scanning Electron Microscope (SEM).An energy dispersive X-ray detector attached to the SEM enabled evaluation of the elemental distribution in the interface (EDS).A Philips X'Pert MRD X-ray Diffractometer was used to evaluate phases in the substrate surface before and after boriding.

Results and Discussion
Figures 1a and 1b show X-ray diffractograms of the WC-Co sample, prior and after the boriding process, respectively.As it can be noticed on Figure 1a, only WC and Co are present in the substrate before boriding, while other phase as CoW 2 B 2 , CoWB and Co 2 B form after boriding treatment.Small fractions of Co 2 B were formed due to free Co binder that migrated to surface from bulk WC-Co.
Our early experiments 2,21 , as other previous reports 22 , showed that a brittle region formed in the substrate surface during salt powders heating stage to the boriding process temperature.Edwards et al. 21used WC-TiC-Co samples with medium grain size of nearly 3μm and did not report this brittle region.However, this brittle region always formed in our experiments with 0.8μm grain size WC-Co substrates 23 .
Figure 2 shows this brittle region after traditional boriding.A WC skeleton without any binder is shown in the figure, in the region indicated as "a".A boride interlayer formed under this skeleton, indicated as region "b" in the figure.This brittle skeleton, clearly seen in Figure 2, is  formed due to cobalt migration to the surface of the sample during the heating stage, before reaching the boriding process temperature.The sample shown in Figure 2 has a large brittle region.However, for samples with small WC grain, as those used in this work, even a very thin brittle interface is unfavorable for tool applications, resulting in fragile coating.Indentation tests confirmed the weakness of the resulting coating -diamond films/boride interlayer as shown in the Figure 3, where the delamination of the 6μm thick diamond film around indentation mark can be seen.This round broad delamination is typical for low adherent films.As previously stated, the low adherence is due to the brittle interface formed in the region with a WC skeleton.
The stable CoW 2 B 2 and CoWB cobalt compounds were not formed in this region, probably due to Co thermal migration to substrate surface.Figure 4 shows the result of our controlled boriding method.Insertion of samples after preheating powders prevented the formation of a brittle region in the boride interlayer.The boride interface starts at the surface without any brittle region.Preheating the powders doesn't affect CoW 2 B 2 and CoWB phases formation.These are the main phases responsible for the boriding barrier.Further, they minimize the presence of unreacted Co on surface.
Even with small amounts of free Co in surface, we further improved the interface by chemical etching.Caro's solution removed cobalt residues and Murakami's solution reacts with WC to get a rougher surface.Figure 5a shows the surface morphology after chemical attack with Caro's and Murakami's reagents.Figure 5b shows a SEM image of diamond film deposited on WC-Co substrate with a boride interlayer.The deposited diamond films had an average thickness of 6μm for the typical 5h growth period.
The diamond film adhesion on the WC-Co substrate significantly increased with the controlled boring process.Figure 6 shows indentation imprints on these samples.No lateral cracks formed at loads up to 700N using the Rockwell C indenter.Figures 6a shows the imprint for 300N load, with a slight crack formed on the CVD diamond film on the sample.Figure 6b shows de imprint for 500N.The diamond film breaks up concentrically around the indentation mark due to boriding interlayer and substrate deformation with load increasing and without peeling off the diamond film.Even at loads of 600N and 700N (Figures 6c and 6d), the diamond films showed similar mechanical behaviors to the one shown in Figure 6b.At critical load of 700N, the diamond film collapsed, as shown in Figure 6d.The good diamond film adhesion demonstrated on Figure 6 suggests that brittle regions on the substrate surface were not formed in the boriding process.During Rockwell C indentation tests, at 600N load, due to flaws on substrate base, a sample broke into two pieces without peeling off, as shown in Figure 7.This is a relevant fact that can be correlated to good adhesion of the diamond film to the substrate surface.This also indicates high interface adhesion for all interfaces -substrate to boride interlayer and boride to diamond film.
Figure 8 shows the Raman spectrum for the diamond film deposited on WC-Co substrate with boride interlayer.Residual stresses in diamond films, typically compressive due to the smaller thermal expansion coefficient of diamond in comparison to WC-Co substrate 24,25 , can be evaluated from Raman peak displacement relative to natural unstressed diamond 26,27 .In this case, the peak is near to 1339 cm -1 , corresponding to a displacement of 7.1 cm -1 from the natural diamond peak at 1332 cm -1 wavenumber.This leads to calculated compressive stress of 4.0GPa in the deposited diamond film.

Conclusion
In this work we have shown that the initial stage of the boriding process is critical to get strong interfaces to sub-micrometric grain size WC-Co substrates.In this stage Co migrates out the WC-Co as the boriding bath is heated up, producing a fragile surface.Inserting the sample at the final boriding temperature, when the bath is already melted and stable, minimizes this brittlening effect and the brittle region disappears.The acid etching after boriding cleans the surface from residual free cobalt.The combined etching effects of the alkaline Murakami solution and the "in situ" CF 4 gas are effective to increase WC surface roughness to improve diamond nucleation.The surface roughness of the substrate also enables formation of better interlocking and hence higher adhesion strength.The association of these good conditions with a tough boride interlayer enabled the deposition of good quality and adherent diamond films on WC-Co substrates, as shown by Raman spectroscopy and indentation tests.

Figure 1 .
Figure 1.(a) X-ray diffractograms of the WC-Co sample prior to the boriding process, (b) after controlled boriding process.

Figure 2 .
Figure 2. SEM image of the cross section of WC-Co substrate after traditional boriding process.Detail (a) shows the brittle region and (b) the boride interlayer that blocks Co diffusion to surface.

Figure 3 .
Figure 3. SEM image of the diamond film deposited on WC-Co substrate with boride interlayer, loaded at 600N using a Rockwell C indenter.

Figure 4 .
Figure 4. SEM image of the cross section of WC-Co substrate after controlled boriding, with samples immersed on powders after heating up to process temperature.Detail (c) shows the boride interlayer without the brittle region.

Figure 5 .
Figure 5. (a) SEM image of the substrate surface etched and (b) diamond film deposited on the WC-Co with the controlled boriding layer.

Table 1 .
Chemical etching parameters used in WC-Co.