Analysis of nanostructured coatings synthesized by ballistic impaction ...
Thin Solid Films 515 (2006) 1147 – 1151 www.elsevier.com/locate/tsf
Analysis of nanostructured coatings synthesized by ballistic impaction of nanoparticles J. Hafiz ⁎, R. Mukherjee, X. Wang, J.V.R. Heberlein, P.H. McMurry, S.L. Girshick Department of Mechanical Engineering, University of Minnesota, 111 Church St. SE, Minneapolis, MN 55455, USA Available online 7 September 2006
Abstract SiC nanostructured coatings were synthesized by ballistic impaction of nanoparticles using a process called hypersonic plasma particle deposition (HPPD). X-ray diffraction spectra of typical samples showed the presence of crystalline SiC and Si. Grain sizes obtained through transmission electron microscopy showed particles in the sub 10 nm range with primarily crystalline β-SiC and some crystalline Si particles present. These results correlate well with particle size distributions measured using an aerosol sampling probe coupled to a scanning electrical mobility spectrometer. Interestingly, particle size distributions indicated only small changes in the particle size distributions when Si deposition was compared to SiC. Examination of adhesion characteristics highlighted the importance of a chemically bound interlayer during SiC deposition on Mo and steel substrates. © 2006 Elsevier B.V. All rights reserved. Keywords: Cubic SiC; Nanoparticle; Inertial impaction; Size distribution; Aerosol sampling probe
1. Introduction A vast and rapidly growing body of research has demonstrated that nanostructured coatings and nanophase materials can exhibit extraordinary properties—mechanical, electronic, magnetic, optical, catalytic, and others. This paper describes the synthesis and in situ assembly of nanostructured coatings using a process known as “hypersonic plasma particle deposition” (HPPD) [1,2]. Nanoparticles are synthesized by injecting gasphase reactants into a thermal plasma that undergoes a rapid expansion through a nozzle, driving nucleation. These particles are then deposited as a spray, after acceleration in a hypersonic expansion, causing them to ballistically impact a substrate at velocities up to ∼2 km/s, forming a dense nanoparticle film. Typical deposition rates of the coatings are in the range of 2– 30 μm/min. Numerical simulations indicate that the cut size for inertial impaction in this process is 3–4 nm [3]. The HPPD process is amenable to scale-up for industrial production as existing and widely-used thermal plasma technology is utilized. In this paper, results are presented on the synthesis of SiC nanostructured coatings using the HPPD process. The structure, composition and morphology of the coatings were analyzed, ⁎ Corresponding author. Tel.: +1 612 626 8605; fax: +1 612 625 4344. E-mail address: [email protected] (J. Hafiz). 0040-6090/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2006.07.121
and the effect of substrate material on coating adhesion was examined. In addition, the properties of individual nanoparticles that form the coatings were studied. Typical grain sizes of individual particles were compared with in situ size distributions measured by a nanoparticle aerosol diagnostic system [4]. Particle diagnostics provided insight into the mechanism of particle formation and growth, and may enable optimization of operating conditions to achieve desired film properties. 2. Experimental details 2.1. Deposition process Fig. 1 shows a schematic of the experimental system. The plasma is generated by a DC arc, operating at ∼200 A current and 8–10 kW power. The main plasma gases are argon, with a flow rate of 30–35 slm, and hydrogen, with a flow rate of 2–6 slm. The deposition rates of the coatings varied from 2 to 30 μm/min, depending on reactant flow rates. Vapor-phase reactants were injected into the plasma through an injection ring. Chloride vapors of Si were injected using a bubbler system, with flow rates in the range 20–80 sccm, while carbon was introduced using methane. Molybdenum and high speed steel (HSS) were used as substrate material. Typical substrate temperature during deposition ranged from 750 to 1000 °C. Experiments were conducted with
1148
J. Hafiz et al. / Thin Solid Films 515 (2006) 1147–1151
Fig. 1. Schematic of the HPPD system.
a copper gettering system to reduce oxygen contamination, which is believed to have a detrimental effect on film hardness [5].
Model 545 Auger electron spectrometer (AES) was utilized to obtain compositional depth profiles of the coatings.
2.2. Characterization techniques
3. Results and discussion
In situ particle size distribution measurements were performed using a sampling probe interfaced to an extraction/ dilution system in series with a scanning electrical mobility spectrometer (SEMS) [4]. During measurements a water-cooled sampling probe is inserted at the same location as the substrate for spray deposition. The aerosol sample is extracted and rapidly diluted by an ejection diluter. Particle size distributions are then measured by the SEMS system. The nanostructured coatings were investigated using various characterization tools. For high resolution imaging a Tecnai G2 F30 FEG transmission electron microscope (TEM) was used. X-ray diffraction (XRD) spectra were obtained with a Bruker AXS microdiffractometer (CuKα). A Physical Electronics
3.1. Nanostructured SiC coatings An XRD spectrum (Fig. 2) of a typical SiC coating deposited at 860 °C (SiCl4 = 40 sccm, CH4 = 280 sccm) on a molybdenum substrate shows the presence of crystalline β-SiC and crystalline Si. A corresponding depth profile obtained by Auger spectrometry is shown in Fig. 3. Since AES is a surface sensitive technique, it is possible to determine the distribution of oxides on the surface as well as the composition of the core coating by repeatedly sputtering away the topmost layers of the sample and then analyzing the freshly exposed surface. Experimental data from the depth analysis provide peaks in Si, C and O concentrations as a function of sputter time, where a
Fig. 2. X-ray diffraction spectrum of a typical SiC coating deposited at 860 °C.
Fig. 3. Depth profile of a SiC coating using Auger electron spectrometry. A minute of sputtering corresponds approximately to a depth of 2 nm of SiC.
J. Hafiz et al. / Thin Solid Films 515 (2006) 1147–1151
Fig. 4. TEM image of β-SiC nanoparticles on a holey carbon grid taken at an accelerating voltage of 300 kV (the scale bar is 10 nm). Top inset is a HRTEM image of one of the particles, which exhibits lattice fringes corresponding to βSiC.
minute of sputtering corresponds approximately to a depth of 2 nm of SiC. It is observed that for both C and Si the atomic concentration reaches a stable state after 20 min of sputtering. On the other hand, oxygen decreases to negligible levels after what is presumed to be a surface oxide layer is sputtered away. This result is consistent with our expectation that the in situ consolidation possible with the HPPD process should restrict oxygen diffusion to the core of the deposit. The characteristics of the synthesized particles are of primary importance in the HPPD process, as the mechanical properties of nanostructured coatings are directly related to particle size. For analysis of individual nanoparticles, TEM samples were prepared by scraping material from the coating into an alcohol solution, which was then repeatedly agitated using an ultrasonic cleaner. After sonication, a few drops of the solution were deposited onto holey carbon TEM grids for analysis purposes. Fig. 4 shows a TEM image of the deposited nanoparticles. It appears that the sizes of the particles are in the sub 10 nm range
Fig. 5. In-situ size distribution measurement for SiC with reactant flow rates.
1149
and relatively uniform. The top inset in Fig. 4 shows a high resolution TEM image of one of the particles, which exhibits lattice fringes corresponding to the (200) plane of β-SiC. Further analysis reveals that the particles are primarily SiC with some Si particles also present. The sizes observed in the TEM image can be compared to particle sizes obtained using the in situ size distribution measurement tool. Measurements performed with the particle sampling probe show the mode of the size distribution to equal 4–5 nm (Fig. 5). Overall, the size distribution of particles in the TEM image corresponds well to that obtained by the SEMS system. It is important to determine the effect of free Si on the mechanical properties of the synthesized coatings. Hardness measurements performed on the SiC coatings by nanoindentation provided an average value of 22 GPa. While this is considerably harder than bulk Si it is softer than commercial polycrystalline SiC. Nanocrystalline Si coatings deposited by our process have yielded a hardness value of 12 GPa, while commercial SiC has hardness values ranging from 26 to 34 GPa. It appears that our coating hardness is affected by the presence of both the Si and the SiC phase, with the measured value in between the Si and SiC hardness numbers. We hypothesize based on these results that eliminating the Si phase might increase the hardness of the SiC coatings to at least that of bulk SiC. Also, given the nanostructured character of our coating we may be able to achieve hardness values higher than bulk SiC if only crystalline SiC is present. This issue is discussed further in the following section. 3.2. Size distribution comparison The SEMS experiments revealed an interesting phenomenon when we compared the size distribution of SiC to the case where we were synthesizing only nanocrystalline Si. Fig. 6 shows a comparison of the size distribution for the same Si reactant flow rate, while CH4 was added for SiC. It can be observed that the mode of the size distribution as well as the overall trend hardly changes, even though for the SiC case we added a significant
Fig. 6. Comparison of size distribution between Si and SiC for the same Si flow rate, while CH4 is added during SiC deposition.
1150
J. Hafiz et al. / Thin Solid Films 515 (2006) 1147–1151
quantity of methane. The slight differences seen in the magnitudes of the particle concentrations can be attributed to inherent variations in day to day running of the experiments. In earlier work with subsonic plasma expansion reactors we had hypothesized that SiC particle formation is initiated by the nucleation of silicon particles which are subsequently carburized in the hydrocarbon rich environment [6]. Similarities in the size distribution between Si and SiC lend credence to this view. More importantly, this suggests that a possible method to eliminate elemental Si may be increasing the ratio of C to Si. However, experiments performed with much higher C to Si ratios still showed the presence of elemental Si. The only noticeable difference was an increase in amorphous material, probably amorphous carbon and amorphous SiC, as evidenced by a representative XRD image (Fig 7). Due to the non-equilibrium conditions present during primary particle growth in the nozzle it is likely that the solution to synthesizing only SiC will depend on a combination of factors, such as H2 concentration and reactant flow rates. The effect of H2 concentration in obtaining crystalline SiC has been reported in the literature [7,8]. It has been shown that increased atomic hydrogen content aids crystallization and selective etching of the amorphous phase during deposition. As previously mentioned, eliminating Si could have a positive effect on the hardness values of the synthesized coatings. Determination of the operational window that would allow us to reproducibly deposit solely SiC is currently ongoing. 3.3. Adhesion characteristics The adhesion of a coating to the substrate is of great concern in ensuring the reliability of coatings. By examining the adhesion characteristics between the substrate material and nanostructured SiC, one can obtain insights regarding how to improve overall coating properties. For deposition on molybdenum substrates, adhesion was not a problem, due to the formation of a chemically bonded Mo2C interlayer. For coatings thinner than 3 μm, XRD was able to detect a crystalline Mo2C interlayer as shown in Fig. 8 (the Mo
Fig. 7. X-ray diffraction spectrum of a SiC coating deposited with high CH4 flow illustrating the amorphous hump.
Fig. 8. X-ray diffraction spectrum of a thin SiC coating deposited on a Mo substrate showing the presence of crystalline Mo2C.
substrate is also detected). This interlayer, presumably formed by interdiffusion between the molybdenum substrate and carbon atoms from either impacting particles or adsorbed gasphase species, is expected to enhance adhesion through chemical bonding [9]. For films thicker than 3 μm, XRD analysis of films deposited at high temperatures was not able to detect crystalline molybdenum due to the limited penetration depth of the probe beam (∼ 3 μm). In contrast to Mo, silicon carbide coatings deposited on steel substrates usually displayed poor adhesion. There exist a couple of major obstacles for depositing SiC on ferrous metals [10]. First, iron has a catalytic effect on the growth of sp2 dominated amorphous and nanocrystalline carbon, which is also known as carbon black. Carbon black can form if the carbon reactant species are not completely consumed by gas-to-particle conversion in the upstream nanoparticle synthesis process, thereby exposing the substrate to high fluxes of monoatomic carbon vapor, which would be expected to be in the form a soft carbon layer [10]. In many such cases SiC films are actually grown on a layer of soft carbon black instead of carburized iron. Second, the
Fig. 9. X-ray diffraction spectrum of a thin SiC coating deposited on a steel substrate showing the presence of Fe2N.
J. Hafiz et al. / Thin Solid Films 515 (2006) 1147–1151
Fig. 10. AES survey of the bottom surface of a SiC coating deposited on a steel substrate showing the presence of N and Fe diffused into the coating.
thermal expansion coefficients of SiC and ferrous metals are not compatible and this mismatch usually causes poor adhesion and high residual stresses. As a consequence, all as-deposited SiC coatings on steel substrates peeled off soon after deposition. These results suggest the use of an interlayer between the deposited SiC film and the iron-based substrate. There have been reports in the literature of utilizing a nitriding pretreatment for enhancing the adhesion of hard coatings deposited on steel substrates [10–12]. For the case of SiC, nitriding prior to nanoparticle deposition results in the formation of an iron nitride interlayer, which improves adhesion of the film to the substrate. Fortunately, due to the flexibility of gaseous precursor injection in the HPPD process, the nitriding pretreatment could be performed in situ without any major modifications to the deposition process. By using a plasma gas mixture of Ar + H2 + N2 (instead of an Ar + H2 mixture) before reactant injection we were able to pretreat the substrate. The N2 gas was shut off prior to reactant injection for SiC deposition. The coatings that resulted from this treatment were strongly adhered to the steel substrate. The corresponding XRD spectrum (Fig. 9) of a thin coating highlights the crystalline Fe2N interlayer that forms due to the nitriding treatment. We also forcibly peeled off a coating for additional analysis of the interface between the substrate and the coating. An AES survey of the bottom surface of the coating confirms the presence of N and Fe diffused into the coating (Fig. 10). AES surveys contain information from only the top 1–2 nm of the coating. This is the reason behind the strong oxygen signal in Fig. 10. Qualitatively, the nitriding pretreatment was successful, as we were only able to delaminate the coatings by extensive polishing with alumina paste. Present investigations are focused on quantitative determination of the adhesive properties of the coatings using a pin-on-disk tribometer. 4. Conclusions SiC nanostructured coatings deposited by high velocity impaction of nanosized particles were investigated. X-ray diffraction spectra of typical samples showed the presence of
1151
crystalline SiC and Si. Grain sizes obtained through transmission electron microscopy showed particles in the sub 10 nm range with primarily crystalline β-SiC and some crystalline Si particles present. These results correlate well with particle size distributions measured using an aerosol sampling probe coupled to a scanning electrical mobility spectrometer. Interestingly, there were only small changes in the particle size distributions when Si deposition was compared to SiC. The aerosol probe measurements demonstrated a possible formation mechanism of SiC in the HPPD system. Examination of adhesion characteristics highlighted the importance of a chemically bound interlayer during SiC deposition in Mo and steel substrates. All these results taken together demonstrate the potential of the HPPD process for high-volume production of nanostructured coatings, while avoiding many of the problems associated with conventional powder processing routes. However, further work is needed to develop procedures to ensure formation of only crystalline SiC deposits. Acknowledgements This work was partially supported by the United States National Science Foundation, under award numbers DMI0103169 and CTS-0506748, and IGERT award number DGE0114372. The assistance of Joysurya Basu of the Characterization Facility during TEM operation, and John Thomas during Auger electron spectroscopy operation is gratefully acknowledged. We would also like to thank William Mook for the nanoindenation measurements. References [1] S.L. Girshick, J.V.R. Heberlein, P.H. McMurry, W.W. Gerberich, D.I. Iordanoglou, N.P. Rao, A. Gidwani, N. Tymiak, F.D. Fonzo, M.H. Fan, D. Neumann, in: K.L. Choy (Ed.), Innovative Processing of Films and Nanocrystalline Powders, Imperial Press, London, 2002, p. 165. [2] J. Hafiz, X. Wang, R. Mukherjee, W. Mook, C.R. Perrey, J. Deneen, J.V.R. Heberlein, P.H. McMurry, W.W. Gerberich, C.B. Carter, S.L. Girshick, Surf. Coat. Technol. 188–189 (2004) 364. [3] A. Gidwani, Studies of Flow and Particle Transport in Hypersonic Plasma Particle Deposition and Aerodynamic Focusing, University of Minnesota, Minneapolis, 2003. [4] X. Wang, J. Hafiz, R. Mukherjee, T. Renault, J. Heberlein, S.L. Girshick, P.H. McMurry, Plasma Chem. Plasma Process. 25 (2005) 439. [5] M. Dayan, M. Shengli, X. Kewei, S. Veprek, Mater. Lett. 59 (2005) 838. [6] N. Rao, S.L. Girshick, J. Heberlein, P.H. McMurry, S. Jones, D. Hansen, B. Micheel, Plasma Chem. Plasma Process. 15 (1995) 581. [7] Q. Zhao, J.C. Li, H. Zhou, H. Wang, B. Wang, H. Yan, J. Cryst. Growth 260 (2004) 176. [8] C. Summonte, R. Rizzoli, M. Servidori, S. Milita, S. Nicoletti, M. Bianconi, J. Appl. Phys. 96 (2004) 3998. [9] D.F. Bahr, D.V. Bucci, L.S. Schadler, J.A. Last, J. Heberlein, E. Pfender, W.W. Gerberich, Diamond Relat. Mater. 5 (1996) 1462. [10] S. Abisset, F. Maury, R. Feurer, M. Ducarroir, M. Nadal, M. Andrieux, Thin Solid Films 315 (1998) 179. [11] C.F.M. Borges, E. Pfender, J. Heberlein, Diamond Relat. Mater. 10 (2001) 1983. [12] A. da Silva Rocha, T. Strohaecker, V. Tomala, T. Hirsch, Surf. Coat. Technol. 115 (1999) 24.
Analysis of nanostructured coatings synthesized by ballistic impaction of nanoparticles J. Hafiz ⁎, R. Mukherjee, X. Wang, J.V.R. Heberlein, P.H. McMurry, S.L. Girshick Department of Mechanical Engineering, University of Minnesota, 111 Church St. SE, Minneapolis, MN 55455, USA Available online 7 September 2006
Abstract SiC nanostructured coatings were synthesized by ballistic impaction of nanoparticles using a process called hypersonic plasma particle deposition (HPPD). X-ray diffraction spectra of typical samples showed the presence of crystalline SiC and Si. Grain sizes obtained through transmission electron microscopy showed particles in the sub 10 nm range with primarily crystalline β-SiC and some crystalline Si particles present. These results correlate well with particle size distributions measured using an aerosol sampling probe coupled to a scanning electrical mobility spectrometer. Interestingly, particle size distributions indicated only small changes in the particle size distributions when Si deposition was compared to SiC. Examination of adhesion characteristics highlighted the importance of a chemically bound interlayer during SiC deposition on Mo and steel substrates. © 2006 Elsevier B.V. All rights reserved. Keywords: Cubic SiC; Nanoparticle; Inertial impaction; Size distribution; Aerosol sampling probe
1. Introduction A vast and rapidly growing body of research has demonstrated that nanostructured coatings and nanophase materials can exhibit extraordinary properties—mechanical, electronic, magnetic, optical, catalytic, and others. This paper describes the synthesis and in situ assembly of nanostructured coatings using a process known as “hypersonic plasma particle deposition” (HPPD) [1,2]. Nanoparticles are synthesized by injecting gasphase reactants into a thermal plasma that undergoes a rapid expansion through a nozzle, driving nucleation. These particles are then deposited as a spray, after acceleration in a hypersonic expansion, causing them to ballistically impact a substrate at velocities up to ∼2 km/s, forming a dense nanoparticle film. Typical deposition rates of the coatings are in the range of 2– 30 μm/min. Numerical simulations indicate that the cut size for inertial impaction in this process is 3–4 nm [3]. The HPPD process is amenable to scale-up for industrial production as existing and widely-used thermal plasma technology is utilized. In this paper, results are presented on the synthesis of SiC nanostructured coatings using the HPPD process. The structure, composition and morphology of the coatings were analyzed, ⁎ Corresponding author. Tel.: +1 612 626 8605; fax: +1 612 625 4344. E-mail address: [email protected] (J. Hafiz). 0040-6090/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2006.07.121
and the effect of substrate material on coating adhesion was examined. In addition, the properties of individual nanoparticles that form the coatings were studied. Typical grain sizes of individual particles were compared with in situ size distributions measured by a nanoparticle aerosol diagnostic system [4]. Particle diagnostics provided insight into the mechanism of particle formation and growth, and may enable optimization of operating conditions to achieve desired film properties. 2. Experimental details 2.1. Deposition process Fig. 1 shows a schematic of the experimental system. The plasma is generated by a DC arc, operating at ∼200 A current and 8–10 kW power. The main plasma gases are argon, with a flow rate of 30–35 slm, and hydrogen, with a flow rate of 2–6 slm. The deposition rates of the coatings varied from 2 to 30 μm/min, depending on reactant flow rates. Vapor-phase reactants were injected into the plasma through an injection ring. Chloride vapors of Si were injected using a bubbler system, with flow rates in the range 20–80 sccm, while carbon was introduced using methane. Molybdenum and high speed steel (HSS) were used as substrate material. Typical substrate temperature during deposition ranged from 750 to 1000 °C. Experiments were conducted with
1148
J. Hafiz et al. / Thin Solid Films 515 (2006) 1147–1151
Fig. 1. Schematic of the HPPD system.
a copper gettering system to reduce oxygen contamination, which is believed to have a detrimental effect on film hardness [5].
Model 545 Auger electron spectrometer (AES) was utilized to obtain compositional depth profiles of the coatings.
2.2. Characterization techniques
3. Results and discussion
In situ particle size distribution measurements were performed using a sampling probe interfaced to an extraction/ dilution system in series with a scanning electrical mobility spectrometer (SEMS) [4]. During measurements a water-cooled sampling probe is inserted at the same location as the substrate for spray deposition. The aerosol sample is extracted and rapidly diluted by an ejection diluter. Particle size distributions are then measured by the SEMS system. The nanostructured coatings were investigated using various characterization tools. For high resolution imaging a Tecnai G2 F30 FEG transmission electron microscope (TEM) was used. X-ray diffraction (XRD) spectra were obtained with a Bruker AXS microdiffractometer (CuKα). A Physical Electronics
3.1. Nanostructured SiC coatings An XRD spectrum (Fig. 2) of a typical SiC coating deposited at 860 °C (SiCl4 = 40 sccm, CH4 = 280 sccm) on a molybdenum substrate shows the presence of crystalline β-SiC and crystalline Si. A corresponding depth profile obtained by Auger spectrometry is shown in Fig. 3. Since AES is a surface sensitive technique, it is possible to determine the distribution of oxides on the surface as well as the composition of the core coating by repeatedly sputtering away the topmost layers of the sample and then analyzing the freshly exposed surface. Experimental data from the depth analysis provide peaks in Si, C and O concentrations as a function of sputter time, where a
Fig. 2. X-ray diffraction spectrum of a typical SiC coating deposited at 860 °C.
Fig. 3. Depth profile of a SiC coating using Auger electron spectrometry. A minute of sputtering corresponds approximately to a depth of 2 nm of SiC.
J. Hafiz et al. / Thin Solid Films 515 (2006) 1147–1151
Fig. 4. TEM image of β-SiC nanoparticles on a holey carbon grid taken at an accelerating voltage of 300 kV (the scale bar is 10 nm). Top inset is a HRTEM image of one of the particles, which exhibits lattice fringes corresponding to βSiC.
minute of sputtering corresponds approximately to a depth of 2 nm of SiC. It is observed that for both C and Si the atomic concentration reaches a stable state after 20 min of sputtering. On the other hand, oxygen decreases to negligible levels after what is presumed to be a surface oxide layer is sputtered away. This result is consistent with our expectation that the in situ consolidation possible with the HPPD process should restrict oxygen diffusion to the core of the deposit. The characteristics of the synthesized particles are of primary importance in the HPPD process, as the mechanical properties of nanostructured coatings are directly related to particle size. For analysis of individual nanoparticles, TEM samples were prepared by scraping material from the coating into an alcohol solution, which was then repeatedly agitated using an ultrasonic cleaner. After sonication, a few drops of the solution were deposited onto holey carbon TEM grids for analysis purposes. Fig. 4 shows a TEM image of the deposited nanoparticles. It appears that the sizes of the particles are in the sub 10 nm range
Fig. 5. In-situ size distribution measurement for SiC with reactant flow rates.
1149
and relatively uniform. The top inset in Fig. 4 shows a high resolution TEM image of one of the particles, which exhibits lattice fringes corresponding to the (200) plane of β-SiC. Further analysis reveals that the particles are primarily SiC with some Si particles also present. The sizes observed in the TEM image can be compared to particle sizes obtained using the in situ size distribution measurement tool. Measurements performed with the particle sampling probe show the mode of the size distribution to equal 4–5 nm (Fig. 5). Overall, the size distribution of particles in the TEM image corresponds well to that obtained by the SEMS system. It is important to determine the effect of free Si on the mechanical properties of the synthesized coatings. Hardness measurements performed on the SiC coatings by nanoindentation provided an average value of 22 GPa. While this is considerably harder than bulk Si it is softer than commercial polycrystalline SiC. Nanocrystalline Si coatings deposited by our process have yielded a hardness value of 12 GPa, while commercial SiC has hardness values ranging from 26 to 34 GPa. It appears that our coating hardness is affected by the presence of both the Si and the SiC phase, with the measured value in between the Si and SiC hardness numbers. We hypothesize based on these results that eliminating the Si phase might increase the hardness of the SiC coatings to at least that of bulk SiC. Also, given the nanostructured character of our coating we may be able to achieve hardness values higher than bulk SiC if only crystalline SiC is present. This issue is discussed further in the following section. 3.2. Size distribution comparison The SEMS experiments revealed an interesting phenomenon when we compared the size distribution of SiC to the case where we were synthesizing only nanocrystalline Si. Fig. 6 shows a comparison of the size distribution for the same Si reactant flow rate, while CH4 was added for SiC. It can be observed that the mode of the size distribution as well as the overall trend hardly changes, even though for the SiC case we added a significant
Fig. 6. Comparison of size distribution between Si and SiC for the same Si flow rate, while CH4 is added during SiC deposition.
1150
J. Hafiz et al. / Thin Solid Films 515 (2006) 1147–1151
quantity of methane. The slight differences seen in the magnitudes of the particle concentrations can be attributed to inherent variations in day to day running of the experiments. In earlier work with subsonic plasma expansion reactors we had hypothesized that SiC particle formation is initiated by the nucleation of silicon particles which are subsequently carburized in the hydrocarbon rich environment [6]. Similarities in the size distribution between Si and SiC lend credence to this view. More importantly, this suggests that a possible method to eliminate elemental Si may be increasing the ratio of C to Si. However, experiments performed with much higher C to Si ratios still showed the presence of elemental Si. The only noticeable difference was an increase in amorphous material, probably amorphous carbon and amorphous SiC, as evidenced by a representative XRD image (Fig 7). Due to the non-equilibrium conditions present during primary particle growth in the nozzle it is likely that the solution to synthesizing only SiC will depend on a combination of factors, such as H2 concentration and reactant flow rates. The effect of H2 concentration in obtaining crystalline SiC has been reported in the literature [7,8]. It has been shown that increased atomic hydrogen content aids crystallization and selective etching of the amorphous phase during deposition. As previously mentioned, eliminating Si could have a positive effect on the hardness values of the synthesized coatings. Determination of the operational window that would allow us to reproducibly deposit solely SiC is currently ongoing. 3.3. Adhesion characteristics The adhesion of a coating to the substrate is of great concern in ensuring the reliability of coatings. By examining the adhesion characteristics between the substrate material and nanostructured SiC, one can obtain insights regarding how to improve overall coating properties. For deposition on molybdenum substrates, adhesion was not a problem, due to the formation of a chemically bonded Mo2C interlayer. For coatings thinner than 3 μm, XRD was able to detect a crystalline Mo2C interlayer as shown in Fig. 8 (the Mo
Fig. 7. X-ray diffraction spectrum of a SiC coating deposited with high CH4 flow illustrating the amorphous hump.
Fig. 8. X-ray diffraction spectrum of a thin SiC coating deposited on a Mo substrate showing the presence of crystalline Mo2C.
substrate is also detected). This interlayer, presumably formed by interdiffusion between the molybdenum substrate and carbon atoms from either impacting particles or adsorbed gasphase species, is expected to enhance adhesion through chemical bonding [9]. For films thicker than 3 μm, XRD analysis of films deposited at high temperatures was not able to detect crystalline molybdenum due to the limited penetration depth of the probe beam (∼ 3 μm). In contrast to Mo, silicon carbide coatings deposited on steel substrates usually displayed poor adhesion. There exist a couple of major obstacles for depositing SiC on ferrous metals [10]. First, iron has a catalytic effect on the growth of sp2 dominated amorphous and nanocrystalline carbon, which is also known as carbon black. Carbon black can form if the carbon reactant species are not completely consumed by gas-to-particle conversion in the upstream nanoparticle synthesis process, thereby exposing the substrate to high fluxes of monoatomic carbon vapor, which would be expected to be in the form a soft carbon layer [10]. In many such cases SiC films are actually grown on a layer of soft carbon black instead of carburized iron. Second, the
Fig. 9. X-ray diffraction spectrum of a thin SiC coating deposited on a steel substrate showing the presence of Fe2N.
J. Hafiz et al. / Thin Solid Films 515 (2006) 1147–1151
Fig. 10. AES survey of the bottom surface of a SiC coating deposited on a steel substrate showing the presence of N and Fe diffused into the coating.
thermal expansion coefficients of SiC and ferrous metals are not compatible and this mismatch usually causes poor adhesion and high residual stresses. As a consequence, all as-deposited SiC coatings on steel substrates peeled off soon after deposition. These results suggest the use of an interlayer between the deposited SiC film and the iron-based substrate. There have been reports in the literature of utilizing a nitriding pretreatment for enhancing the adhesion of hard coatings deposited on steel substrates [10–12]. For the case of SiC, nitriding prior to nanoparticle deposition results in the formation of an iron nitride interlayer, which improves adhesion of the film to the substrate. Fortunately, due to the flexibility of gaseous precursor injection in the HPPD process, the nitriding pretreatment could be performed in situ without any major modifications to the deposition process. By using a plasma gas mixture of Ar + H2 + N2 (instead of an Ar + H2 mixture) before reactant injection we were able to pretreat the substrate. The N2 gas was shut off prior to reactant injection for SiC deposition. The coatings that resulted from this treatment were strongly adhered to the steel substrate. The corresponding XRD spectrum (Fig. 9) of a thin coating highlights the crystalline Fe2N interlayer that forms due to the nitriding treatment. We also forcibly peeled off a coating for additional analysis of the interface between the substrate and the coating. An AES survey of the bottom surface of the coating confirms the presence of N and Fe diffused into the coating (Fig. 10). AES surveys contain information from only the top 1–2 nm of the coating. This is the reason behind the strong oxygen signal in Fig. 10. Qualitatively, the nitriding pretreatment was successful, as we were only able to delaminate the coatings by extensive polishing with alumina paste. Present investigations are focused on quantitative determination of the adhesive properties of the coatings using a pin-on-disk tribometer. 4. Conclusions SiC nanostructured coatings deposited by high velocity impaction of nanosized particles were investigated. X-ray diffraction spectra of typical samples showed the presence of
1151
crystalline SiC and Si. Grain sizes obtained through transmission electron microscopy showed particles in the sub 10 nm range with primarily crystalline β-SiC and some crystalline Si particles present. These results correlate well with particle size distributions measured using an aerosol sampling probe coupled to a scanning electrical mobility spectrometer. Interestingly, there were only small changes in the particle size distributions when Si deposition was compared to SiC. The aerosol probe measurements demonstrated a possible formation mechanism of SiC in the HPPD system. Examination of adhesion characteristics highlighted the importance of a chemically bound interlayer during SiC deposition in Mo and steel substrates. All these results taken together demonstrate the potential of the HPPD process for high-volume production of nanostructured coatings, while avoiding many of the problems associated with conventional powder processing routes. However, further work is needed to develop procedures to ensure formation of only crystalline SiC deposits. Acknowledgements This work was partially supported by the United States National Science Foundation, under award numbers DMI0103169 and CTS-0506748, and IGERT award number DGE0114372. The assistance of Joysurya Basu of the Characterization Facility during TEM operation, and John Thomas during Auger electron spectroscopy operation is gratefully acknowledged. We would also like to thank William Mook for the nanoindenation measurements. References [1] S.L. Girshick, J.V.R. Heberlein, P.H. McMurry, W.W. Gerberich, D.I. Iordanoglou, N.P. Rao, A. Gidwani, N. Tymiak, F.D. Fonzo, M.H. Fan, D. Neumann, in: K.L. Choy (Ed.), Innovative Processing of Films and Nanocrystalline Powders, Imperial Press, London, 2002, p. 165. [2] J. Hafiz, X. Wang, R. Mukherjee, W. Mook, C.R. Perrey, J. Deneen, J.V.R. Heberlein, P.H. McMurry, W.W. Gerberich, C.B. Carter, S.L. Girshick, Surf. Coat. Technol. 188–189 (2004) 364. [3] A. Gidwani, Studies of Flow and Particle Transport in Hypersonic Plasma Particle Deposition and Aerodynamic Focusing, University of Minnesota, Minneapolis, 2003. [4] X. Wang, J. Hafiz, R. Mukherjee, T. Renault, J. Heberlein, S.L. Girshick, P.H. McMurry, Plasma Chem. Plasma Process. 25 (2005) 439. [5] M. Dayan, M. Shengli, X. Kewei, S. Veprek, Mater. Lett. 59 (2005) 838. [6] N. Rao, S.L. Girshick, J. Heberlein, P.H. McMurry, S. Jones, D. Hansen, B. Micheel, Plasma Chem. Plasma Process. 15 (1995) 581. [7] Q. Zhao, J.C. Li, H. Zhou, H. Wang, B. Wang, H. Yan, J. Cryst. Growth 260 (2004) 176. [8] C. Summonte, R. Rizzoli, M. Servidori, S. Milita, S. Nicoletti, M. Bianconi, J. Appl. Phys. 96 (2004) 3998. [9] D.F. Bahr, D.V. Bucci, L.S. Schadler, J.A. Last, J. Heberlein, E. Pfender, W.W. Gerberich, Diamond Relat. Mater. 5 (1996) 1462. [10] S. Abisset, F. Maury, R. Feurer, M. Ducarroir, M. Nadal, M. Andrieux, Thin Solid Films 315 (1998) 179. [11] C.F.M. Borges, E. Pfender, J. Heberlein, Diamond Relat. Mater. 10 (2001) 1983. [12] A. da Silva Rocha, T. Strohaecker, V. Tomala, T. Hirsch, Surf. Coat. Technol. 115 (1999) 24.
