ELSEVIER
Surface and Coatings Technology lOO&lOl ( 1998) 45-48
Scratch-resistant transparent boron nitride films E.H.A. Dekempeneer
a-*, S. Kuypers a, K. Vercammen a, J. Meneve a, J. Smeets a, P.N. Gibson b, W. Gissler b
a Vlaamse Instelling voor Technologisch Onderzoek, Boeretang 200, B-2400 Mel, Belgium b Joint Research Centre of the Commission of’ the European Communities, Institute,for Advanced Materials, Ispra (Va), Italy
Abstract Transparent boron nitride (BN) coatings were deposited on glass and Si substrates in a conventional capacitively coupled RF PACVD system starting from diborane (diluted in helium) and nitrogen. By varying the plasma conditions (bias voltage, ion current density), coatings were prepared with hardness values ranging from 2 to 12 GPa (measured with a nano-indenter). Infrared absorption measurements indicated that the BN was of the hexagonal type. A combination of glancing-angle X-ray diffraction measurements and simulations shows that the coatings consist of hexagonal-type BN crystallites with different degrees of disorder (nanocrystalline or turbostratic material ). High-resolution transmission electron microscopy analysis revealed the presence of an amorphous interface layer and on top of this interface layer a well-developed fringe pattern characteristic for the basal planes in h-BN. Depending on the plasma process conditions, these fringe patterns showed different degrees of disorder as well as different orientational relationships with respect to the substrate surface. These observations were correlated with the mechanical properties of the films. 0 1998 Elsevier Science S.A. Keywords:
Boron nitride;
Microstructure;
Transparent
films; PACVD
1. Introduction As part of a broad investigation of light-elementbasedcoatings prepared within the boron-nitrogen-carbon (BNC) composition triangle [l-4], boron nitride materials deposited in a conventional RF PACVD process were studied. The aim was to develop smooth, transparent, wear-resistant coatings using low-temperature Plasma Assisted Chemical Vapour Deposition (PACVD) processes. Hard “amorphous” BN films deposited at low temperatures have been reported previously [4-61. However, the microstructure of these films, and in particular the correlation with other material properties, is not well understood. In the crystalline state, BN has two principal forms [ 71: hexagonal (h-BN ) and cubic (c-BN). Moreover, h-BN-type films often develop a disordered microstructure referred to as turbostratic, showing a tendency to disorder in the stacking of the layers. In the latter case, diffraction from (hkl) crystal planes with 1>O is weak or absent and the diffraction pattern starts to resemble the c-BN pattern. * Corresponding author. Tel: + 32 14 335623: Fax: + 32 14 321186; e-mail:
[email protected] 0257~8972/98/$19.00 0 1998 Elsevier Science B.V. All rights reserved. PII SO257-8972(97)00585-9
films
This ambiguity in the identification of diffraction spectra is an important factor explaining the different interpretations of the microstructure of these BN films [8,9]. In this paper, more insight was obtained from a combination of glancing angle X-ray diffraction (GAXRD) measurements,cross-sectional high-resolution transmission electron microscopy (HRTEM ) and Fourier transformed infrared spectroscopy (FTIR).
2. Experimental details Hydrogenated BN films were prepared in a capacitively coupled RF PACVD reactor starting from N, and B,H, ( 10% diluted in He) gasmixtures at a pressure of 1 Pa. The coatings were deposited on Si and glass substrates which were placed on the powered electrode. The bias voltage was varied between -200 and -600 V by changing the RF power. In addition, two electromagnetic coils were placed on opposite sides of the reactor, generating a magnetic field parallel to the substrate electrode [lo]. The magnetic field was varied between 0 and 5 mT by controlling the coil current, I,. between 0 and 4 A. This configuration enabled the ion current
density to be varied independently of the bias voltage. This was verified qualitatively by measuring the sputter rate on Si substrates placed on the powered electrode in an Ar plasma at a fixed bias voltage (- 200 V ) and gas pressure. This sputter rate increased by a factor ~7 when increasing 1, from 0 to 4 A. The film composition (B, N and impurity 0) was measured with electron probe microanalysis (EPMA). The microstructure of the coatings was investigated using FTIR, GAXRD and HRTEM. GAXRD spectra were obtained on 1 x 2 cm2 coated Si substrates at an incident angle of 0.5’ [8]. HRTEM analysis was carried out on cross-sections prepared from coated Si substrates on a JEOL JEM-4000EX at 400 kV accelerating voltage. The magnification was accurately determined using the substrate lattice spacings as an internal reference. The hardness of the films was measured with a depth-sensing indentation instrument [4].
3. Results and discussion It is well known that the transition from h-BN to c-BN in a vacuum deposition process is strongly controlled by the total momentum of the incoming ions [ 111. Therefore, experiments were conducted in which bias voltage and ion current density were varied while other process parameters were kept fixed (Table 1). Due to the absence of heavy ions (e.g. argon) and the relatively low temperature, the formation of c-BN was not expected [IO]. Instead, the properties of various hexagonal-type BN films were investigated. According to EPMA, all films were close to stoichiometric BN with, however, a tendency for a slight excess in N content of a few at.%. The 0 contamination was less than 1 at.% and the H content was not measured. The coatings on the glass specimens looked completely Table 1 BN coatings
deposited
Bias voltage
(V )
- 200
-400
~ 600
under Coil
different
current.
process conditions
(A-G
transparent. According to Table 1, the different coatings can be separated into two groups: one group with a high hardness value (Z 12 GPa) and the other with much lower hardness values. The infrared absorption spectra of all films A-G (not shown here) showed the characteristic features of h-BN, with the main absorption peaks around 780 cm ’ and 1380 cm-‘. These peak position values indicated that there was a good local ordering in the hexagonal basal planes [ 121. The GAXRD diffraction spectra, however, showed a distinctly different behaviour for the harder and softer coatings ( Fig. 1). The softer coatings (D and F) showed strong diffraction peaks around 26” and 42’ and a weaker peak around 78 , whereas for the harder coatings (A, C, E and G) the 26~ peak was very weak or absent. The simulated spectra shown in Fig. 1 were calculated using a Rietveld-type analysis programme using two different models. For the harder coating (C) the simulation was performed by assuming a hexagonal BN structure, but the crystallite size in the c-axis direction was reduced to an unrealistically low value (0.025 nm). This effectively caused the disappearance of the 26” (002) diffraction peak and suggested that the coatings were of a mainly turbostratic nature. For the softer coating (D) it was necessary to include a second phase, randomly orientated nanocrystalline BN, to obtain a reasonable approximation to the experimental spectra. The crystallite size in all directions was about 3 nm and a relatively large strain (6%) as well as a lattice parameter increase of 7% in the c-axis direction were used to obtain a good simulation. Representative cross-sectional HRTEM results for both hard and soft coatings are shown in Fig. 2. Both coating types showed an initial amorphous interface layer of around 5 nm thickness. On top of this amorphous interface layer, the hard coating showed well-
)
I, (A)
0
I
2
A 12.0 GPa 20 w E 12.7 CPa 60 W G 11.9 GPa 105 w
B 12.1 GPa 30 w
C 11.6 GPa 40 w F
4 D 2 GPd
IIOW
4 GPd
140 w
J--. ~~~~~ 0
The process parameters that were varied were bias voltage and coil current, I,. Tabulated values are: the hardness (GPa) of the deposited coatings and the RF power ( W) dissipated in the plasma. Each hardness value is the average of IO measurements (SD 2 10%). The bold area marks films with a higher hardness, typically around I2 GPa.
~~______ 20
40 60 2 theta (degrees)
80
100
Fig. I. Measured (thick line) and simulated (thin line) GAXRD spectra of a typically hard BN coating (C) and a typically soft BN coating (D). These samples were also used for the HRTEM analysis in Fig. 2.
41
Fig. 2. Cross-sectional HRTEM images of a hard coating prepared under condition C (a, b) and a soft coating prepared under condition D (c, d). (a, c) Images from the central regions of the respective BN layers revealing preferentially orientated (a) and randomly orientated (c) crystallites. (b, d) Images taken near the coating/substrate interface. The latter reveal a thin ( ~5 nm) amorphous layer at the interface and a preferentially orientated h-BN layer on top. The GAXRD spectra of these films are shown in Fig. 1.
developed parallel fringes corresponding to the basal planes of h-BN. The fringes were aligned perpendicular to the substrate surface (Fig,. 2b), probably as a result of stress in the coatings [ 131. This preferential orientation remained constant throughout the whole layer thickness (Fig. 2a). Strong curvatures in the fringe pattern indicated a turbostratic structure. In contrast, in the case of the soft coating, after initially developing a fringe pattern which was similarly aligned perpendicular to the surface (Fig. 2d), the preferential orientation of the basal planes was suddenly disrupted and tended to develop a more random character (Fig. 2~). Similar layered structures were observed in [ 141. The transition to a randomly orientated structure occurred after a layer thickness of around 40 nm. The fringe patterns observed in the softer coating tended to be less disordered (less
curved and extended over larger areas), indicative of a more nanocrystalline material. The spacings between the basal planes extracted from these HRTEM images were around 0.36 nm for both coating types, which was about 9% larger than the h-BN bulk value of 0.33 nm. This observation is in good agreement with the large lattice parameter increase in the c-direction (7%) extracted from the GAXRD analysis. The HRTEM analysis seems to confirm the differentiation between hard turbostratic and soft nanocrystalline coatings. However, it also clearly demonstrated that orientational effects are very important in these layers. The assumption of a random orientation of crystallites, as for the GAXRD simulations, therefore, is not totally correct. The observed variations from a fully orientated layer with its basal planes perpendicular to the surface (harder
48
E. H.A.
Dekrmpenrer
et al. 1 Surf&e
and Coatings
films) to a randomly orientated layer (softer films) may also contribute to the observed amplitude variations of the 26’ peak in the GAXRD spectra.
Tdm~log~
100-101
Mobility Programme CT94-0575.
(1998)
43-48
(networks),
contract
no. CHRX-
References 4. Conclusions [l]
Smooth, transparent BN coatings with different mechanical properties were prepared by RF PACVD. The harder coatings appeared to be of the turbostratic type, with the hexagonal BN planes being preferentially orientated perpendicular to the substrate surface. The softer coatings, although initially developing a similarly preferentially orientated microstructure, were shown to undergo a transformation towards a different microstructure in which the crystallites become randomly orientated. Moreover, these crystallites appear to be slightly less disordered. These different microstructures were obtained by varying the bias voltage and the ion current density during the deposition process, under conditions (relatively low total momentum and deposition temperature) that are not suitable for making c-BN.
[2] [3] [4] [5] [6] [7] [8] [9]
[IO] [I I]
Acknowledgement [ 131
We appreciate the technical assistance of Mr. Rossou (EMAT, University of Antwerp, Belgium) the TEM sample preparation, and acknowledge the for supporting this work via the Human Capital
L. for EC and
[I31 [ 141
E.H.A. Dekempeneer. V. Wagner. L.J. van Ijzendoorn. J. Meneve. S. Kuypers. J. Smeets. J. Geurts, R. Caudano. Surf. Coat. Technol. 8687 (1996) 581. E.H.A. Dekempeneer. J. Meneve, S. Kuypers. J. Smeets, Thin Solid Films 381282 (1996) 331. E.H.A. Dekempeneer, J. Meneve, J. Smeets, S. Kuypers, L. Eersels. R. Jacobs. Surf. Coat. Technol. 6869 (1994) 621. E.H.A. Dekempeneer. J. Meneve, S. Kuypers, J. Smeets. Surf. Coat. Technol. 7475 ( 1995) 399. K. Montasser. S. Morita. S. Hattori, Mater. Sci. Forum 5455 (1990) 295. M.A. Tamor. W.C. Vassell. US Patent 5518780. D.R. McKenzie, W.G. Sainty. D. Green. Mater. Sci. Forum 5455 (1990) 193. W. Gissler. J. Haupt. A. HoEmann. P.N. Gibson. D.G. Rickerby. Thin Solid Films 199 ( 1991) 113. E. Dekempeneer. V. Wagner. P.N. Gibson, J. Meneve, S. Kuypers. L.J. van Ijzendoorn. J. Smeets. J. Geurts. R. Caudano. Diamond Films Technol. 7 (1997) 181- 193. W. Dworschak, K. Jung, H. Ehrhardt. Thin Solid Films 254 (1995) 65. P.B. Mirkarimi. K.F. McCarty. D.L. Medlin, W.G. Wolfer. T.A. Friedman. E.J. Claus. J. Mater. Res. 9 ( 1994) 2925. K.F. McCarty, P.B. Mirkarimi, D.L. Medlin, T.A. Friedman, J.C. Barbour. Diamond Relat. Mater. 5 (1996) 1519. D.R. McKenzie, W.D. McFall. W.G. Sainty. C.A. Davis. R.E. Collins. Diamond Relat. Mater. 2 (1993) 970. J.L. Andujar. E. Bertrdn, Y. Maniette. J. Appl. Phys. 80 ( 1996) 6553.
