Characterization of ion-assisted pulsed laser deposited cubic boron nitride films

Transcription

1 1 S. Weissmantel, G. Reisse, U. Falke, presented at the International Conference on Metallurgical Coatings and Thin Films, ICMCTF 99, April 12-16, 1999, San Diego, California, Thin Solid Films (1999). Characterization of ion-assisted pulsed laser deposited cubic boron nitride films Steffen Weissmantel and Guenter Reisse, Hochschule für Technik und Wirtschaft Mittweida, Technikumplatz 17, Mittweida, Germany Abstract Cubic boron nitride films were deposited by pulsed laser deposition from a boron nitride and a boron target using a KrF-excimer laser, where the growing films were bombarded by a nitrogen or a nitrogen/argon ion beam. The variation of the film properties with laser and ion beam parameters and substrate temperature has been investigated by infrared spectroscopy, cross-section and plan-view high-resolution transmission electron microscopy, electron-energy-loss spectroscopy and in situ ellipsometry. It will be shown that c-bn films with high phase purity can be prepared at sufficiently strong ion bombardment as well as substrate temperatures above 160 C. The c-bn phase was found to grow exclusively on top of the well-known hexagonal interlayer with c-axis orientation parallel to the substrate surface. Two types of nucleation were observed, the first characterized by c-bn (111) and the second by c-bn (001) lattice planes growing parallel to the (0002) lattice planes of the initially formed h-bn layer. c-bn films were prepared at maximum growth rates of 16 nm /min. Additional uv-photon irradiation of the growing films results in distinct modifications of the microstructure of the BN films. Using laser pulse energy densities on the substrate surface above 200 mj/cm 2 the laser irradiation leads to the formation of turbostratic h-bn even though the unirradiated film regions of the same sample show the cubic structure. In contrast, films irradiated at 100 to 160 mj/cm 2 are cubic. Electron microscopic observations show that in this range the mean diameter of crystallite in the excimer laser irradiated regions increased by a factor of 2 in comparison with unirradiated regions of the same sample. The experimental results will be discussed in context with the results of temperature field calculations. 1. Introduction Due to its unique properties, cubic boron nitride (c-bn) is a potential candidate for widespread applications. The preparation of that material in form of thin films has, so far, succeeded exclusively by various ion assisted methods [1-7]. However, there are quite a number of problems with the deposition of those films, which will have to be solved prior to their use as mechanical, optical and electrical coatings. One main problem is the poor adherence of c-bn films caused by the high intrinsic stresses in the films. Another problem for some applications is the small size of the c-bn crystallites of only several 10 nm. In this paper, we will present the results of our studies on the deposition and microstructural characterization of c-bn films prepared by ion-assisted pulsed laser deposition from boron as well as boron nitride targets. One of our goals is the investigation of the influence of additional UV-photon irradiation of the growing films on the microstructural properties of the c-bn films. 2. Experimental details The experimental set-up used for the preparation of boron nitride films by ion-assisted pulsed laser deposition is illustrated and described elsewhere [8]. The deposition parameters used for the preparation of boron nitride films are shown in Table 1. To investigate enhanced energetic activation of film growth by ultraviolet photon bombardment, certain regions of the growing films were additionally irradiated with a KrF excimer laser beam.

2 2 Film growth was controlled by means of an in situ ellipsometer operating at nm wavelength, where the angle of incidence was 70 to the normal of the substrate surface. The films were investigated by infrared spectroscopy using a spectrophotometer Carl-Zeiss M80 operating from 200 to 4000 cm -1 wavenumber and transmission electron microscopy (TEM) as well as electron-energyloss spectroscopy (EELS) using a 200 kev high-resolution electron microscope Philips CM 20. The IR spectra were measured in transmission at normal incidence and ratioed to the spectrum obtained from the uncoated silicon substrates. 3. Results and discussion 3.1. Deposition and characterization of c-bn films Deposition from boron target The preparation of boron nitride films by pulsed laser deposition using boron targets requires strong nitrogen or nitrogen/argon ion bombardment in order to supply sufficient nitrogen of enhanced reactivity. Moreover, a sufficient momentum and energy transfer to the growing film as well as a sufficiently high substrate temperature are necessary for the formation of the cubic phase. As already known from the literature [1,4], the nucleation of the cubic phase occurs only after the formation of a hexagonal interlayer. In our experiments the formation of cubic films on such hexagonal interlayers took place at substrate temperatures above 160 C and a minimum ion current density of 200 µa/cm 2 at 350 ev ion energy was found to be necessary at growth rates up to 1 nm/min. At the maximum laser pulse energy density of 3.6 J/cm 2 and the maximum pulse repetition rate of 50 Hz used for the ablation of the boron target, an ion energy of 500 ev and an ion current density of 450 µa/cm 2 were found to be the optimum ion beam parameters with respect to maximum growth rate and high phase purity. At those optimum parameters, the ion supplied momentum per deposited BN was 1160 (ev. amu) 1/2, the ion supplied energy per deposited film volume was ev/nm 3 during the h-bn interlayer growth and ev/nm 3 during the c-bn growth (see also table 2), where the momentum and energy transfer from the particles arriving from the target was not considered. Fig. 1 shows an IR transmittance spectrum of a BN film deposited at the optimum parameters indicating that apart from c-bn also h-bn is present in the film. The h-bn and c-bn film growth can well be distinguished and controlled by means of the in-situ ellipsometer. The variation of refractive index with film thickness (see Fig. 2), derived by simulation of the in-situ measured Psi- Delta-curve, indicates to the following layer sequence of that BN film: At first a 14 nm thick layer, showing a gradient of refractive index from the Si-substrate value (determined after Ar + ion cleaning using also the ellipsometer) to a value as low as 1.75, is formed by nitridation of the silicon surface with increasing boron content. This low effective refractive index was found to be typical for h-bn nucleation layers and should be related to a low packing density of that hexagonal phase, considering the angle of incidence of the ellipsometer beam, the anisotropy of h-bn, the preferred orientation of the c-axis parallel to the substrate surface and the random orientation of the h-bn crystallites about the substrate normal. The minimum thickness of those layers amounted to 15 nm with our deposition method. After an 11 nm thick layer with the refractive index increasing to the value of c-bn, which is due to laterally different nucleation of c-bn, a 75 nm thick c-bn layer has grown. The growth rates of the h-bn interlayer was 4.8 nm/min and of the actual c-bn film 2.7 nm/min, respectively. The ratio of these growth rates is nearly the same as the inverse ratio of the h-bn to c-bn density and is, therefore, mainly related to the difference in the number of atoms per volume in both phases. Consequently, the difference in resputtering rate of both phases by ion bombardment should be small. The ellipsometrically determined layer sequence is confirmed by the cross-sectional TEM studies performed on the same c-bn film. The two dark field images taken from the same cross-sectional region, but one formed with a h-bn (0002) reflex and the other with a c-bn (111) reflex (see Figs. 3a and b), clearly show a thin initially formed layer, which can be seen to be amorphous in high resolution (see Fig. 4a), followed by a well-oriented hexagonal interlayer. This h-bn nucleation layer shows the well-known c-axis orientation parallel to the substrate surface (see Figs. 3c and 4a). The

3 3 (0002) lattice plane spacing was found to be in the range of to nm, being larger compared to that of the ideal crystal (d 0002 =0.333nm). On that specific h-bn interlayer, a nearly phase pure and strongly oriented c-bn layer grows in columns (see Figs. 3a to c). This is supported by the EELS spectra from the same cross-section (see Fig. 5), showing a nearly vanishing π * -peak. The small remaining π * -peak might be the result of some sp 2 -bound material in the grain boundaries between c-bn crystallites. Two types of nucleation and growth of the cubic phase upon the hexagonal phase can be found in the HRTEM micrographs. One type, shown in Fig. 4c, is characterized by the growth of c-bn (111) lattice planes parallel to the h-bn (0002) lattice planes with a lattice matching of 3:2. With the second type, occurring at least with the same abundance, the c-bn (001) lattice planes are parallel to the h-bn (0002) lattice planes, allowing theoretically an even better lattice matching of 1:1 for the larger values of the h-bn (0002) spacing (see Fig. 4b, showing c-bn (111) planes and h-bn (0002) planes at an angle of nearly 54.7, which is the theoretical angle for this type of nucleation). The evaluation of the c-bn diffraction pattern yielded a <110> preferred orientation of the crystallites perpendicular to the substrate (see Fig. 3c). This orientation of c-bn crystallites remains the same over the whole thickness of the c-bn film. At the growing film surface, however, a thin sublayer of hexagonal boron nitride can be observed (see Fig. 4 a and f), which supports the cylindrical spike model proposed by Hofsäss et.al. [10] as well as the subplantation model from Robertson [11]. Plan-view TEM investigations have shown that the h-bn and c-bn layers are randomly oriented about the surface normal. A known problem with the deposition of c-bn films, which we experienced too, is the low adhesion of the films to the substrates due to the poor mechanical stability of the h-bn nucleation layer. In another paper of this conference [8], we report on the improvement of the adhesion of c-bn films by using pulsed laser deposited h-bn intermediate layers Deposition from boron nitride target Though films deposited by pulsed laser ablation from boron nitride targets without ion bombardment may also be nitrogen deficient (in dependence of nitrogen background pressure and growth rate), the nitrogen supply to the growing films needed for stoichiometric BN films is relatively low. This should allow, in principle, the preparation of c-bn films at a lower ion bombardment or higher growth rates, which could indeed be achieved. However, as we learned from experiments, the growth rate of the initial hexagonal interlayer must not exceed relatively low values for the nucleation of the cubic phase, whereas the growth rate of the cubic phase can be increased after nucleation. Using constantly high laser pulse energy densities, the growth rate was varied by controlling the pulse repetition rate of the target laser beam. The dependency of the maximum growth rates, at which the cubic phase nucleates, and the maximum growth rates, at which the c-bn films could be grown after nucleation, on the ion beam parameters and, in particular, on the necessary ion-to-arriving-target-atom ratio as well as the momentum and energy transfer into the films are presented in Table 2. It should be noted that the momentum and energy transfer supplied into the films by the particles arriving from the target was not considered in the given values. Calorimetric measurements have shown that the mean kinetic energy of those arriving particles amounts to about 30 ev per atom at 7.3 J/cm 2 laser pulse energy density. As can be seen in the right column of Table 2 the resputtering rate is only insignificantly higher for the h-bn phase than for the c-bn phase. In comparison to c-bn growth from the boron target, a four times higher growth rate could be achieved at the same ion energy. The structure of c-bn films deposited from the BN-target is identical to that of films deposited from the boron target, as we concluded from the corresponding IR spectroscopic and TEM investigations [8, 9]. First investigations using SiO 2 substrates have shown that c-bn films can be deposited on those substrates, too, and that they have the same layered structure as on silicon.

4 Modification of c-bn films by ultraviolet laser irradiation during film growth Though it is possible to deposit nearly pure c-bn films, the mean size of crystallite is only in the range from 10 to 20 nm (see Fig. 6a). In order to investigate the influence of ultraviolet photons on the size of crystallite, we have additionally irradiated a part of the growing films with a ultraviolet laser beam of 248 nm wavelength, where the angle of incidence was 45. No modification of the irradiated film area could be observed below an energy density of 0.07 J/cm 2. In the energy density range of 0.07 to 0.11 J/cm 2 we reproducibly observed nearly a doubling of the mean size of crystallite in comparison to the unirradiated part of the same film (see Figs. 6a and b). Additionally, on some irradiated samples a certain preferred orientation of the crystallites of the nucleation layer as well as of the cubic film about the substrate surface normal was observed in the diffraction patterns (see Fig. 6b). Increasing the energy density to values in the range of 0.14 to 0.19 J/cm 2 led to the growth of purely hexagonal films with turbostratic microstructure (see Fig. 6c), even though the unirradiated part of the film showed the cubic microstructure. Though the main goal was the photoexcitation of the electrons in the growing films, temperature field calculations using a numerical method [12] show that the photon induced temperature rise may also play a role for the observed modifications, though the temperature rise is of only short duration in the nanosecond range (see Figs. 7a and b). The calculations have been carried out onedimensionally using optical material parameters determined by photometry (n Si =1.80, k Si =3.95, n h- BN=2.1, k h-bn =0.04, n c-bn =2.1, k c-bn =0). The used reflectivity is the mean value of the reflectivities calculated for orthogonal and parallel polarized radiation. The other boron nitride material parameters (h-bn: λ th =0.63 W cm -1 K -1 perpendicular to the c-axis, ρ=2.2 g cm -3, c-bn: λ th =2.0 W cm -1 K -1, ρ=3.49 g cm -3, ε=0.7) were taken from the literature [13]. The initial temperature was 500 K according to the substrate heating and the thickness of the substrate was 1 mm. In the higher energy density range the calculated maximum temperatures per laser pulse exceeded 1600 K, approaching the melting temperature of silicon. Such temperature rise apparently prevents the hexagonal interlayer with c-axis orientation parallel to the substrate surface from forming and with it the nucleation of the cubic phase. As can be seen in Fig. 7b, the laser induced maximum temperature per pulse strongly depends on the BN film thickness, which is due to the film thickness dependent reflectivity. For that reason, laser modification of growing films at constant temperature must be performed with thickness dependent control of the laser pulse energy. 4. Conclusions Nearly phase-pure c-bn films have been prepared by ion-assisted pulsed laser deposition. The cubic phase grows exclusively on top of the well-known h-bn nucleation layer with c-axis orientation parallel to the substrate surface. It nucleates either with (111) lattice planes or with (001) lattice planes growing parallel to the h-bn (0002) lattice planes, showing lattice matching of 3:2 and 1:1, respectively. A thin h-bn layer can be observed at the surface of the growing c-bn film. Additional UV-laser irradiation of the growing films results in an increasing mean size of crystallite by a factor of two. Acknowledgement The authors gratefully acknowledge financial support of the present work by the Deutsche Forschungsgemeinschaft (Project Numbers RE883/3-1 and RE883/3-2) carried out under the auspices of the trinational D-A-CH German, Austrian and Swiss co-operation on the Synthesis of Superhard Materials and by the Sächsisches Staatsministerium für Wissenschaft und Kunst (LIST08). References 1. D.L. Medlin, T.A. Friedmann, P.B. Mirkarimi, P. Rez, M.J. Mills, K.F. McCarty, J. Appl. Phys., 76 (1994) F.Quin, V. Nagabushnam, R.K. Singh, Appl. Phys. Lett., 63 (1993) 317.

5 5 3. T. Klotzbücher, W.Pfleging, M. Mertin, D.A. Wesner, E.W. Kreutz, Appl. Surf. Sci. 86 (1995) D.J. Kester, K.S. Ailey, R.F. Davis, Diamond and Related Materials, 3 (1994) J. Robertson, Advances in Physics 35 (1986) S. Reinke, M. Kuhr, W. Kulisch, Diamond and Related Materials, 3 (1994) G. Reisse, S. Weissmantel, B. Keiper, A. Weber, Appl. Surf. Sci (1998) S. Weissmantel, G. Reisse, The use of pulsed laser ablated boron nitride interlayers for improving the adhesion of cubic boron nitride films, presented also at the ICMCTF99, San Diego 1999, to appear in the Proceedings and in Thin Solid Films. 9. S. Weissmantel, G. Reisse, B. Keiper, S.Schulze, Diamond Relat. Mater. 8 (1999) H. Hofsäss, H. Feldermann, R. Merk, M. Sebastian, C. Ronning, Appl. Phys. A 66 (1998) J. Robertson, Diamond Relat. Mater. 2 (1993) S. Weissmantel, thesis, TU Chemnitz, Properties of Group III Nitrides, J.H. Edgar ed., emis Datareviews Series No. 11 INSPEC, London 1994.

6 6 Captions Table 1. Deposition parameters used for the preparation of cubic boron nitride films by ion-assisted pulsed laser deposition. (λ-wavelength, τ-laser pulse duration, f T -laser pulse repetition rate, A T /A S - laser spot size on target/substrate surface, H 0T /H 0S laser pulse energy density at the target/substrate, E I -ion energy, j S -ion current density at the substrate). Table 2: Parameters for the deposition of c-bn films with high phase purity from boron nitride as well as boron target. The number of atoms arriving from the target was determined from the growth rates without ion bombardment. Densities used for the calculations were 2.28 g/cm 3 for h-bn and 3.49 g/cm 3 for c-bn. The resputtering rate was determined from the ratio of deposited atoms and atoms arriving from the target. Fig. 1. IR spectrum of a c-bn film deposited from a boron target showing a large peak at 1080 cm -1 attributed to c-bn and two peaks at 1400 cm -1 and 780 cm -1 attributed to h-bn (Deposition parameters: H 0T = 3.6 J/cm 2, f T = 50 Hz, E I = 500 ev, j S = 450 µa/cm 2, T S = 220 C). Fig. 2. Refractive index versus thickness of the growing film simulated from the experimental Psi- Delta-curve of Fig. 2. Fig. 3. Cross-sectional TEM dark-field images formed with a h-bn (0002) reflex (a) and a c-bn (111) reflex (b) as well as the corresponding cross-sectional diffraction pattern (c) from a c-bn film (boron target, H 0T = 3.6 J/cm 2, f T = 50 Hz, E I = 500 ev, j S = 450 µa/cm 2, T S = 220 C). The c-bn crystallites are <110> oriented perpendicular to the substrate surface and randomly oriented about the substrate normal. In case a the (111) or the (111) planes and in case b the (111), (111) and (001) planes are parallel to the electron beam. The lattice plane spacings determined from (c) are shown in the table (except for silicon). Fig. 4. Cross-sectional HRTEM images of the c-bn film of Figs. 2 and 3. a) General view showing the layered growth (silicon, amorphous layer, h-bn, c-bn) b) Transitional region with c-bn (001) lattice planes growing parallel to h-bn (0002) lattice planes showing the h-bn planes in the lower part and the c-bn (111) planes at an angle of nearly c) Transitional region from h-bn to c-bn showing 3:2 lattice matching of the c-bn (111) and the h- BN (0002) lattice planes. d,e) c-bn (111) lattice planes, d 111 =0.208 nm. f) Upper film region showing an h-bn surface layer. Fig. 5. EELS spectra measured at different regions of the cross-section of the BN film of Fig. 3 (electron energy 200 kev, electron beam diameter 10 nm). Fig. 6. Dark field TEM micrographs and diffraction patterns of BN films deposited from a boron target (H 0T = 3.6 J/cm 2, f T =50 Hz, E I = 500 ev, j S = 450 µa/cm 2, T S =260 C) grown without and with laser irradiation during deposition. a) Plan view of an unirradiated region. Dark field image formed with a segment of the (111) c-bn ring. b) Plan view of a laser irradiated region of the same film as a) (H 0S = 0.09 J/cm 2, f S =150 Hz). Dark field image formed with a segment of the (111) c-bn ring. The mean size of crystallite is about twice the size as in the unirradiated region. c) Cross-sectional view of a laser irradiated region of another film (H 0S = 0.18 J/cm 2, f S =150 Hz). Dark field image formed with a segment of the (0002) h-bn ring. The film is completely hexagonal.

7 7 Fig. 7. Calculation of KrF excimer laser induced temperature fields in c-bn/h-bn/si-systems. a - Time dependence of a one-dimensional temperature field induced by a pulse of the shown form (H=0.091 J/cm 2 ). b - Maximum film temperature during a laser pulse versus BN film thickness for two laser pulse energy densities. Table 1. Target laser beam λ = 248 nm, τ = 30 ns, f T = Hz, A T = 3 mm 2 (B-target), A T = 4.1 mm 2 (BN-target), H 0T = J/cm 2, spiral motion with constant vector velocity (12.5 mm/s) across a target area 10 mm in diameter Substrate ion beam E I = ev, j S = µa/cm 2, feeding gas: Ar/N 2 -mixture (ratio of partial pressure p N2 /p Ar = 2:1) Substrate laser λ = 248 nm, τ = 20 ns, f S = 150 Hz, A S = 20 mm 2, H 0S = J/cm 2 beam Base pressure p 0 = Pa Working pressure p w = Pa Substrate T S = C temperature Targets hot-pressed B and hot-pressed as well as pyrolytical h-bn (both 99.9 % purity) Substrate materials Si (100), stainless steel, SiO 2 Table 2. Boron nitride target f T Hz growth rate nm/min atoms arriving from BN target cm -2. s -1 ion-toarriving target atom ratio ion supplied momentum per deposited BN (ev. amu) 1/2 ion supplied energy per deposited volume BN ev/nm 3 ev resputtered B and N atoms per incident ion E I = 300 ev, j S = 75 µa/cm 2, H 0T = 6.6 J/cm 2 until c-bn nucleation after c-bn nucleation E I = 500 ev, j S = 310 µa/cm 2, H 0T = 7.3 J/cm 2 until c-bn nucleation after c-bn nucleation E I = 700 ev, j S = 570 µa/cm 2, H 0T = 6.6 J/cm 2 until c-bn nucleation after c-bn nucleation Boron target E I = 500 ev, j S = 450 µa/cm 2, H 0T = 3.6 J/cm 2 until c-bn nucleation after c-bn nucleation

8 8 Fig Transmittance 0,9 0,8 0,7 h-bn c-bn h-bn 0, Wavenumber [cm -1 ] Fig. 2. Refractive index h-bn c-bn Film thickness [nm]

9 9 Fig. 3. c-bn (220) a,b c-bn (111) b h-bn (0002) c-bn (111) b c-bn (111) a c-bn (111) a [c-bn (001)] Lattice plane spacing [nm] Assignment / theoretical value [nm] h-bn (0002)/ c-bn (111) / c-bn (220) / c-bn (311) / c c

10 10 Fig. 4.

11 11 Fig. 5. Fig. 6. a 70 nm 70 nm c 20 nm b

12 12 Fig. 7. a) Si-substrate BN-films Laser pulse b) Maximum temperature per laser pulse [K] 1700 H=0.091 J/cm^2 H=0.182 J/cm^ Film thickness [nm]