Materials Chemistry and Physics 44 (1996) 9-16
Microstructure and properties of Ti-Si-N films prepared by plasma-enhanced chemical vapor deposition J.L. He”>*, C.K. Chenb, M.H. bDepartntent
aDepartmenr of Materials
of Materials Science, Feng Chia Uniuersity, Taiclmng, Tailvan, ROC Science and Eilgineeriizg, National Cherzg Kung Ufziuersity, Tainan, Taiwan,
Received 11 March 1994; accepted2 August 1995
The promising hardnessof multicomponent Ti-Si-N films reported recently encouragesa systematicinvestigation on the microstructure of the films. In this paper, an attempt is made to correlate the surfacehardnesswith the microstructural change asa function of Si concentration controlled by SiH4flow rate. The Ti-Si-N films weredepositedusingplasma-enhanced chemical vapor deposition(PECVD) with SiH,, TiCl,, N, and H, as gassources.Resultsof microstructural analysissuggestthat a ‘solid solution strengthening’promotesthe hardnessof a Ti-Si-N film depositedat a lower SiH, flow rate. A higher SiH, flow rate inducesthe precipitation of an a-Si3N4phasein the depositedTi-Si-N films and reducesthe hardnessvalue. Keywords:
1. Introduction Current trends in coating techniques involve multilayer or multicomponent coatings which are expected to have tailor-made properties for some specific applications. A large variety of coating material combinations was investigated over the years to reveal the outstanding properties that one coating material alone can never have. Films consisting of Si3N4 and TIN recently attracted interest because both materials are commonly used as protective materials, and are compatible with each other to a certain extent. In the early years of the latest decade, Hirai et al. prepared Si3N4-TiN composites by chemical vapor deposition (CVD) [ 11,characterized them by transmission electron microscopy (TEM)  and megavolt TEM [ 31, and determined some of their properties . Posadowski  evaluated the electrical properties of Ti-Si-N films deposited by the sputtering of two-component targets. A fYm consisting of S&N4 and TiN showed a relatively low resistance (AR/R> to a temperature variation of l%, which presumably resulted from * Corresponding
0254-0584/96/S15.00 0 1996 Elsevier CCnTrs-lc” nc~dloz\nl6
good protection of the films against any environmental effects. Taniuchi et al.  chemically vapor deposited Si,N, onto metal substrates by using TiN as an intermediate layer which contributed to the good adherence of Si,N4 and acted as the diff&ion barrier for silicon and substrate elements, thus promoting oxidation resistance. Li et al.  examined a Ti-Si-N &-I-I prepared by plasma-enhanced chemical vapor deposition (PECVD) and found that the Ti-Si-N films containing lo- 15 at.% Si showed a surprisingly high hardness value of about 6350 kg mm-‘, much higher than that of TIN films. Films obtained from conventional CVD usually show a well-defined crystallinity and can be clearly identified, while those obtained from plasma-enhanced techniques, multicomponent in particular, show an ambiguous structure as concluded by Li et al. consisting of crystalline TiN, amorphous S&N4 and a small amount of free Si. The surprising hardness value of Ti-Si-N films reported by Li necessitatesa systematic investigation. In this study, the microstructure was determined as a function of Si concentration controlled by SiH, flow rate during deposition. An attempt was also made to correlate the microstructure with the surface hardness of the deposited Ti-Si-N film.
J.L. He et al. 1 Materials Chemistry rind Physics 44 (1996) 9-16
Table 1 Deposition parameters for Ti-Si-N
SiH, flow rate TiCl, flow rate N, flow rate H, flow rate Substrate temperature Discharge power Working pressure Deposition time
O-7.13 ml min-’ 7.13 ml min-’ 325 ml min-’ 273 ml min-’ 600 “C 200 w 5 torr 180 min
The Ti-Si-N films were deposited using an RFPECVD apparatus in which the internal electrodes were capacitively coupled. The r.f. electrode at a relatively lower position was also used a as substrate table and was auxiliarily heated by a graphite heater. The ground electrode, made of stainless steel, was also used as a gas shower that allowed reactive gases to flow uniformly through the substrate table. Tic&, 5% SiH4 in Ar, N2 and H, were used as reactive gases. The deposition parameters are shown in Table 1, and the SiH4 flow rate was changed from 0 to 7.13 ml min-’ in order to reveal the effect of Si concentration on the microstructure of Ti-Si-N. Silicon wafer, silicon nitride and alumina substrate were used as substrates. The coated specimens are described in Table 2 with respect to the SiH4 flow rate and the corresponding SiH,/TiCl, flow ratio. The higher designation numbers represent the higher SiH4 flow rates used. The specimens were analyzed using scanning electron microscopy (SEM), X-ray diffraction (XRD), TEM, and electron spectroscopy for chemical analysis (ESCA). For the precision lattice constant measurement by XRD, silicon powder was used for calibration. The surface hardness was measured by a microhardness
50 60 cnglc
50 60 angle
Fig. 1. Dependence of XRD patterns of Ti-Si-N on SiH, flow rate deposited on (a) silicon nitride and (b) alumina substrate.
tester to determine the dependence on the Si concentration in the film. 3. Results and discussion
Table 2 Designation of specimens with respect to SiH, flow rate and SiH,/ TiCI, flow ratio Silicon nitride
SiH, flow rate (ml min-‘)
SiH,/TiCI, flow ratio
SB so Sl s2 s3 s4 s5 S6 s7 S8 s9
AB A0 Al A2 A3 A4 A5 A6 A7 A8 A9
WB wo Wl W2 w3 w4 w5 W6 WI W8 w9
blank 0.00 0.60 0.71 0.91 1.19 1.43 1.78 2.38 3.51 7.13
blank 0:l 1:12 1:lO 1:8 1:6 1:5 1:4 1:3 1:2 1:l
3.1. XRLI analysis Fig. 1 shows the dependence of XRD patterns of Ti-Si-N films on SiH4 flow rate deposited on (a) silicon nitride and (b) alumina substrates. Comparing silicon nitride specimens SO with SB, an extra peak at about 26’ = 42.3”, corresponding to TiN(200), was observed. This is due to the typical preferred orientation of growth of TiN films in a glow discharge process. The same result, shown in Fig. l(b), was obtained in the case of alumina substrates. As the SiH4 flow rate was increased, the peak intensity at about 20 = 34.5”, corresponding to a-Si,N,( 102), was gradually increased due to the gradual appearance of a-SiaNd phase in the Ti-Si-N films as a function of SiH, flow rate. Note that the peak a-Si,N,(102) is invisible until specimens S3
He et al. 1 Materials
in the TiN lattice in our study. The difference may come from the reactivity of SiH4 being much higher than that of Sic&. A Si-H bond is much easier to break than a Si-Cl bond in plasma, so that with Si from SiH4 it is much easier to overcome the energy barrier for formation of either the solid solution or the a-Si,N, phase.
3.2. SEM analysis 27
(4 TlN(2CQ) 4
Fig. 3 shows cross-sectional micrographs of the Ti-Si-N films, i.e., WO to W9, deposited with different
Fig. 2. Precision lattice constant measurement of TiN(200) for (a) silicon nitride and (b) alumina specimens at a scan speed of 0.25 deg min-’ using silicon powder for calibration.
and A3, corresponding to the films deposited at a SiH, flow rate higher than 0.91 ml mm-‘. This implies that a solid solution of Si is present in the TiN lattice if the SiH4 flow rate is kept low. The evidence for a solid solution of Si in the TIN lattice was obtained from precision lattice constant measurements and TEM selected area diffraction (SAD). The ionic radius of Si4+ (0.41 A) is smaller than that of Ti3+ (0.75 A) . Substitution of an Si ion for a Ti ion leads to a reduction in the TIN lattice constant, which shifts the TiN(200) peak in the XRD pattern to a higher angle, as indicated in Fig. 2 for (a) silicon nitride and (b) alumina substrates, However, the solubility is limited and the extra amount of Si reacts with nitrogen to form an a-Si,N, phase if a higher SiH, flow rate is employed. The results of XRD analysis show that the amount of u-S&N4 phase increases and TiN phase decreases as the SiH4 flow rate is increased. It is also obvious that Si dissolves into the TiN lattice. This is somewhat different from the results of Li et al. [ 71. They found a broadening instead of shifting of the TiN(200) peak as the SiClJ Tic& flow ratio was increased, without any crystal growth of a-S&N, for the deposition parameters used. This means that the addition of Si degrades the crystallinity of the TIN lattice in terms of the ‘nano particle’, so called by Li et al. rather than by a substitution of Si
(4 Fig. 3.
J.L. He et al. 1 Materials Chemistry and Physics 44 (1996) 9-16
(3 Fig. 3. Cross-sectional micrographs of Ti-Si-N filmsdeposited on a siliconwaferat differentSiH, Rowratesfor specimens (a) WO,(b) WI, (c) W2,(d) W3,(e) W4,(f) W5,(g) WG,(h) W7,(i) WS,and(j) w9.
SH4 flow rates. It can be seen that the film thickness and columnar grain sizechanged as the SH4 flow rate increased.Since all deposition batcheswere kept at 180 min, the growth rate, in terms of film thicknessper unit time versus SiH4 flow rate, can be seenin Fig. 4. In the case of TiN deposition without SiH4 addition, the growth rate was approximately 0.66 /lrn h-‘. By adding SiH, to the deposition system, the growth rate was reduced to a value lower than 0.66 jtrn h-’ in the
J.L. He et al. 1 Materials
Fig. 4. Growth rate of Ti-Si-N
film as a function of SiH, flow rate.
beginning, then increased linearly as a function of SiH, flow rate. The reason probably is that the growth rate of a-S&N4 is higher than that of TIN; however, at a low SiH4 flow rate, the Si, with a smaller radius, enters into the TiN lattice to form a solution, so that a reduction in film volume, i.e., the subsequent film thickness, was observed. Once the amount of added Si exceeds the solubility of Si in the TiN lattice, an a-S&N4 phase appears and a positive effect on film growth is observed, which is exactly what happens to those specimens that are deposited at a SiH4 flow rate higher than 0.91 ml mm-‘. Columnar grains were obtained in all the deposited films. The grain size of the films deposited at low SiH4 flow rates decreased to a very fine grain level compared with the TiN film without Si addition, as observed in Fig. 3(a), (b) and (c), which agreed with the results of Li’s SEM observation. As the SiH4 flow rate was further increased, the grain size increased rapidly and even pores appeared among the columnar grains when the films were deposited at a SiH4 flow rate as high as 7.13 ml min-‘. The coarsening of grain size and pores was expected to degrade the hardness of the deposited Ti-Si-N films. 3.3. TEM analysis
Fig. 5(a) and (b) shows the TEM SAD patterns from two different arbitrarily selected areas of specimen S3. The former ring pattern is, no doubt, diffracted by the TIN phase of which the corresponding planes are labelled on the figure. Comparing the ring pattern from S3 in Fig. 5(a) with that of SO, not shown here, it was found that the two ring patterns were basically the same except that the diameter from S3 was larger, indicating that the TiN lattice constant decreased due to the smaller Si atom substituting for the Ti atom. This
evidence for Si substitution in the TIN lattice agrees with the results of XRD analysis. The spot pattern in Fig. 5(b) was possibly from the phases a-Si,N4 and ,0-Si,N4, which have a h.c.p. structure in common and an almost equal length in the a-axis . The way to distinguish them is to have diffraction from another zone axis, other than [OOOI]. Fig. 5(c) shows the diffraction pattern using [Ol lo] as the zone axis. By calculation, the c-axis length obtained was 5.708 A, which is close to the value of a-S&N, phase identified in the JCPDS card . Qualitatively, the Ti-Si-N film of specimen S3 observed in TEM analysis is the solid soluted TIN as the major phase with a-Si,N, as the minor phase, whose amount is too small to be detected by XRD analysis. The bright field image shown in Fig. 6(a) was obtained from the Ti-Si-N film of specimen S9. The pores existing in between the columnar grains appear as the white region after ion thinning for TEM specimen preparation, while the columnar grains were thinned to form the dark network as shown. The diffraction pattern observed from the dark network is shown in Fig. 6(b), which is the same as the one in Fig. 5(b). 3.4. ESCA analysis
The Si 2p spectrum for the Ti-Si-N film of specimen S2 in Fig. 7 shows that the Si atom in the Ti-Si-N film is in a multi-chemical state, leading to the peak splitting in the spectrum. The suggested designation of these slightly split peaks shows that no free silicon (binding energy = 99.28 eV) was present. Instead, this convoluted peak can be more likely attributed to the Si 2p in S&N, (binding energy = 101.6 eV) and in SiOz (binding energy = 103.5 eV), the latter resulting from the oxygen incorporation usually inevitable in the PECVD process. A semi-quantitative analysis of the deposited films using ESCA as a function of SiH4 flow rate is shown in Fig. 8. It is clear that the atomic concentration of titanium decreases and silicon increases as the SiH4 flow rate is increased, while nitrogen is almost kept constant. The Si concentration in films deposited at low SiH4, though low, confirms that Si definitely incorporates into the TiN film as a solid solution. 3.5. Hardness measurements
Fig. 9(a) and (b) shows the microhardness of the deposited Ti-Si-N film as a function of SiH, flow rate. The trend of hardness as a function of Si concentration in the study is basically the same as the results of Li et al., except that a lower value is obtained. This is due to the higher indentation load employed in this study. The maximum hardness value of the Ti-Si-N film deposited at low SiH, flow rates is far beyond the
J.L. He et al. / Materials Chemistry and Physics 44 (1996) 9-16
TiN .(111) .(200) (220) S(3ll)
, Oil0 I(
loi0 I 1 iicx, 2iio .
Fig. 5. (a), (b) TEM SAD patterns of Ti-Si-N zone axis.
film obtained from two arbitrarily selected area of specimen S3; (c) the same area as (b) with [OllO]
intrinsic hardness value of the deposited TiN. For the films deposited on silicon nitride or alumina, it is clear that the maximum hardnessvalue appears at a flow rate of 0.6-0.7 ml min-’ for specimensAl, A2 or Sl, S2, for which the films were characterized to have the fine grains
of a Si-TiN solid solution as their major phase
with approximately 7-8 at.% Si. The films deposited at
lower SiH4 are thinner than those deposited at higher SiH, flow rates for the same period. Therefore, the intrinsic hardnessvalue of films deposited at lower SiH4 flow rates should be higher than the measured value, becausethe substrate effect is more pronounced. As the SiH, flow rate was further increased,the hardnessvalue abruptly dropped. Grain size coarsening, pores in the
J.L. He et al. / Materials Chemistry and Physics 44 (1996) 9-16
, I 170
eoml To10 *
Fig. 6. (a) TEM bright field image and (b) the corresponding SAD patterns of a Ti-Si-N
f?hn and the appearance of a-S&N, phase are the three possible reasons responsible for this. Unfortunately, the dominant one cannot be distinguished because all of them show up simultaneously. However, the hardness of a TiN f?hn by %-doping is by no means promoted by the appearance of a-S&N,. Instead, it is the solution strengthening by Si atoms, as shown in XRD and TEM
film for specimen S9.
analysis, which comes to the same result if a solid solution of Si is present in the TiN lattice and if the SiH4 flow rate is kept low. This explanation for the promotion of hardness in this study in terms of solution strengthening, discussed above, seems to be more reasonable than the ‘nano-particle hardening’ proposed by Li. The promising hardness value of the TIN film
Fig. 7. ESCA Si 2p spectrum for specimen S3.
SiH4 ‘flow Fig. 8. Ti-Si-N
film composition as a function of SiH, flow rate.
J.L. He et al. / Materials Chemistry and Physics 44 (1996) 9-16
: . :
Ii i ‘,Iiardr1ess 1500
In terms of more detailed characterization and analysis, the microstructure of the multicomponent Ti-Si-N films deposited by RF-PECVD using SiH4 as a silicon source suggeststhat a ‘solution strengthening’ promotes hardness of the Ti-Si-N films deposited at lower SiH, flow rates. Higher SiH4 flow rates induce the precipitation of an a-S&N4 phase in the deposited Ti-Si-N films with pores and coarsened grains and reduce the hardness value. This implies that a superhard coating can be achieved simply by doping trace amounts of Si into the conventional TiN hard coatings deposited by the plasma process. Acknowledgements
The authors wish to thank the National Science Council of Taiwan for their financial support for this study, No. NSC82-0112-COO6-006. References
Fig. 9. Microhardness of the Ti-Si-N film deposited on (a) alumina and (b) silicon nitride as a function of SiH, flow rate.
obtained by Si doping, whatever Si sources are employed, may have potential applications in wear-resistant coatings.
[l] T. Hirai and S. Hayashi, J. Mater. Sci., 17 (1982) 1320-1328.  S. Hayashi, T. Hirai, K. Hiraga and M. Hirabayashi, J. Ma&r. Sci., 17 (1982) 3336-3340.  K. Hiraga, K. Tsuno, D. Shindo, M. Hirabayashi, S. Hayashi and T. Hirai, Philos. Msg. A, 47 (1983) 483-496.  T. Hirai and S. Hayashi, J. Mater. Sci., 18 (1983) 2401-2406.  W. Posadowski, Thin Solid Films, 162 (1988) 111-117.  T. Taniuchi, T. Adachi and K. Kobayashi, &I$ Cont. Techno/., 49 (1991) 13-17.  S. Li, Y. Shi and H. Peng, Plnsmn C/fan, Pl~snrn Process., I.2 (3) (1992) 287-297.  R.C. Evans, An Introduction to Crystnl Chcr~istry, College Book Store, Taipei, 2nd edn., 1969, p, 36.  Joint Committee on Powder Diffraction Standards, No. 9-250.