Biaxially aligned titanium nitride thin films deposited by reactive unbalanced magnetron sputtering

Transcription

1 Surface & Coatings Technology 200 (2006) Biaxially aligned titanium nitride thin films deposited by reactive unbalanced magnetron sputtering S. Mahieu a, *, P. Ghekiere a, G. De Winter a, R. De Gryse a, D. Depla a, G. Van Tendeloo b, O.I. Lebedev b a University of Gent, Department of Solid State Sciences, Krijgslaan 281/S1, 9000 Gent, Belgium b University of Antwerp, Electron Microscopy for Material Research (EMAT), Groenenborgerlaan 171, 2020 Antwerpen, Belgium Received 6 July 2004; accepted in revised form 27 September 2004 Abstract Texture control of sputter-deposited nitride films has provoked a great deal of interest due to its technological importance. However, to our knowledge, studies on the influence of the crystallographic texture of nitride films on the thin film properties have only investigated the influence of the preferred out-of-plane orientation. In this study, we investigated the mechanism responsible for the biaxial alignment in TiN layers. These biaxially aligned TiN layers were deposited by reactive unbalanced (type II) magnetron sputtering on a polycrystalline substrate (stainless steel). The preferred out-of-plane orientation was investigated by carrying out depositions on a nontilted substrate. It was observed that this preferred out-of-plane orientation changed from a (111) to a (002) out-of-plane orientation by increasing the N 2 partial pressure and was unaffected by the variation in substrate bias (0 or +10 V). It was also noticed that biaxially aligned layers can be obtained by tilting the substrate with respect to the incoming material flux. A model for the development of the preferred out-of-plane and in-plane orientation (biaxial alignment) is proposed. D 2004 Elsevier B.V. All rights reserved. PACS: d; Jk; Aj; Cd Keywords: Growth models; Grain growth; Reactive sputtering; Direct current; Magnetron; Titanium nitride 1. Introduction A sputter-deposited polycrystalline layer generally consists of a large amount of small grains. The crystallographic orientation of these grains is fully determined by two different orthogonal crystallographic axes. As a reference, we determine the crystallographic orientation of a grain with his direction perpendicular to the substrate (out-of-plane orientation) and a crystallographic orientation parallel to the substrate (in-plane orientation). The polycrystalline layer has a preferred out-of-plane orientation if nearly all grains have the same out-of-plane crystallographic orientation (which means that all grains have the same crystallographic plane parallel to the substrate plane). When all these grains * Corresponding author. Tel.: ; fax: address: stijn.mahieu@ugent.be (S. Mahieu). not only have the same out-of-plane orientation, but also the same in-plane orientation, the layer is called to be biaxially aligned. To obtain biaxially aligned layers on a nonaligned polycrystalline or amorphous substrate, specific deposition techniques such as ion beam assisted deposition (IBAD) [1] or inclined substrate deposition (ISD) [2] are necessary. In our group, an unbalanced planar magnetron type II is used to deposit biaxially aligned layers [3]. Biaxially aligned thin films of several materials have already been deposited by this deposition technique [4]. In this research, we investigated the mechanism of biaxial alignment in TiN layers. The resulting microstructure and preferred out-of-plane orientation has been investigated by varying the reactive gas partial pressure and the substrate bias. Biaxially aligned layers were obtained by tilting the substrate with respect to the material flux /$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi: /j.surfcoat

2 S. Mahieu et al. / Surface & Coatings Technology 200 (2006) Experimental A planar circular unbalanced magnetron type II (more specific when the outer magnetic field is stronger than the inner magnetic field) is used to deposit biaxially aligned films. The substrate (mirror-like polished nonaligned polycrystalline stainless steel or glass) was mounted on a substrate holder that could be tilted over an angle a (a is the angle between the substrate normal and the direction of the incoming material flux). Uniaxially aligned layers are obtained when the deposition is carried out at a nontilted substrate (a=08), while biaxially aligned layers can be obtained when the substrate is tilted with respect to the material source (ap08). A 2-in. Ti target was sputtered in reactive DC mode, with a mixture of Ar and N 2 (total gas flow=60 sccm). The target substrate distance was 13 cm. The discharge current (I d ) was kept at 0.9 A during deposition. The discharge potential changed from 380 to 392 V with increasing N 2 partial pressure. These sputter conditions resulted in a racetrack of about 10F0.5 cm 2. The base pressure was below 10 3 Pa and the sputter pressure was kept at 0.55 Pa. The substrate temperature was floating, which results in a temperature of about 200 8C [5]. The substrate bias was varied between ground potential and +10 V. The deposition geometry is shown in Fig. 1. A planar disc-shaped probe (built to the model of Greene et al. [6]) with a diameter of 12 mm was used to measure the flux of ions towards the substrate during deposition. By applying a negative bias to this probe, the saturation current of positive ions towards the substrate during deposition could be measured. In this way, the ion-to-atom ratio (this is the ratio between the flux of positive ions arriving at the substrate to the flux of incorporated metallic atoms during the deposition) could be calculated. This ion-to-atom ratio is believed to be an important parameter, influencing the adatom mobility during the deposition [7]. Langmuir probe measurements (Hiden Analytical, with a tungsten wire of 9-mm length and 0.15-mm diameter) have been carried out to determine the plasma potential (V p )and floating potential (V f ) at the substrate position during the deposition. The V p as well as the V f remained nearly constant in the used Ar/N 2 region (V p =+5 Vand V f = 11 V). Fig. 1. Diagram of the used deposition system. 3. Experimental results for depositions on a nontilted substrate 3.1. Observations Two series of TiN layers (~0.85-Am thick) were deposited on a nontilted nonaligned polycrystalline stainless-steel substrate (at ground potential or at a positive bias of +10 V). XRD angular scans (h/2h) revealed that the out-of-plane orientation changed from nearly perfect [111] out-of-plane oriented to nearly perfect [002] out-of-plane oriented with increasing N 2 partial pressure, regardless of the substrate bias. A similar transition from [111] to [002] out-of-plane orientation with increasing reactive gas partial pressure has also been observed by Li et al. [8] and Banerjee et al. [9]. Fig. 2 gives the fraction of [111], [002], [220], and [311] out-of-plane oriented grains, showing the resulting preferential orientation of the thin films grown on a substrate at ground potential or positively biased. This fraction of grains with the (hkl) plane parallel to the substrate surface was calculated by I hkl /I hkl V where I hkl is the measured peak intensity for the (hkl) reflection and I hkl V is the standard peak intensity (ICDD database). TEM cross sections of a TiN layer deposited at Ar/ N 2 =52/7.5 and at Ar/N 2 =43/17 (both grown on a substrate at ground potential) are shown in Fig. 3. The deposition at low reactive gas partial pressure shows a zone T structure (clear overgrowth and faceted grains) while the deposition at high reactive gas partial pressure reveals a zone II structure (no overgrowth and less faceting). It is also visible that the column diameter increased from ~35 to ~120 nm. Two selective area diffraction (SAD) patterns are shown in Fig. 4. These SAD patterns correspond with the lower and upper region of the deposition at Ar/N 2 =43/ 17. As seen in the SAD of the lower region, the preferred out-of-plane orientation is already well established in the first 100 nm of the layer, which is typical for a zone II growth. The SAD pattern of the upper region shows that the [002] out-of-plane alignment is nearly perfect at a layer thickness of ~0.85 Am. The deposition rate and the plasma density in the vicinity of the substrate have been measured for all used deposition conditions, in order to calculate the ion-to-atom ratio. Increasing the reactive gas partial pressure resulted in a decreasing deposition rate, due to poisoning of the target (see Fig. 5). The ion density (sum of all positive ions: Ar +, N +,N + 2,...) in the vicinity of the substrate seemed to be relatively unaffected (within 5%) by the reactive gas partial pressure. As such, the ion-to-atom ratio, which is believed to influence the adatom mobility during the growth in a positive way, increased with increasing reactive gas partial pressure (see Fig. 5), which is in accordance to the observed transition from a zone T to a zone II structure. This increase in ion-to-atom ratio is caused by a reduction in metallic atom flux.

3 2766 S. Mahieu et al. / Surface & Coatings Technology 200 (2006) Fig. 2. XRD h/2h results showing the influence of the reactive gas partial pressure on the preferential out-of-plane orientation of TiN, deposited on a nontilted substrate at ground potential and at a positive bias Discussion Several models of preferred out-of-plane orientation in TiN layers deposited by magnetron sputtering have already been discussed in literature. A first model was described by Pelleg et al. [10] and Oh et al. [11]. They proposed that the preferred orientation is determined by the competition between two thermodynamic forces: the aim to lower the surface energy and the aim to lower the strain energy. They assume that the strain energy is linear with the layer thickness. As such, a [002] out-ofplane oriented layer should be observed in thin the layers, because the [002] plane is the lowest surface energy plane. However, at a critical thickness, the strain energy would exceed the surface energy and because the elastic moduli are anisotropic (E 100 NE 111 ), a transition to a [111] out-of-plane oriented layer should be observed at that critical thickness. However, all XRD results shown in Fig. 2 are from TiN layers with the same thickness (~0.85 Am). Moreover, the predicted transition from [002] to [111] out-of-plane orientation was checked by depositing TiN layers at several thicknesses. Nevertheless, this transition was never observed even in coatings thicker than 1.5 Am. As such, the transition from [111] to [002] observed in Fig. 2 is not the result of a competition between the two thermodynamic forces (aim to lower the surface energy and the strain energy). A second, well-described model has been proposed by Gall et al. [12] and Shin et al. [13]. They investigated the influence of a flux of low energetic N 2 + (E i =~20 ev) on the out-of-plane orientation of TiN and TaN, respectively. A transition from a [111] to a [002] out-of-plane orientation was observed when increasing the ion-to-atom ratio. They assumed in their model that a 20 ev N 2 + will undergo dissociative chemisorption before taking part in the growth. As such, an excess N coverage will be established when the deposition is carried out at high ion(n 2 + )-to-atom ratio. The transition from a [111] to a [002] out-of-plane orientation was calculated to be a consequence of this excess N coverage during the growth. To check this model, the ion energy (e[v p V s ]) as well as the ion-to-atom ratio has been measured for all deposition conditions, used in Fig. 2. The V p was +5 ev, while the substrate bias V s was 0 or +10 V. Hence, N 2 + ions will be accelerated by only 5 ev (which is less than the dissociation energy of N 2 ) or will be even slowed down by 5 ev. Although a quite large ion-to-atom ratio was measured, it is believed that almost no N 2 + ions can undergo dissociative chemisorption at the growing layer. Therefore, the observed transition from a [111] to a [002] out-of-plane orientation in Fig. 2, cannot be explained by the model of Gall and Shin. Moreover, the observation that the substrate bias (0 or +10 V) has no influence on the out-of-plane orientation (see Fig. Fig. 3. TEM cross section of a TiN layer deposited at Ar/N 2 =52.5/7.5 (left) and at Ar/N 2 =43/17 (right). Fig. 4. Selective area diffraction patterns of a TiN layer deposited at Ar/ N 2 =43/17 on a substrate at ground potential, corresponding with Fig. 2, right.

4 S. Mahieu et al. / Surface & Coatings Technology 200 (2006) Fig. 5. Ion-to-atom rate (left) and deposition rate (right) as function of the N 2 partial pressure. 2) emphasize the disagreement with the model of Gall and Shin. A clear transition from a zone T to a zone II microstructure is observed in the TEM cross section of TiN layers deposited at a low and at a high N 2 partial pressure (see Fig. 3). Moreover, the SAD patterns of the corresponding regions of the TEM cross section of the deposition at high N 2 partial pressure, clearly show that the out-of-plane alignment is already present at a layer thickness of 100 nm, which is typical for a zone II structure. This transition in microstructure coincides with a change in preferred orientation from [111] to [002] as shown in Fig. 2. Generally, the transition from a zone T to zone II growth is understood in terms of adatom mobility during the growth. More specific, the transition is believed to occur at a critical ionto-atom ratio, because the ion-to-atom ratio influences the adatom mobility during the growth in a positive way. As such, the transition from a zone T to a zone II structure (and corresponding transition from [111] to [002] out-of-plane orientation) is believed to occur at a certain N 2 partial pressure, because the ion-to-atom ratio increases with increasing N 2 partial pressure as shown in Fig. 5. All depositions carried out with a low N 2 partial pressure, reveal a zone T structure, which means that the [111] out-of-plane orientation is a result of an evolutionary overgrowth. Shin [13] reasoned that the [111] direction is the fastest growing direction of TiN, which is in accordance with the observations. However, the transition to a [002] out-ofplane orientation is believed to be a consequence of the aim to lower the surface energy, which is possible due to a higher adatom mobility during the growth. 4. Depositions on a tilted substrate: biaxial alignment The influence of the target substrate angle (a) was investigated for depositions at low N 2 partial pressure. Earlier depositions on a tilted substrate revealed a clear inplane alignment for yttria-stabilized zirconia and indium tin oxide thin films, deposited by the same deposition technique [4]. In this research, the in-plane alignment of [111] out-ofplane oriented TiN layers has been investigated by optimizing the substrate target angle (a), the target substrate distance, substrate bias, and working pressure. Optimal deposition conditions are summarized in Table 1.A [111] out-of-plane oriented layer with a random in-plane orientation exhibits a ring of non-modulated intensity in a (002) pole figure (see Fig. 6, left). However, in case of a deposition on a tilted substrate, there are three separate peaks visible in the (002) pole figure, indicating a preferential in-plane alignment (hence, a biaxially aligned layer, see Fig. 6, right). The full width at half maximum (FWHM) in phi of the poles in the (002) pole figure of the biaxially aligned layers reached a value of 198. This value of Fig. 6. (002) pole figure of TiN layer deposited on a nontilted (left) and a tilted (right) substrate. The arrow indicates the direction of the incoming material flux. Table 1 Deposition conditions of TiN layer with in-plane alignment of 198 in phi T S distance 9 cm discharge current 0.9 A T S angle a 658 substrate bias +15 V Working pressure 0.48 Pa Ar/N 2 55/5 Fig. 7. SEM plan-view and cross section of biaxially aligned TiN. The arrow indicates the direction of the incoming material flux.

5 2768 S. Mahieu et al. / Surface & Coatings Technology 200 (2006) FWHM is normally used as a measure for the degree of inplane alignment. The smaller the FWHM of the peaks, the better the in-plane alignment. A SEM plan-view and cross section of this layer are shown in Fig. 7. The h/2h (not shown here) revealed that tilting the substrate did almost not influence the out-of-plane alignment [only a small tilt of ~48 of the (111) direction away from the incoming material flux]. It is visible in Fig. 7 that the TiN layer deposited on a tilted substrate also reveals a zone T structure, but that the columns have a tilt angle towards the substrate. Hence, the tilt angle of the (111) orientation and the tilt angle of the columns are not the same (tilted in the opposite direction). It can be concluded that tilting the substrate with respect to the material flux mainly influences the microstructural and crystallographic in-plane alignment (as seen in the SEM plan view and pole figure). It is believed that this in-plane alignment is caused by an overgrowth mechanism in which the [111] out-of-plane oriented grains with a good in-plane alignment overgrow the other [111] out-of-plane oriented grains. This is based on an anisotropy in growth rate, caused by the shape of the grains. These grains grow namely according to a specific crystal habit (as seen in the plan view). A sketch of this specific crystal habit is drawn in Fig. 8, based on Figs. 6 and 7. Evidently, the grains are terminated by planes of lowest energy ({001} planes). It can be calculated that the grains with a {001} plane facing the material flux will grow faster perpendicular to the substrate than the other in-plane oriented grains, because they catch more metallic adatoms. An analogous calculation has been done and described in Ref. [14]. As such, the microstructural and crystallographic in-plane alignment are correlated due to the fact that both inplane alignments are a direct consequence of the fact that all grains in zone T grow according to a specific crystal habit. 5. Conclusions Fig. 8. Sketch of the observed TiN crystal habit. Biaxially aligned TiN layers have been deposited by unbalanced magnetron sputtering. The mechanism of outof-plane alignment is investigated by performing depositions on a nontilted substrate. A transition from a [111] to a [002] out-of-plane alignment has been observed with increasing N 2 partial pressure. This transition is believed to be the consequence of increasing adatom mobility during the deposition. The [111] out-of-plane orientation is a result of evolutionary overgrowth, while the [002] out-of-plane orientation is a result of the aim to lower the surface energy. Depositions at low N 2 partial pressure on a tilted substrate revealed a microstructural and crystallographic in-plane alignment, conserving the [111] out-of-plane orientation. The in-plane alignment is caused by anisotropy in growth rate of the different in-plane-oriented grains. This anisotropy in growth rate is a consequence of the fact that all grains grow according to a specific crystal habit, if the deposition is carried out in zone T regime. Acknowledgments The authors would like to thank the biwt-vlaanderenq for funding this research. We also would like to thank Olivier Janssens of the University of Gent for the XRD measurements. References [1] Y. Ijima, N. Tanabe, O. Kohno, Y. Ikeno, Physica, C (1991) [2] K. Hasegawa, Y. Nakamura, T. Izumi, Y. Shiohara, Physica, C (2001) 967. [3] G. De Winter, J. Denul, R. De Gryse, IEEE Trans. Appl. Supercond. 11 (1) (2001) [4] G. De Winter, S. Mahieu, I. De Roeck, R. De Gryse, J. Denul, IEEE Trans. Appl. Supercond. 13 (2) (2003) [5] G. De Winter, J. Denul, R. De Grys, IEEE Trans. Appl. Supercond. 11 (1) (2001) [6] I. Petrov, F. Adibi, J.E. Greene, W.D. Sproul, W.D. Mqnz, J. Vac. Sci. Technol., A 10 (5) (1992) [7] L. Hultman, G. Hâkansson, U. Wahlstrfm, J.E. Sundgren, I. Petrov, F. Adibi, J.E. Greene, Thin Solid Films 250 (2) (1991) 153. [8] T.Q. Li, S. Noda, Y. Tsuji, T. Ohsawa, H. Komiyama, J. Vac. Sci. Technol., A 20 (3) (2002) 583. [9] R. Banerjee, K. Singh, P. Ayyub, M.K. Totlani, A.K. Suri, J. Vac. Sci. Technol., A 21 (1) (2003) 310. [10] J. Pelleg, L.Z. Zervin, S. Lungo, N. Croitoru, Thin Solid Films 197 (1991) 117. [11] U.C. Oh, J.H. Je, J. Appl. Phys. 74 (3) (1993) [12] D. Gall, S. Kodambaka, M.A. Wall, I. Petrov, J.E. Greene, J. Appl. Phys. 93 (11) (2003) [13] C.S. Shin, D. Gall, Y.W. Kim, N. Hellgren, I. Petrov, J.E. Greene, J. Appl. Phys. 92 (9) (2002) [14] S. Mahieu, G. De Winter, D. Depla, R. De Gryse, J. Denul, Surf. Coat. Technol. 187 (2004) 122.