APL Materials

Anomalous epitaxial stability of (001) interfaces in ZrN/SiNx multilayers

Naureen Ghafoor, Hans Lind, Ferenc Tasnâdi, Igor A. Abrikosov, and Magnus Odén

Citation: APL Materials 2, 046106 (2014); doi: 10.1063/1.4870876 View online: http://dx.doi.Org/10.1063/1.4870876

View Table of Contents: http://scitation.aip.org/content/aip/joumal/aplmater/2/4?ver=pdfcov Published by the AIP Publishing

Articles you may be interested in

Self-organization during growth of ZrN/SiNx multilayers by epitaxial lateral overgrowth J. Appl. Phys. 114, 224302 (2013); 10.1063/1.4838495

Growth of Co Si 2 on Si(001) by reactive deposition epitaxy J. Appl. Phys. 97, 044909 (2005); 10.1063/1.1774263

First-principle study on GaN epitaxy on lattice-matched ZrB 2 substrates Appl. Phys. Lett. 83, 2560 (2003); 10.1063/1.1613353

Epitaxial structure SrTiO 3 011 on Si001

J. Appl. Phys. 89, 5421 (2001); 10.1063/1.1357461

Preparation and characterization of superhard CN x / ZrN multilayers J. Vac. Sci. Technol. A 15, 946 (1997); 10.1116/1.580784

SUBSCRIBE TO

/V\ CrossMark

VH# ^-c^ckfor updates

Anomalous epitaxial stability of (001) interfaces in ZrN/SiNx multilayers

Naureen Ghafoor,1 Hans Lind,2 Ferenc Tasnadi,2 Igor A. Abrikosov,2 and Magnus Oden1

1 Department of Physics, Chemistry, and Biology (IFM), Nanostructured Materials, Linkoping University, SE-581 83 Linkoping, Sweden

2 Department of Physics, Chemistry, and Biology (IFM), Theory and Modelling Division, Linkoping University, SE-581 83 Linkoping, Sweden

(Received 20 February 2014; accepted 31 March 2014; published online 14 April 2014)

Isostructural stability of B1-NaCl type SiN on (001) and (111) oriented ZrN surfaces is studied theoretically and experimentally. The ZrN/SiNx/ZrN superlattices with modulation wavelength of 3.76 nm (dSiNx~0.4 nm) were grown by dc-magnetron sputtering on Mg0(001) and MgO(111). The results indicate that 0.4 nm thin SiNx layers utterly influence the preferred orientation of epitaxial growth: on Mg0(001) cube-on-cube epitaxy of ZrN/SiNx superlattices were realized whereas multilayers on Mg0(111) surface exhibited an unexpected 002 texture with a complex fourfold 90°-rotated in-plane preferred orientation. Density functional theory calculations confirm stability of a (001) interface with respect to a (111) which explains the anomaly. © 2014 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution 3.0 Unported License. [http://dx.doi.org/10.1063/L4870876]

The structural understanding of transition metal nitride nanocomposites is of interest due to their use as protective hard coatings at extreme conditions, e.g., metal cutting tools. The nc-TiN-SiNx (nc: nanocrystalline) is considered as a model system in this field and it has been studied extensively.1-4 The hardness enhancement in superhard nanocomposites is attributed to a percolative nature of phase segregation, which results in coverage of the transition metal nanocrystals with approximately one monolayer (ML) of the non-polar Si3N4. The morphology and bonding relationship of a crystalline SiNx tissue phase with nc-TiN is thus debated.5,6 Alling et al. predicted the dynamical theoretical stability of two pseudo-B3 Si3N4 phases derived from L12- or D022-type distribution of Si vacancies7 with respect to the B1-TiN lattice. In a follow-up study, Marten et al. showed that the stoichiometric interface of 1:1 SiN on TiN(001) is dynamically unstable, as indicated by the presence of imaginary frequencies in the phonon dispersion relations calculated for this system, whereas on (111) interface B1-SiN is stable.8 The calculations were performed on a monolayer of SiNx sandwiched isostructurally between B1-TiN(001) and (111) oriented slabs. Experiments also confirmed that the main crystallographic relationship between polycrystalline Si3N4 and TiN determined from HRTEM analysis is [0001] Si3N4||[110] TiN with (10 10) Si3N4||(111) TiN.9

The Zr-Si-N composites are gaining large scientific interest due to their potential usage in industrial protective coatings for metal cutting tools. Structurally they are very similar to nc-TiN/a-Si3N4 (a: amorphous), yet behave differently during phase segregation leading to composite formation.10,11 Owing to a larger atomic radius compared to N, it is energetically unfavorable for Si atoms to occupy either N sites or interstitials in ZrN lattice. Substitutional replacement of Zr by Si causes a large lattice distortion as Si content increases up to 3 at. %. The enhanced structural distortions in c-ZrN compared to c-TiN when adding 2 at. % Si, illustrated by Martin et al.,12 support a strong immiscibility of Si in c-ZrN. A comparatively stronger percolation of c-ZrN and SiNx-rich phases with larger lattice mismatch could lead to different morphology and bonding structure of the SiNx

tissue phase than what have been discussed so far for nanocomposites of nc-TiN/SiNx. There is one

|@ ® |

2166-532X/2014/2(4)/046106/5 2,046106-1 ©Author(s) 2014 IbiH^

scientific report in which Dong et al. have shown the stabilization of a rock salt-like pseudocrystal structure of 0.6 nm thin SiNx layer in ZrN/SiNx multilayers13 similar to what was previously shown for TiN/SiNx multilayers.14'15 However, unlike TiN/SiNx multilayers which preferably attain 111 texture, ZrN/SiNx multilayers exhibited 001 preferred orientation along the growth direction when deposited without any intentional heating or ion assistance and no pseudomorphic epitaxial forces of the substrates were present.

In this letter we report experimental and theoretical investigation of the stability of (001) and (111) interfaces of ZrN with ~1 ML of SiNx. We observe experimentally that the multilayers exhibit strong 002 texture regardless of the single crystal MgO substrate orientation. First principle density functional theory (DFT) calculations confirm much higher stability of 001 ZrN interfaces over 111 interfaces with SiNx, indicating that the effect of lattice vibrations is a cause of such a preference in growth orientation.

Periodic stacks of ZrN/SiNx multilayers consisting of 30 bilayers with a modulation period of 3.76 nm with approximately 0.4 nm thin SiNx layer (corresponding to 1ML of B1-SiN) were deposited at 800 °C on polished 10 x 10 mm2 MgO (001) and MgO (111) substrates. The temperature was optimized for good epitaxy of ZrN layers on MgO substrates. Deposition was made simultaneously on both substrates using high vacuum dc reactive unbalanced magnetron sputtering from pure Zr (99.9%) and Si (99.999%) targets in an Ar/N2 atmosphere at a total pressure of 4.5 mTorr. A high flux of low energy ion bombardment16 of the sputtering gas ions was utilized and substrates were held at floating potential of ~-18 V. The X-ray diffractometry (XRD) and pole figure analysis was performed with a PANalytical Empyrean diffractometer using Cu-Ka radiation. The diffractometer was configured with a mirror and a 0.5° divergence slit as primary optics and a parallel plate col-limator (0.27°) with a PIXcel3D detector on the secondary side. A FEI Tecnai G2 TF 20 UT FEG microscope operated at 200 kV was used to obtain high-resolution TEM cross sectional microscopy (HR-XTEM) images and selected area electron diffraction (SAED) patterns.

Figure 1 shows 20-m (20 = 30°-90°) scans of ZrN/SiNx layers grown on Mg0(001) and Mg0(111) substartes. The layers on Mg(001) in Fig. 1(a) show only 002 and 004 ZrN reflections represnting strong 002 texture of the film. The first order superlatice reflection (*) is the evidence for a strong compositional modulation of the periodicity of 3.76 nm calculated by using the expression

A = (X/20* - sin 0ZrN002)-

The film grown on Mg0(111) substrate shows a weak 111 reflection while the dominant reflections still are from 002 and 004 ZrN in Fig. 1(b). The inset (in red) shows the diffractogram (20 = 30°-45°) from a single layer of 1.2 ¡m thick ZrN film grown on Mg0(111). The film exhibits prominent 111-reflection and a comparatively low intensity 002 reflection. This confirms the possibility of stabilizing a 111-texture of ZrN on Mg0(111). The 111 pole figure, Fig. 1(c), of the ZrN/SiNx film on Mg0(111) contains 6 peaks (12 in a complete figure) located at ^ = 54.7° and separated by y = 30°. These peaks emanate from (111) planes of 002 oriented grains in three equivalent directions opposite to what has been observed for 111 oriented film of ScN on Mg0 (001).17 Here, 002 textured film grows with a complex fourfold 90°-rotated in-plane preferred orientation such that square {001} ZrN/SiNx fit with triangular (111) planes of Mg0. The central strong intensity peak at (^ = 0°, y = 0°) is originating from the nearby substrate peak as no strong 111 reflection is observed in Fig. 1(b).

The lattice resolved TEM image and superlattice reflections (*) in SAED in Figure 2(a) are evident of 002 epitaxial growth of ZrN/SiNx layers with the underlying substrate, supporting the XRD. However, unlike TiN/SiNx structure18 which grows with abrupt interfaces, nonhomogeneous distributions of dark regions interlaced in an overall cubic lattice, investigated as segregated SiNx precipitates,19 are present here. No distinct SiNx layer sites can be marked in multilayers grown on Mg0(111) substrate in the overview image in Figure 2(b). The overall lattice does not mimic the (111) orientation of the substrate and instead still exhibits 002 texture in the growth direction evident in the SAED pattern. The less pronounced periodic structure is reflected by the weak superlattice reflections around 002 spots in SAED.

In order to understand the anomalous epitaxial stability of (001) interfaces in ZrN/SiNx multilayers, we have performed ab initio calculations in the framework of DFT as implemented in

ZrN/MgO(111)

A.......

N=30 L=3.76 nm

Mg0(001)

30 35 40 45 50 55 60 65 70 75 80 85 90 Scattering angle, 20

FIG. 1. X-ray diffractogram from 30 periods of ZrN/SiNx superlattices grown on (a) Mg0(001) and (b) MgO(111) substrates (marked S). The periodicity is measured to 3.76 nm from the superlattice (*) andZrN(002)/Mg0(001) Bragg reflection peaks' positions. Inset (red profile) in (b) shows a fraction of X-ray diffractogram from a 1.2 /u,m thick ZrN/Mg0(111) film. The dashed lines are drawn at the positions of relaxed lattice spacing of ZrN lattice. (c) Two top quadrants of XRD 111 pole figure (< = 0°-180°, ^ = 0°-90°), obtained with 26 = 33.89° fromZrN/SiN film grown on Mg0(111).

the QUANTUM-ESPRESSO (QE) package.20 The exchange-correlation energy functional was approximated by the generalized gradient approximation according to Perdew-Burke-Ernzerhof.21 The pseudo-potentials were taken from the library linked to QE. The phonon spectra were calculated within the harmonic approximation by density functional perturbation theory.22

Interfaces in the calculations were modeled by periodic slab structures along the surface normal with (1x1) and (2x2) in-plane periodicity. The latter gives larger degree of freedom for the atomic relaxations and allows modeling the 3:4 SiN stoichiometry by introducing a Si vacancy, as the 1:1 stoichiometric B1-SiN was shown to be unstable thermodynamically.7 Using the (1x1) supercell we found that the Si-atoms in the (001) interface are highly unstable in the ideal B1 position, similar to the TiN/SiN (001) case.5,6,23 The Si-atom after full relaxation moved within the (x-y)-plane a distance of 14% of the cell size. The same amount of Si relaxation, though with a more complex but still regular pattern, was obtained by using (2 x 2) in-plane supercell. Figure 3(a) shows the significant shift of the Si atoms off their ideal positions in the (001) interface. At the same time, the phonon calculation for the (1 x 1) in-plane supercell simulating (001) ZrN/SiN interface showed that the fully relaxed system with 1:1 SiN stoichiometry is dynamically stable, though structurally disordered. On the contrary, the (111) interface with 1:1 stoichiometry was found to show significantly different behavior. For example, the convergences of the ionic relaxations were slow and the calculations resulted in heavy and irregular distortions. The increase of the in-plane supercell from (1 x 1) to (2x2) did not help in

. V . - —

y&-. . aagù sm-

• J: , .¡àj •

_w.v ' IKÉ! :?! -

, ¡yiSiii'!'-'-'5' %

-►sin

a- tsishaai«

-yr" dpi S » '• iHt^fS®^»'» •:. .-¿ft;' Mg0(001) [100]

FIG. 2. Transmission electron micrographs showing first 6 periods of ZrN/SiN superlattices grown on (a) Mg0(001) and (b) Mg0(111) substrates. Position of SiNx layers is marked with arrows. The corresponding selected area electron diffraction patterns (SAED) and lattice resolved images are also shown.

(a) (001) 2x2

&..............m

(c) (001) 2x2 vac

%............«P

(b) (111) 2x2

(d) (111) 2x2 vac

- -m-*

FIG. 3. Atomic configuration of the interfacial Si-atoms (green) and N-atoms (blue) in the (2x2) in-plane model of ZrN/SiN interface relative to the basal lattice vectors (empty circles): (a) (001) interface without vacancy, (b) (111) interface without vacancy, (c) (001) interface with vacancy, and (d) (111) interface with vacancy. Empty circles mark ideal crystal positions and filled circles are positions after relaxations. Note that, while in the (001) interface all displayed atoms are in the same plane, in the (111) interfaces they are distributed over three layers (see supplementary material24 for more discussion). The image shows the atoms' coordinates relative to the given basis vectors, in the (111) interface structures the basis is actually hexagonal even though the coordinates are represented on a square basis here.

finding a stable relaxed Si positions but further complicated the situation by showing highly irregular atomic configurations. Moreover, the relaxation path and final configuration of Si positions were greatly dependent on the initial applied distortion of the atomic positions (see the supplementary material24) and slab thickness. Figure 3(b) shows one of the obtained interface structure for (111)

ZrN/SiN interface. The calculated phonon spectra of the (111) interface for (1x1) supercell, both with ideal lattice configuration and with the configurations obtained after static relaxations, showed imaginary phonon frequencies across the entire spectra. This further confirms extreme instability of (111) ZrN/SiN interface.

Taking Si vacancies into consideration is known to stabilize the interface in TiN/SiN system.8,23 However, our calculations with 3:4 SiN stoichiometry also showed the behavior for the interface stability that was similar to the one obtained with 1:1 stoichiometry. Namely, it was found that the (001) interface relaxed easily into a stable configuration, see Fig. 3(c), while in the case of (111) we failed to determine a distinct converged atomic configuration, see Fig. 3(d). We conclude that both the slow convergence of the forces and the strong dependence of the relaxed atomic positions on the initial conditions observed for the (111) ZrN/SiN interface indicate a very complex energy landscape for this system (see the supplementary material24 for further details about the structural models, phonon calculations, and energy landscape). In fact, this complexity results in the dynamical instability of the (111) ZrN/SiN interface, identified by the presence of imaginary frequencies in this phonon dispersion relations (see Fig. S2 in the supplementary material24). The latter could lead to the observed effects during deposition experiments in which one attempts to grow ZrN/SiN superlattices with (111) orientation.

In conclusion we have substantiated the dynamical stability of SiNx in ZrN/SiNx superlattices grown on 111 and 001 oriented MgO template. Both XRD and TEM investigations strongly indicate that one monolayer of SiNx stabilizes the epitaxial growth on 001 oriented substrates where on MgO(111) it transforms the growth from 111 to a strong 002 texture with a complex fourfold 90°-rotated in-plane preferred orientation such that square {001} ZrN/SiNx fit with triangular (111) planes of MgO. First principles DFT calculations support the dynamical instability of the (111) ZrN/SiN interface compared to (001). These results are unlike the bonding behavior known for nc-TiN/SiNx interfaces and thus can help in designing nc-ZrN/SiNx nanocomposites and nanolaminates.

The Strategic Research (SSF) project Designed Multicomponent Coatings, Multifilms and VINNOVA Strategic Faculty Grant VINNMER—Marie Curie Chair are acknowledged for financial support. I. Abrikosov is grateful to the SSF program SRL10-0026 and the Swedish Research Council (VR) Project No. 621-2011-4426. The computations were performed on resources provided by the Swedish National Infrastructure for Computing (SNIC) at the National Supercomputer Center (NSC) and PDC Center for High Performance Computing.

1 S. Veprek, M. G. J. Veprek-Heijman, P. Karvankova, and J. Prochazka, Thin Solid Films 476(1), 1-29 (2005).

2 S. Veprek and S. Reiprich, Thin Solid Films 268(1-2), 64 (1995).

3L. Hultman, J. Bareno, A. Flink, H. Soderberg, K. Larsson, V. Petrova, M. Oden, J. E. Greene, and I. Petrov, Phys. Rev. B 75, 155437 (2007).

4 A. Niederhofer, T. Bolom, P. Nesladek, K. Moto, C. Eggs, D. S. Patil, and S. Veprek, Surf. Coat. Technol. 146-147, 183 (2001).

5 R. F. Zhang, A. S. Argon, and S. Veprek, Phys. Rev. Lett. 102, 015503 (2009).

6R. F. Zhang, A. S. Argon, and S. Veprek, Phys. Rev. B 79, 245426 (2009).

7 B. Alling, E. I. Isaev, A. Flink, L. Hultman, and I. A. Abrikosov, Phys. Rev. B 78, 132103 (2008).

8 T. Marten, E. I. Isaev, B. Alling, L. Hultman, and I. A. Abrikosov, Phys. Rev. B 81, 212102 (2010).

9C. Iwamoto and S. Tanaka, J. Am. Ceram. Soc. 81(2), 363-368 (1998).

10 T. Mae, M. Nose, M. Zhou, T. Nagae, and K. Shimamura, Surf. Coat. Technol. 142-144, 954 (2001).

11 S. H. Sheng, R. F. Zhang, and S. Veprek, Acta Mater. 59, 297 (2011).

12 P. J. Martin, A. Bendavid, J. M. Cairney, and M. Hoffman, Surf. Coat. Technol. 200, 2228 (2005).

13 Y. Dong, W. Zhao, J. Yue, and G. Li, Appl. Phys. Lett. 89, 121916 (2006).

14 X. Hu, H. Zhang, J. Dai, G. Li, and M. Gu, J. Vac. Sci. Technol. A 23, 114 (2005).

15 H. Soderberg, A. Flink, J. Birch, P. O. A. Persson, M. Beckers, L. Hultman, and M. Oden, J. Mater. Res. 22, 3255 (2007).

16 N. Ghafoor, F. Eriksson, P. O. A. Persson, L. Hultman, and J. Birch, Thin Solid Films 516, 982 (2008).

17 D. Gall, I. Petrov, L. D. Madsen, J.-E. Sundgren, and J. E. Greene, J. Vac. Sci. Technol., A 16, 2411 (1998).

18 H. Soderberg, J. M. Molina-Aldareguia, T. Larsson, L. Hultman, and M. Oden, Appl. Phys. Lett. 88, 191902 (2006).

19 A. Fallqvist, N. Ghafoor, H. Fager, L. Hultman, and P. O. A. Persson, J. Appl. Phys. 114, 224302 (2013).

20P. Gianozzi etal., J. Phys.: Condens. Matter 21, 395502 (2009).

21 J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865 (1996).

22 S. Baroni, S. de Gironcoli, A. D. Corso, and P. Gianozzi, Rev. Mod. Phys. 73, 515 (2001).

23 T. Marten, B. Alling, E. I. Isaev, H. Lind, F. Tasnadi, L. Hultman, and I. Abrikosov, Phys. Rev. B 85, 104106 (2012).

24 See supplementary material at http://dx.doi.org/10.1063/1.4870876 for details about the structural models, phonon calculations, and energy landscape.