molybdenum thin films for application as interdigital transducers on high temperature stable piezoelectric substrates La3Ga5SiO14 and Ca3TaGa3Si2O14

molybdenum thin films for application as interdigital transducers on high temperature stable piezoelectric substrates La3Ga5SiO14 and Ca3TaGa3Si2O14

Materials Science and Engineering B 202 (2015) 31–38 Contents lists available at ScienceDirect Materials Science and Engineering B journal homepage:...

2MB Sizes 2 Downloads 62 Views

Materials Science and Engineering B 202 (2015) 31–38

Contents lists available at ScienceDirect

Materials Science and Engineering B journal homepage: www.elsevier.com/locate/mseb

Tungsten/molybdenum thin films for application as interdigital transducers on high temperature stable piezoelectric substrates La3 Ga5 SiO14 and Ca3 TaGa3 Si2 O14 Gayatri K. Rane a,b,∗ , Siegfried Menzel a,b , Marietta Seifert a,b , Thomas Gemming a,b , Jürgen Eckert a,c a

Leibniz Institute for Solid State and Materials Research Dresden, Institute for Complex Materials, P.O. Box 27 01 16, D-01171 Dresden, Germany SAWLab Saxony, Dresden, Germany c Technical University Dresden, Institute of Materials Science, D-01062 Dresden, Germany b

a r t i c l e

i n f o

Article history: Received 18 May 2015 Received in revised form 24 July 2015 Accepted 18 August 2015 Available online 29 August 2015 Keywords: High temperature sensor IDT material Refractory metal films Electrical resistivity Thin film microstructure Multilayers

a b s t r a c t Sputter-deposited single, bi- and multilayers of W and Mo on Si substrate and high temperature stable piezoelectric substrates La3 Ga5 SiO14 (LGS) and Ca3 TaGa3 Si2 O14 (CTGS) have been studied as electrode material for high temperature applications of surface acoustic wave (SAW) devices up to 800 ◦ C. We show for the first time that the film resistivity lowers with decreasing the individual layer thickness of W in the W/Mo multilayer stack. This has been attributed to the low electron mean free path of W of about ∼4 nm as well as low electron scattering of the electrons at the W–Mo interface as a result of the formation of coherent interfaces. The stability of the films on Si and CTGS has been demonstrated up to 800 ◦ C while the films on the LGS substrate fail already at 600 ◦ C due to the inherent instability of the LGS substrate under vacuum annealing. © 2015 Published by Elsevier B.V.

1. Introduction Most metallic films undergo severe creep deformation even under low external stress which makes them unsuitable for applications already at intermediate temperatures ([1–3] and references therein). In this aspect, in contrast to the occurrence of creep in lowmelting metals, refractory metals exhibit lower rate of thermally activated degradation and as such low creep behavior. In particular, W–Mo multilayers and their alloys have been studied for their excellent thermal conductivity, high-temperature strength, good mechanical properties and low electrical resistivity making them technologically relevant [4–8]. With similar material requirements in the field of surface acoustic wave (SAW) devices as temperature sensors, recent investigations are focused on the search for electrode materials for high-temperature (>500 ◦ C) applications [9–13]. These device consists of metallic interdigital transducer (IDT) electrodes combs structured on a piezoelectric substrate. An electric RF

∗ Corresponding author at: Leibniz Institute for Solid State and Materials Research Dresden, Institute for Complex Materials, P.O. Box 27 01 16, D-01171 Dresden, Germany. Tel.: +49 3514659842. E-mail address: [email protected] (G.K. Rane). http://dx.doi.org/10.1016/j.mseb.2015.08.007 0921-5107/© 2015 Published by Elsevier B.V.

voltage applied between the two opposing electrodes is converted into a SAW and vice versa because of the piezoelectric effect of the substrate material [14,15]. The sensor works on the principle that any physical quantity influencing the SAW can be sensed based on the modification of the electric signal. For over two decades, research in this field has been predominantly focused on noble metals, especially Pt, IDTs on the high temperature stable piezoelectric substrate La3 Ga5 SiO14 (LGS) [11,12,16–18]. Most of these studies have been dedicated toward stabilizing the Pt thin film on LGS to avoid diffusion related damaging effects such as delamination and agglomeration which get more pronounced at higher temperatures. Addition of pinning and adhesion layer leads to improvement in the film stability, however, the deterioration is intrinsic to the inert Pt film. Other electrode materials have been studied such as Ir which undergoes severe oxidation above 700 ◦ C also upon vacuum annealing by taking up oxygen from the LGS substrate [19]. Thus the realization of high temperature stable metallizations has limitations with noble metals and additionally the film-substrate composite stability is also subject to the chemical interaction in between the two. Although LGS has been utilized for high temperature applications, studies have indicated that its chemical stability is highly sensitive to the surrounding atmosphere. Especially under vacuum or at low oxygen

32

G.K. Rane et al. / Materials Science and Engineering B 202 (2015) 31–38

partial pressures, high rate of diffusion of oxygen and gallium atoms was observed already by about 650 ◦ C [20–22]. A relatively new substrate for SAW applications, Ca3 TaGa3 Si2 O14 (CTGS), belonging to the LGS family has been recently studied for its higher stability, superior piezoelectric properties (higher effective piezoelectric coefficient at high temperature) and better crystal quality (with respect to stoichiometry) as compared to LGS [23–25]. This relatively lesser known substrate has a higher mechanical strength and a lower thermal expansion anisotropy as compared to LGS. In the present paper, we report studies on W and Mo multilayer thin films as electrode material on Si (reference substrate), LGS and CTGS substrates. These two high melting metals have exceptional thermal shock resistance due to their low coefficients of thermal expansions (CTE) combined with their high thermal conductivity (W: 174 W m−1 K−1 , Mo: 139 W m−1 K−1 ) besides having high-temperature creep resistance. In addition, the CTE of W (4.5 × 10−6 K) and Mo (4.8 × 10−6 K) are close to that of LGS (a11 = 5.63 × 10−6 K) and CTGS (a11 = 3.3 × 10−6 K). W and Mo are also interesting because they are mutually soluble over the entire composition range with a low lattice misfit over a large temperature range due to very similar lattice constants and an excellent match of the CTEs. The many advantages of this system, especially for high temperature applications and the lack of knowledge on these thin film multilayers have initiated these studies for further application of these materials as IDTs in SAW devices. The paper focuses on the thermal stability and electrical characteristics of these films up to a temperature of 800 ◦ C under vacuum condition. 2. Experimental procedure 2.1. Film deposition Thin films of Mo, W and their multilayers with different layer thicknesses making up to a total film thickness of 100 nm were prepared by magnetron sputtering in a HV chamber with base pressure of 1.7 × 10−4 Pa using two dc planar magnetrons with a tungsten and a molybdenum target (both with 99.95% purity). Pure Ar (99.999%) was used as the sputtering gas. The metals were deposited onto (100)-oriented single crystalline silicon substrates with 1000 nm of thermally grown silicon oxide layer which is used as a reference substrate for the SAW technology (henceforth referred to as Si). The oxide layer acts as a diffusion barrier during annealing to avoid undesirable reaction of the film material with the Si substrate, e.g. silicide formation. The substrate was rotated at a constant speed of 10 rpm during the deposition in order to improve the homogeneity of the deposited film. The films were deposited at 400 ◦ C substrate temperature. Annealing experiments on the deposited films were carried out in a tube furnace under high-vacuum (∼10−3 Pa) by heating at 400 ◦ C, 600 ◦ C and 800 ◦ C for 12 h each. Based on the results on Si substrate, some of the films were studied on the piezoelectric substrates LGS (138.5◦ cut) and thereafter on CTGS (Y-cut). 2.2. Measurements The film thickness was determined by a profilometer and using X-ray reflectivity (XRR) measurements. XRR measurements were carried out on a Panalytical X’Pert MRD thin film device with parallel beam geometry and equipped with an Eulerian cradle. The diffractometer consists of a copper X-ray tube, an X-ray lens in the primary beam and a parallel-plate collimator in the diffracted beam. The estimates for the film thickness, roughness and density were obtained by fitting of the measured profile using the X’Pert Reflectivity v1.1 program. The electrical resistivity of the films was determined by the four point probe measurement technique (van

Table 1 Electrical resistivity of the as-deposited films on Si/SiO2 , LGS and CTGS substrates. The residual stress obtained by the wafer curvature method is tabulated for the films on the Si/SiO2 substrate. The subscript denotes the individual layer thickness and S denotes the substrate. Film

Stress/MPa

Electrical resistivity/␮ cm

Si(1 0 0)

Si(1 0 0)

LGS

CTGS

W100 –S (@ 0.18 nm/s) Mo100 –S (@ 0.92 nm/s) W50 –Mo50 –S Mo50 –W50 –S W20 –Mo20 –W20 –Mo20 –W20 –S (Mo10 –W10 )5 –S (W10 –Mo20 )3 –W10 –S (Mo20 –W5 )4 –S

−2100 −15 −560 300 −1200 −1550 −1320 −1020

13.6 15 17.5 12.4 12.8 12.2 10.3 10.5

15.4 17.8 20.1 12.1 12.3 12 11.1 10.9

14.2 14.6 15.4 11.7 11.6 11.6 10.7 10.5

der Pauw method) on a 1 × 1 cm2 sample [26]. Evaluation of the data showed that the electrical measurement errors are in the order of ±0.3 ␮ cm and the film thickness could be determined to an accuracy of ±1–2 nm using XRR and profilometry which is of the same range as the roughness. The measurement of residual stresses was carried out only for the films deposited on the Si wafers, based on the wafer curvature technique, on a Tencor Flexus-2320 system using dual wavelengths (670 nm and 750 nm). The average stress in the film was then calculated from the change in the curvature of the Si substrate (300 ␮m thickness) before and after film deposition using the Stoney’s equation [27,28]. The surface topography of the as-deposited and annealed samples were imaged using atomic force microscopy (AFM, DI Dimension 3100) and scanning electron microscopy (Zeiss Ultra 55 Plus equipped with an in-lens detector which makes it possible to obtain high-resolution surface images at low working distance <10 mm using secondary electrons and operated at 10 keV accelerating voltage). The film-substrate interface was viewed from the cross sections prepared by focussed ion beam (FIB, Zeiss 1540 XB Crossbeam) cuts. Auger electron spectroscopy (AES) sputter depth profiling was used to gather information on the contaminants on the films with 1 kV Ar+ ions at 0.7 ␮A. 3. Results and discussion 3.1. As-deposited films Table 1 shows a summary of the thin films that were deposited (the subscript denotes the individual layer thickness in nm and S denotes the substrate) and the measured properties. The criteria for selection of the layer thickness in the multilayer stack was based on achieving the lowest electrical resistivity. Low electrical resistivity was achievable up to a minimum layer thickness of 10 nm for Mo and as low as 5 nm for W. Fig. 1 shows the SEM micrographs of the as deposited 100 nm thick films on Si. The morphology of the bilayer and multilayer replicates the structure of the first layer adjacent to the substrate, exhibiting a template-like effect. Thus the W–Mo–Si bilayer topology replicates that of the single layer Mo film while that of Mo–W–Si and all the other multilayer films (with W as the first layer) are similar to that of the single layer W film with approximately the same grain size in all the films. The microstructure of the films on LGS and CTGS are similar to that on Si and hence have not been shown in this paper. The films have a preferred out-of-plane {110} orientation of the W and Mo grains. The residual stresses in the films on the Si substrate are given in Table 1. Multilayer stacking of the layers leads to a decrease in the compressive stress in the film as compared to the highly stressed single layer W film. A summary of the electrical resistivity of the thin films on Si, LGS and CTGS substrates are also given in Table 1. The resistivity of the bilayer and multilayer films is the lowest when deposited on

G.K. Rane et al. / Materials Science and Engineering B 202 (2015) 31–38

33

Fig. 1. In-lens SEM micrographs of the as-deposited films on Si/SiO2 substrate.

CTGS. On all the substrates, the resistivity of the multilayer films is lower than that of the single layer W or Mo films. The resistivity of the bilayer W50 –Mo50 –S is the highest among all the films. The higher resistivity of the bilayer stack W50 –Mo50 –S films as compared to that of the oppositely stacked Mo50 –W50 –S films has been attributed predominantly to the higher interfacial roughness of the prior in addition to the smaller grain size [29]. A higher interfacial roughness results from the impact of the bombardment of Ar gas neutrals (the sputtering gas used) on the underlying film. These neutrals reflected from the heavier W target have higher energy (almost 3 times higher) as compared to those reflected from the lighter Mo target resulting in a higher interfacial roughening in the prior film.

The electron mean free path (EMFP) of Mo is reported to be about 40 nm [30] while that of W [1,3] is reported as low as 4 nm. Hence, the layer thickness of Mo and W was reduced to observe its effect on the resistivity. Reduction of the individual layer thickness of Mo and W in the multilayer stack to 20 nm led to only a minor/negligible increase in the resistivity. Reduction of the W layer thickness to 10 nm while keeping the Mo layer thickness at 20 nm, led to a drastic decrease in the resistivity. For the film with 10 nm layers of both W and Mo, the resistivity increases again, however it is still slightly lower than that for the 20 nm W/Mo layer film. Thus, the Mo layer thickness was kept at 20 nm and that for W was reduced to 5 nm which again yielded the lowest resistivity amongst all the layer stack combinations. Although the EMFP of Mo is reported at

34

G.K. Rane et al. / Materials Science and Engineering B 202 (2015) 31–38

∼40 nm, no increase in the resistivity of the multilayers with even 10 nm of Mo is observed. The BCC metals Mo and W have a very low lattice mismatch of about 0.6% and are reported to grow coherent interfaces forming a kind of superlattice with columnar grains going through the layers [7,31]. The reduction in the resistivity regardless of the reduction in the individual layer thickness of Mo can be attributed to this effect. However, the higher the number of interfaces, the larger is the electrical resistivity. The number of interfaces in (Mo10 –W10 )5 –S stack is larger than for (Mo20 –W5 )4 –S and consequently the prior has the higher resistivity. Although W20 –(Mo20 –W20 )2 –S has lower number of interfaces, it does not follow the above trend and has a higher resistivity than the (Mo10 –W10 )5 –S film. The density (top layer value from XRR) of this film (Mo/W: 20/20) is lower than the latter. A reason for this could be due to the growth conditions. This film has the highest amount of W which means that a longer sputtering time was involved. Sputtering of W atoms is accompanied by back-scattering of highly energetic Ar neutrals from the W target (as was described already in [29]). These neutrals bombard the growing film surface for a longer time giving rise to larger defect content which leads to higher resistivity of this film. Thus, in the W–Mo multilayer system, the electrical resistivity shows a stronger dependence on the number of interfaces and the sputtering time for W than on individual layer thickness. Thus a 5 nm thick W layer was sufficient to provide a template-like effect while retaining the low resistivity due to its lower EMFP and the low sputtering time leads to lower interfacial damage, a higher coherency at the Mo–W interface, resulting in the low electrical resistivity value for the (Mo20 –W5 )4 –S sample.

4. Annealing studies Films on Si: The electrical resistivity evolution upon annealing at 400, 600 and 800 ◦ C each for 12 h is shown in Fig. 2a Annealing led to a decrease in the electrical resistivity of all the films with the (Mo20 –W5 )4 –Si film yielding the lowest resistivity of about 9 ␮ cm. A slight improvement in the crystallinity of the multilayer films is seen from the XRD measurements. Quantification of the coherently diffracting domain size and microstrain from the XRD data by the single line analysis method [32] for the single elemental film of W showed an almost double increase in the outof-plane domain size after annealing to 800 ◦ C (to ∼50 nm) with a strong decrease of the defect content (interpreted from the microstrain content) as compared to the as-deposited state. For Mo single films, the grain growth was far more appreciable with an almost 4 times increase (to ∼100 nm). Similar analysis for the multilayer films from the XRD data was not possible due to the close overlap of the W and Mo diffraction peaks. The in-plane grain growth can be better appreciated from the SEM micrographs of the annealed films on Si substrate shown in Fig. 3 as compared to the as-deposited films in Fig. 1. The out-of-plane {1 1 0} orientation of the crystallites improved slightly for the single layer W film while a considerable improvement was observed for the Mo film (see Fig. 4). The larger extent of texture improvement of the Mo film (along with larger grain growth) should be responsible for the higher drop in the resistivity upon annealing as compared to that of W. Further on, the SEM images (Fig. 3) show that the grain size of the bilayer W50 –Mo50 –S film is the lowest among all the annealed films which is responsible for it having the highest electrical resistivity. Among the 4 multilayer films, no appreciable grain growth can be seen except for the (Mo20 –W5 )4 –Si film which shows a large grain growth. This film also exhibits the lowest resistivity among all the films. In addition to these results, XRR measurements show an improvement in the reflectivity data with better delineation of

Fig. 2. Resistivity evolution as a function of annealing at 400 ◦ C, 600 ◦ C and 800 ◦ C for 12 h each (measurements made at room temperature after the heat treatment) for the films on Si/SiO2 , LGS and CTGS.

G.K. Rane et al. / Materials Science and Engineering B 202 (2015) 31–38

35

Fig. 3. In-lens SEM micrographs of the films on Si/SiO2 substrate after annealing to 800 ◦ C under vacuum.

the multilayer satellites of the annealed films as compared to the as-deposited films. Films on LGS: In contrast to the films on Si substrate, it was observed that the film resistivity started increasing upon annealing (see Fig. 2b). While for Mo–LGS the resistivity increased already at 400 ◦ C, in the other films the increase starts at or above 600 ◦ C. Resistivity of the W–LGS film remains almost unchanged upon annealing up to 800 ◦ C. In all the films after annealing to 800 ◦ C, the surface is completely damaged with the W film being the least affected. The SEM micrograph of an exemplary film (Mo50 –W50 –LGS) is shown in Fig. 5a. The surface of all the films are similarly damaged to more or less the same extent. The FIBcut cross section view (Fig. 5b) of this film shows that the Mo

layer above the W layer is destroyed with certain regions where the Mo layer is entirely absent. Correspondingly, the SEM image shows circular gaps all over the film. Sequential FIB-cut through such a gap showed that the defect propagates from the substrate reaching up to the film surface. The LGS substrate is damaged to a depth of almost 100 nm below the film at such defects (see Fig. 5b). All the samples show such a damage of the film-substrate system but to a lower extent when W is the top-most layer. The chemical composition of the Mo50 –W50 –LGS sample was studied using AES measurements at several locations on the sample. The film surface shows the presence of Ga (<1 at.%) and La (2–5 at.%) besides O and C. Depth profiling revealed the presence of oxygen throughout the film. It was seen that measurements at different locations required

36

G.K. Rane et al. / Materials Science and Engineering B 202 (2015) 31–38

Fig. 4. {1 1 0} pole figure section for as-deposited and 800 ◦ C annealed W and Mo films on Si/SiO2 substrate.

Fig. 6. FIB cut cross-section view (samples are tilted by 54◦ ) of the films deposited on CTGS after annealing to 800 ◦ C.

Fig. 5. (a) SEM (in-lens) micrograph and (b) FIB-cut cross section view of the Mo50 –LGS50 –LGS film after annealing up to 800 ◦ C. The positions marked 1 and 2 show the locations where the Mo layer is damaged or completely absent. The position 3 shows the region where the substrate defect reaches into the film.

different sputtering times to reach to the substrate. A clear demarcation of the start of the substrate was not possible from the depth profiles (data scattering is seen close to the substrate since the substrate is insulating and electric charging inhibits measurements and detection of the amounts of the elements in the substrate). These two observations result from both, the change in the film structure at different locations (e.g. marked region 1 and 2 in Fig. 5b where the uneven Mo layer is seen from the FIB-cut) and due to the substrate defects that reach into the film (e.g. region 3 in Fig. 5b). The defects in the substrate region have been previously reported to arise due to the diffusion of the Ga and O from the LGS substrate forming blister-like features on the surface when annealed under low oxygen partial pressures [22] and thus making the substrate unstable under such conditions. Since the films were annealed under vacuum, the oxidation of the film can be a result of diffusion of O from the substrate, accompanying the diffusion of Ga. Since bulk phase

diagram of Mo–Ga exhibits a small solubility of Ga in Mo already at 530 ◦ C [33], the Mo–LGS film can be expected to have the least stability. A solubility of 1 at.% Ga in Mo is suggested by 1000 ◦ C. In contrast, W has no solubility for Ga and evidently the higher stability of the W–LGS sample can be seen (the electrical resistivity was least affected and the structure of the film was intact). Previous studies on Ir electrodes showed a similar oxidation of the film on LGS when annealed under vacuum conditions by taking up oxygen from the substrate [21] (several Ga–Ir phases are possible). The destruction of the films on LGS can be attributed to this instability of the substrate. Films on CTGS: The resistivity evolution upon annealing the films deposited on CTGS is shown in Fig. 2c. A small reduction in the film resistivity is observed upon annealing to 600 ◦ C similar to that on the Si substrate. Further annealing to 800 ◦ C led to an almost 10% reduction in the resistivity for all the films. Similar to the films on Si, the highest resistivity is of the W50 –Mo50 –CTGS film while (Mo20 –W5 )4 –CTGS has the lowest value. The layer stacks in the multilayer films on CTGS, after the annealing process up to 800 ◦ C, can be seen in the FIB-cut cross section view shown in Fig. 6. It is noted that the individual layers in the multilayer stacks can still be clearly demarcated after the annealing process. XRR measurements were made on the films before and after annealing. It is interpreted from the XRR data that upon annealing the individual layers in the multilayer system do not intermix, based on the profile of the Bragg reflections as well as the clearer Kiessig oscillations, but rather become more distinctly defined. Both, the reflected intensity and the critical angle increase. These changes can be appreciated from Fig. 7 which shows the XRR data for the (Mo20 –W5 )4 –CTGS sample, as an example, before and after the annealing procedure.

G.K. Rane et al. / Materials Science and Engineering B 202 (2015) 31–38

37

can be used to reduce the residual stresses in the film. Annealing of the films on Si and CTGS lowers the electrical resistivity of the films which is majorly an outcome of the improved interfacial structure which becomes sharp and coherent along with the formation of columnar grains that span the entire film thickness. The failure of the films on LGS starts already by 600 ◦ C due to the destruction of the substrate and the film. Changes in the chemical composition of the substrate at the interface to the film occurs due to the diffusion of the Ga and O ions upon annealing under vacuum. In contrast, all the films studied on the relatively new substrate CTGS are stable up to 800 ◦ C. Annealing improves the film quality and lowers the resistivity of the W/Mo films. W–Mo multilayer films show promising results for high temperature application as IDT electrodes on the piezoelectric substrate CTGS.

Acknowledgements Fig. 7. XRR data for the (Mo20 –W5 )4 -CTGS films as-deposited (black) and after the annealing procedure up to 800 ◦ C (gray).

Table 2 AFM measured surface roughness (rms values) of the films on CTGS before and after annealing up to 800 ◦ C. Films on CTGS

W100 –CTGS Mo50 –W50 –CTGS W50 –Mo50 –CTGS W30 –Mo40 –W30 –CTGS W20 –Mo20 –W20 –Mo20 –W20 –CTGS (Mo10 –W10 )5 –CTGS (W10 –Mo20 )3 –W10 –CTGS (Mo20 –W5 )4 –CTGS

Financial support by German BMBF (InnoProfile-Transfer grant 03IPT610Y) is gratefully acknowledged. The authors would like to express their gratitude to E. Brachmann for help with the thin film depositions, T. Wiek for FIB preparation and M. Hoffmann for AES measurements.

Roughness (rms)/nm As-deposited

Annealed

2.1 1.2 1.3 1.6 1.9 1.9 1.4 1.2

1.7 1.3 0.9 1.8 1.8 1.7 1.2 1.3

Qualitatively, this can be interpreted as a lowering of the interface roughness as well as a de-mixing of the adjacent layers at the interface and a slight increase in the density of the film (top layer). According to the phase diagram, although W and Mo are mutually soluble over the entire composition range, phase separation at the Mo–W interface is preferred instead of intermixing. This has been shown to be true also by first principle calculations [8] wherein a positive interface energy was calculated signifying that the Mo–W interface mixing is energetically unfavorable. The FIB cuts show columnar grains spanning the entire film thickness suggesting the formation of coherent interfaces with the grains spanning the entire thickness. This has also been shown by Geyang et al. on modulated W/Mo multilayers [7]. This effect results from the low lattice mismatch and similar thermal expansion coefficient of the two metals. All the film surfaces are rather smooth (<2 nm roughness) with the roughness decreasing in some of the films upon annealing (see Table 2). Thus all the films are shown to be stable on the CTGS substrate with a high quality of the film up to a temperature of 800 ◦ C. 5. Conclusion W and Mo single, bilayer and multilayer films have been studied for the first time on Si and the piezoelectric substrates LGS and CTGS to reveal their high temperature characteristics up to 800 ◦ C under vacuum. The as-deposited multilayers exhibit lower electrical resistivity as compared to the single layer or bilayer films. This is due to the lower EMFP of W as well as the formation of coherent interfaces as a result of the low lattice mismatch between W and Mo. We show that multilayer stacking of the hard W and softer Mo

References [1] D. Choi, B. Wang, S. Chung, X. Liu, A. Darbal, A. Wise, N.T. Nuhfer, K. Barmak, A.P. Warren, K.R. Coffey, M.F. Toney, J. Vac. Sci. Technol. A 29 (2011) 051512. [2] J. Ligot, S. Benayoun, J.J. Hantzpergue, J. Vac. Sci. Technol. A 19 (2001) 798–804. [3] S.M. Rossnagel, I.C. Noyan, J.C. Cabral, J. Vac. Sci. Technol. B 20 (2002) 2047–2051. [4] D.P. Adams, M. Vill, J. Tao, J.C. Bilello, S.M. Yalisove, J. Appl. Phys. 74 (1993) 1015–1021. [5] S.I. Jun, P.D. Rack, T.E. McKnight, A.V. Melechko, M.L. Simpson, J. Appl. Phys. 97 (2005). [6] T. Namazu, N. Maruo, S. Inoue, J. Mater. Sci. 47 (2012) 2725–2730. [7] L. Geyang, X. Junhua, Z. Liuqiang, W. Liang, G. Mingyuan, J. Vac. Sci. Technol. B 19 (2001) 94–97. [8] H.R. Gong, K. Cho, Appl. Phys. Lett. 91 (2007) 092106. [9] J.W. Mrosk, C. Ettl, L. Berger, P. Dabala, H.J. Fecht, A. Dommann, G. Fischerauer, J. Hornsteiner, K. Riek, E. Riha, J. Auersperg, E. Kieselstein, E. Born, B. Michel, M. Werner, A. Mucha, IECON ‘98, in: Proceedings of the 24th Annual Conference of the IEEE Industrial Electronics Society, vol. 1–4, 1998, pp. 2386–2390. [10] O. Elmazria, T. Aubert, Smart Sens. Actuators Mems V 8066 (2011) 806602. [11] T. Aubert, O. Elmazria, B. Assouar, L. Bouvot, M. Hehn, S. Weber, M. Oudich, D. Geneve, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 58 (2011) 603–610. [12] D. Richter, S. Sakharov, E. Forsen, E. Mayer, L. Reindl, H. Fritze, Procedia Eng. 25 (2011) 168–171. [13] S. Moulzolf, D. Frankel, M. Pereira da Cunha, R. Lad, Microsyst. Technol. 20 (2014) 523–531. [14] R.M. White, F.W. Voltmer, Appl. Phys. Lett. 7 (1965) 314–316. [15] R. Fachberger, G. Bruckner, R. Hauser, L. Reindl, Proceedings of the 2006 IEEE International Frequency Control Symposium and Exposition, vol. 1–2, 2006, pp. 358–367. [16] M.P. da Cunha, T. Moonlight, R. Lad, D. Frankel, G. Bernhard, Proc. IEEE Sens. (2008) 752–755. [17] J. Hornsteiner, E. Born, E. Riha, Phys. Stat. Solidi A 163 (1997) R3–R4. [18] X. Ji, T. Han, W. Shi, G. Zhang, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 52 (2005) 2075–2080. [19] T. Aubert, O. Elmazria, J. Bardong, G. Bruckner, IEEE Int. Ultrason. Symp. (2011) 2065–2068. [20] J. Bardong, G. Bruckner, M. Kraft, R. Fachberger, IEEE Int. Ultrason. Symp. (2009) 1680–1683. [21] T. Aubert, J. Bardong, O. Elmazria, G. Bruckner, B. Assouar, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 59 (2012) 194–197. [22] H. Fritze, J. Electroceram. 26 (2011) 122–161. [23] Z. Wang, D. Yuan, X. Cheng, D. Xu, M. Lv, L. Pan, X. Duan, H. Sun, X. Shi, Y. Lv, X. Wei, Z. Sun, C. Luan, S. Guo, G. Zhang, X. Wang, J. Cryst. Growth 253 (2003) 378–382. [24] O.M. Kugaenko, S.S. Uvarova, S.A. Krylov, B.R. Senatulin, V.S. Petrakov, O.A. Buzanov, V.N. Egorov, S.A. Sakharov, Bull. Russ. Acad. Sci. Phys. 76 (2012) 1258–1263. [25] D. Puccio, D.C. Malocha, N. Saldanha, M.P. da Cunha, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 54 (2007) 1873–1881. [26] L.J. van der Pauw, Philips Tech. Rev. 20 (1959) 220–224. [27] G.G. Stoney, Proc. R. Soc. Lond. Ser. A 82 (1909) 172–175. [28] L.B. Freund, J.A. Floro, E. Chason, Appl. Phys. Lett. 74 (1999) 1987–1989.

38

G.K. Rane et al. / Materials Science and Engineering B 202 (2015) 31–38

[29] G.K. Rane, S. Menzel, T. Gemming, J. Eckert, Thin Solid Films 571 (Part A) (2014) 1–8. [30] E. Fawcett, D. Griffiths, J. Phys. Chem. Solids 23 (1962) 1631–1635. [31] E.B. Svedberg, J. Birch, I. Ivanov, E.P. Munger, J.-E. Sundgren, J. Vac. Sci. Technol. A 16 (1998) 633–638.

[32] T.H. de Keijser, J.I. Langford, E.J. Mittemeijer, A.B.P. Vogels, J. Appl. Crystallogr. 15 (1982) 308–314. [33] B. Predel, Ga-Mo (Gallium-Molybdenum), in: O. Madelung (Ed.), Ga-Gd–Hf-Zr, Springer, Berlin Heidelberg, 1996, pp. 1–2.