Mo multilayers

Mo multilayers

Thin Solid Films 373 Ž2000. 287᎐292 Structure determination of Ž Ti,Al. NrMo multilayers C.J. Tavares a,U , L. Reboutaa , E.J. Alves b a Departament...

1MB Sizes 0 Downloads 19 Views

Thin Solid Films 373 Ž2000. 287᎐292

Structure determination of Ž Ti,Al. NrMo multilayers C.J. Tavares a,U , L. Reboutaa , E.J. Alves b a

Departamento de Fısica, Uni¨ ersidade do Minho, Azurem, ´ ´ 4800-058 Guimaraes, ˜ Portugal b ITN, Departamento Fısica, E.N. 10, 2685 Saca¨´ em, Portugal ´

Abstract ŽTi,Al.NrMo multilayers have been deposited by dc magnetron sputtering on high-speed steel and silicon substrates. Experimental X-ray diffraction ŽXRD. and computational modelling of these patterns has been carried out to achieve the basics in elucidating their structural properties. The layers were designed with modulation periods of approximately 13 nm, up to a total thickness of 2.8 ␮m. Residual stress experiments revealed a compressive stress state that prevailed in these structures, ranging from y0.2 to y1.3 GPa. This in turn is in good agreement with the XRD-refined expanded values of the out-of-plane interplanar distances. RBS spectra provided the film composition and a qualitative evaluation of the waviness of the interfaces with increasing substrate bias potential. 䊚 2000 Elsevier Science S.A. All rights reserved. Keywords: Multilayers; Roughness; XRD; RBS

1. Introduction Multilayers are one-dimensional artificial structures comprising a number of alternate layers. They fit in a vast range of applications. Wear prevention in steel tools and cutting applications w1,2x was our major concern while attempting to produce ŽTi,Al.NrMo coatings. Ti 0.4 Al 0.6 N Žfcc. and Mo Žbcc. both possess a cubic crystal structure and share similar properties, such as: high melting point; good chemical and thermal stability; and comparable elastic properties. Since the most interesting and notable properties of multilayers derive from the close proximity of different materials, it is not surprising that these properties are often strongly sensitive to the nature of the interfaces between them. Therefore, in order to understand and manipulate the physical behaviour of multilayers, it is crucial to determine the detailed structure of the layers and inter-

U

Corresponding author. E-mail address: [email protected] ŽC.J. Tavares..

faces. Moreover, it is essential to correlate the analysed structure with the measured properties w3x. 2. Experimental details 2.1. Fabrication of the samples ŽTi,Al.NrMo coatings were deposited using a custom-made sputtering system. An ArrN2 atmosphere was present in the chamber, with an argon flow rate Žpressure. of 140 cm3rmin Ž0.35 Pa. and a nitrogen flow rate Žpressure. of 8.4 cm3rmin Ž0.13 Pa. for growing TiAlN; while to produce Mo, the Ar Žpressure. flow was varied from 160 to 250 cm3rmin Ž0.5᎐0.7 Pa.. Pure 200 =100= 6 mm TiAl and Mo targets were used. A current density of approximately 0.01 Arcm2 was applied to both magnetrons. The substrate bias voltage was changed from 0 to y120V while the targetto-substrate distance was kept at 110 mm in all depositions. The base pressure was typically of the order of 5 = 10y5 Pa, while the substrate temperature during deposition was 200⬚C. Before deposition the substrates

0040-6090r00r$ - see front matter 䊚 2000 Elsevier Science S.A. All rights reserved. PII: S 0 0 4 0 - 6 0 9 0 Ž 0 0 . 0 1 1 0 9 - 3

C.J. Ta¨ ares et al. r Thin Solid Films 373 (2000) 287᎐292

288

were in situ sputter-etched in an argon atmosphere of 7 Pa with a dc power of 100 W for 20 min. Two different series of ŽTi,Al.NrMo multilayers were produced with the intention of maintaining the modulation periodicity constant, and varying the bias voltage and the argon partial pressure during Mo deposition in the first ŽDB series.. In the second series ŽDC series. a set of thinner multilayers were prepared for RBS measurements. The experimental details are displayed in Table 1. 2.2. Rutherford backscattering spectrometry (RBS) Rutherford backscattering spectrometry ŽRBS. was used to determine the film composition. A 2 MeV Heq beam in a 3.0 MV Van de Graaff accelerator w4x was used to analyse samples deposited on Si wafers. The backscattered particles were detected by a surface barrier detector placed at 160⬚ to the beam direction in the Cornell geometry and with an energy resolution FWHM of 14 keV and a beam spot of 0.2= 0.6 mm. The RBS spectra were fitted with the RUMP code w5x.

layers requires the modelling of their superlattice structure. This modelling includes lattice strains, and intra- and inter-layer disorder. Thus, by adjusting the structural parameters of the model to best fit the measured intensities, we can determine the structural parameters of the layers w6᎐8x. The SUPREX refinement program was used to fit the experimental highangle XRD patterns of the ŽTi,Al.rMo multilayers w6,9x. The kinematical step-model w10,11x approach enables the calculation of the average number of atomic planes Ž N . in an individual layer ŽTiAlN or Mo. and its associated discrete Gaussian width Ž ␴N . and also the atomic interplanar distances in each layer Ž d TiAlN and d Mo .. In addition, the interfacial distance Ž d int ., whose height varies continuously due to lattice mismatch between the two materials, and its characteristic continuous atomic level of disorder Ž ␴int . can also be determined. A value for the modulation period Ž ⌳ . may be calculated through the formula: ⌳ s Ž NTiAlN y 1 . d TiAlN q Ž NMo y 1 . d Mo q 2 d int

Ž1.

2.3. High-angle XRD and refinement model

2.4. Low-angle XRD

For the XRD scans a Philips PW3040r00 X’Pert diffractometer was used with the standard Bragg᎐ Brentano geometry. The specular resolution is 0.002⬚ and the integration time was varied from 1.25 to 5 s with a 2␪ step of 0.01⬚, both for low-angle and high-angle diffraction experiments. Knowledge acquired for the crystal texture was derived from XRD at high-angles. The multilayer peak positions of the high-angle diffraction spectra are dependent solely on the average lattice spacing of the constituent layers and the modulation periodicity. Structural information regarding the ŽTi,Al.N and Mo

In the low-angle regime, the length scales that are probed are greater than the lattice spacing of the constituent layers. Therefore, the scattering solely arises from the chemical modulation of the structure. In low-angle spectra, the modulation period can be assessed through the position of the Bragg diffraction peaks w12x: ns

2⌳ ␭

'cos Ž ␪ . y cos Ž ␪ . 2

2

c

Ž2.

n

and compared to that determined by high-angle formal-

Table 1 Experimental details and calculated results regarding the deposition of different series of samples Sample

Type

PTiAlN ŽAr.a ŽPa.

PMo ŽAr.b ŽPa.

tTiAlN rtMoc

⌳ Žnm.

Biasd ŽV.

Stress ŽGPa.

␴Totale Žnm.

DB1 DB2 DB3 DB4 DB5 DB6 DC1 DC2 DC3

ŽTi0.4 Al0.6 NrMo. = 200 ŽTi0.4 Al0.6 NrMo. = 200 ŽTi0.4 Al0.6 NrMo. = 200 ŽTi0.4 Al0.6 NrMo. = 200 ŽTi0.4 Al0.6 NrMo. = 200 ŽTi0.4 Al0.6 NrMo. = 200 ŽTi0.4 Al0.6 NrMo. = 50 TiAlrŽTi0.4 Al0.6 NrMo. = 50 MorŽTi0.4 Al0.6 NrMo. = 50

0.35 0.35 0.35 0.35 0.35 0.35 0.35 0.3 0.35

0.6 0.6 0.6 0.7 0.7 0.7 0.5 0.5 0.7

1.1 1.0 1.1 1.1 1.0 1.1 1.1 1.1 1.1

13.6 13.3 13.6 13.5 12.8 13.6 13.6 14.1 12.3

y60 y80 y120 y60 y80 y120 0 y60 y120

y0.19 y0.24 y1.33 y0.31 y0.48 y1.13 ᎐ ᎐ ᎐

0.8 0.6 0.5 0.8 0.7 0.6 ᎐ ᎐ ᎐

a

The values of P TiAlN ŽN2 . were kept constant in all depositions at 0.13 Pa. P Mo ŽAr. is the partial pressure of argon relative to the deposition of Mo. c t TiAlN rt Mo refers to the thickness ratio of the constituent materials in a bilayer of Ti 0.4 Al 0.6 N and Mo, and together with ⌳ was obtained from studying the refinements of both the low- and high-angle XRD patterns. d Bias is the polarisation potential applied to the substrate holder. e The experimental compressive residual stress values and simulated interfacial roughness Ž ␴Total . are also shown for the thicker samples. b

C.J. Ta¨ ares et al. r Thin Solid Films 373 (2000) 287᎐292

289

ism. n represents the order of diffraction, related to the Bragg peak positioned at ␪ n ; ⌳ is the modulation period of the multilayer; ␪ c f 0.4⬚ is the critical angle, below which all radiation is totally reflected; and ␭ corresponds to the CoK ␣ radiation wavelength. The low-angle results are very sensitive to changes in the electronic density throughout the sample and less sensitive to the intra-layer disorder. Since the detected radiation results from the contribution of the reflected radiation on the interfaces, the low-angle patterns are, above all, sensitive to the interface quality and chemical modulation. From the low-angle modelling we can estimate an average rms roughness present at the interfaces Ž ␴ Total ., however limited by the coherence length of the probing radiation w9x: ␴Total s

)

2 2 ␴DW q ␴c2 q t i-d q Ž ␴TiAlN = d TiAlN .

q Ž ␴Mo = d Mo .

2

2

Ž3.

In this last equation the modelled parameters from the SUPREX program include: the width of the interdiffusion zone at the interface Ž t iy d .; a Debye᎐Waller Ž ␴ DW . coefficient that accounts for the waviness of the interfaces; and the width of the continuous distribution of the layer thickness Ž ␴c .. Surface roughness will be determined by kinetic limitations, a competition between various rates w13x. Moreover, the coherence length of the X-rays at low-angles is higher than at high-angles, therefore the region that is probed contains several grains where the wave profile of the interfaces is better manifested. Thus, the calculated roughness from the modelling of the low-angle XRD spectra should be higher than that from the high-angle modelling. 3. Results and discussion These samples exhibit a columnar structure that was clearly observable through SEM analysis. Fig. 1 illusFig. 2. Morphology of three Ti 0.4 Al 0.6 NrMo samples wfrom top to bottom: Ža. DB1, Žb. DB2 and Žc. DB3x from the same series varying only the bias potential applied to the substrates. As the bias is increased from y60 to y120 V Žfrom top to bottom. the surface becomes cleaner with fewer defects in its surface. The magnification shown in these pictures is 50 = .

Fig. 1. Cross-section SEM analysis of a Ti 0.4 Al 0.6 NrMo multilayer showcasing its typical columnar structure.

trates this characteristic morphology, for the particular case of a 2.4-␮m-thick sample. One apparent effect of the negative potential that was applied to the substrates during the deposition of these multilayers is related to the reduction in the number and size of defects that arise at the surface. The possible mechanism responsible for this has to do with the disorientation of columns during growth. These protrusions often arise from grains exposing a higher surface energy to the deposition flux. In Fig. 2 it is obvious how the film

C.J. Ta¨ ares et al. r Thin Solid Films 373 (2000) 287᎐292

290

Table 2 Composition of the DC series of Ti 0.4 Al 0.6 NrMo samples obtained by refining the RBS spectra Sample

Total thickness Žatrcm2 .

Mo Žat.%.

Ti Žat.%.

Al Žat.%.

N Žat.%.

Mo thickness Žatrcm2 .

Ti0.4 Al0.6 N thickness Žatrcm2 .

DC1 DC2 DC3

4.9= 1018 4.4= 1018 4.2= 1018

37 42 36

13 11.5 13

19 17.5 19

31 29 32

3.6= 1016 3.7= 1016 3.2= 1016

6.1= 1016 5.1= 1016 5.2= 1016

surface became cleaner with protrusions as the bias voltage was increased from 0 to y120 V. From AFM images these protrusions can be at most 350 nm wide and 300 nm tall. An RBS experiment, with respective fit, was performed on a thick and homogeneous ŽTi,Al.N coating in order to determine the composition of the individual layers. From this result it was assumed that those layers consist of 30 at.% aluminium, 20 at.% titanium and approximately 50 at.% nitrogen, hence we have Ti 0.4 Al 0.6 N. According to Y. Setsuhara et al. w14x, and from the XRD analysis of a thick sample, this stoichiometry is evidence of a NaCl type structure. RBS spectra only allow analysis of the outermost layers of the multilayers. The evolution of the RBS spectra, and their respective fits with RUMP, in respect to the different deposition parameters Žtaken at an incidence angle of 10⬚. is illustrated in Fig. 3 for samples DC1, DC2 and DC3, which have a thickness of approximately 0.7 ␮m and a modulation period of approximately 13 nm. This structural data was useful in the determination of the film’s composition and individual layer thickness Žin atrcm 2 .. Sample DC1, prepared without substrate bias, appears to have rougher or wavier interfaces since the individual layers cannot be separated using the RBS technique. Samples DC2 and DC3 were prepared with a substrate bias of y60 V and y120 V, respectively, having in the former a buffer layer of TiAl with a thickness of ; 100 nm and in the latter a buffer layer of Mo with a thickness of ; 39 nm. An increase in bias voltage and the consequent increase in the Arq ion energy have the effect of decreasing the multilayer roughness Žsee Fig. 3.. Ion bombardment in these samples was relatively high with an ion-to-atom flux ratio Ž JirJa . at the substrate in the order of 2 for Mo and 12 for Ti 0.4 Al 0.6 N. These values were calculated from the saturation current density Žflux of incoming atoms. in the substrate holder w15x and from the deposition rate Žflux of atoms. estimated according to the corresponding RBS-calculated thickness in atrcm 2 Žsee Table 2.. The decrease in the multilayer roughness is associated with an increase in the residual stress of the multilayer structure. Regarding the individual layer thickness, a reasonable fluctuation is evidenced. This can be explained by the fact that from DC1 to DC2 the bias voltage was increased from

0 to y60 V, hence the heavier ions flux rate and Arq ion energy was enhanced. In addition, the argon partial pressure for Ti 0.4 Al 0.6 N dropped from 0.35 to 0.3 Pa, which led to a decrease in the deposition rate for this material. When comparing samples DC1 and DC3 we can conclude that the decrease in the thickness was due to re-sputtering effects derived from such a high voltage present in the substrate holder. Both low-angle and high-angle scans provided us with the modulation periodicity. The Bragg peaks associated with the low-angle patterns are identified up to the eighth order. This allowed the precise determination of the modulation periodicity Ž ⌳, see Table 1., and also verified a good chemical modulation in the multilayers. In the high-angle XRD scan of Fig. 4 for sample DC2 no central peak is evident but two main peaks instead, indicating a multilayer preferential growth direction. This is due to the large difference in the scattering factors between the multilayer constituents. Since only the first-order satellite peaks are visible in

Fig. 3. RBS spectra and the respective RUMP refinement corresponding to the DC series of Ti 0.4 Al 0.6 NrMo samples showing a decrease in the interfacial roughness with the value of the negative bias potential applied to the substrates during deposition.

C.J. Ta¨ ares et al. r Thin Solid Films 373 (2000) 287᎐292

Fig. 4. XRD high-angle spectrum from sample DC2 consisting of a 50-bilayer Ti 0.4 Al 0.6 NrMo structure with a period of 14.1 nm. Both Ti 0.4 Al 0.6 N Ž111. and Mo Ž110. growth directions are viewed.

these patterns in approximately the Ti 0.4 Al 0.6 N Ž111. and Mo Ž110. direction, it is indicative that a possible mechanism that lowers the intensity of these peaks has to be directly related to the rough wavy interfaces. In addition, other growth directions were visible, namely the Ti 0.4 Al 0.6 N Ž200. and Mo Ž200., albeit associated with lower intensities and not completely textured. This last fact implies a grain-related interfacial roughening that enhances the overall disorder and contributes to the absence of higher-order satellite peaks. While analysing the output disorder parameters from the SUPREX XRD refinement program, both discrete and continuous levels of fluctuation are considerable, enhancing the total disorder, which lead to roughness rms maximum values of 0.8 nm for bias voltages of y60 V and rms minimum values of 0.5 nm for bias

291

voltages of y60 V Žsee Table 1.. This fact is in agreement with the decrease of the interfacial roughness with increasing bias voltage. In addition, an improvement in the resolution of the Bragg peaks in the low-angle XRD scans was experimentally noticeable with increasing polarisation potential, leading to sharper interfaces. From the refinement of the low-angle XRD patterns, inter-diffusion distances of the order of 0.4 nm were obtained for all samples. Therefore, a mixing of both species ŽTi 0.4 Al 0.6 N and Mo. occurs at the interface. In addition, the main peaks in the highangle scans are very broad, indicating in turn a considerable amount of layering defects. The values of ⌳ and t TiAlNrt Mo Žindividual layer thickness ratio. shown in Table 1 were obtained from separately refining the low- and high-angle scans with SUPREX. A good indicator of the reproducibility of these structures is that the modulation period was confirmed as almost constant at approximately 13 nm throughout the DB series, fluctuating slightly in the thinner set ŽDC series.. Fig. 5 shows both refinements for the low- and high-angle patterns associated with sample DB1. In the case of the modulation period, the refined values ŽEq. Ž1.. coincide with the predicted ones using Bragg’s equation with a correction for the critical angle ŽEq. Ž2... Residual in-plane stress was calculated using Stoney’s equation w16x after measuring the deflection of the substrates prior to and after deposition w17x. All samples were under a compressive residual stress, these values ranging typically from approximately y0.2 to y0.5 GPa. A couple of higher values of the order of y1 GPa, for particular experimental conditions, are related to a high Žy120 V. bias voltage during deposition. From the results in Table 1 we can conclude that

Fig. 5. XRD low- and high-angle spectra and respective SUPREX refinement for sample DB1 consisting of a 200-bilayer Ti 0.4 Al 0.6 NrMo structure with a period of 13.6 nm.

292

C.J. Ta¨ ares et al. r Thin Solid Films 373 (2000) 287᎐292

both an increase of the bias voltage and, to some extent, the argon partial pressure, to produce Mo increases the residual stress values. This compressive in-plane stress field is compatible with the out-of-plane tensile Žrelaxed. stress field detected in the XRD profiles. Both refined perpendicular interplanar distances ˚ and 2.242 A, ˚ Ž d . for Ti 0.4 Al 0.6 N and Mo Ž2.392 A . respectively are expanded relative to the bulk ones ˚ and 2.219 A, ˚ respectively.. Hence, the effect Ž2.387 A of the bias voltage on these coatings is to increase the state of residual stress as the voltage is elevated from 0 to y120 V. The lattice mismatch between Ti 0.4 Al 0.6 N and Mo is relatively high Ž; 7%. and therefore the possibility of coherency strain effects should not be ruled out. 4. Conclusion We report in this paper the production of Ti 0.4 Al 0.6 NrMo multilayers. In order to study their structural properties, RBS, XRD low- and high-angle experiments were conducted. Furthermore, to study the structural parameters, computational modelling of the experimental spectra was carried out. An estimation of the interfacial roughness was obtained, however this was limited by the coherency of the X-rays. A higher cumulative roughness, which was not estimated, exists in these coatings, since the interfacial disorder increases with film thickness, as proven by RBS experiments. The latter spectroscopy was used to determine the film composition.

References w1x K.J. Ma, A. Bloyce, T. Bell, Surf. Coat. Technol. 76r77 Ž1995. 297. w2x L.A. Donahue, J. Cawley, D.B. Lewis, J.S. Brooks, W.-D. Munz, ¨ Surf. Coat. Technol. 76r77 Ž1995. 149. w3x D.G. Stearns, J. Appl. Phys. 71 Ž1992. 4286. w4x M.F. da Silva, M.R. da Silva, E.J. Alves et al., in: R. Kossowsky, S. Singhal ŽEds.., Surface Engineering, NATO ASI, Les Arcs, Nijhoff, 1984, p. 74. w5x L.R. Doolittle, Nucl. Instrum. Meth. B9 Ž1985. 344. w6x E.E. Fullerton, I.K. Schuller, H. Vanderstraeten, Y. Bruynseraede, Phys. Rev. B 45 Ž1992. 9292. w7x K. Temst, M.J Van Bael, B. Wuyts et al., Appl. Phys. Lett. 67 Ž23. Ž1995. 3429. w8x C.J. Tavares, L. Rebouta, B. Almeida, J. Bessa e Sousa, Surf. Coat. Technol. 100r101 Ž1998. 65. w9x H. Vanderstraeten, Ph.D. Thesis, Katholieke Universiteit Leuven, Belgium, 1991. w10x A. Segmuller, A.E. Blakeslee, J. Appl. Cryst. 6 Ž1973. 19. ¨ w11x E. Spiller, in: P. Dhez, C. Weisbuch ŽEds.., Physics, Fabrication and Applications of Multilayered Structures, Plenum, New York, 1987, p. 271. w12x Y.S. Gu, W.P. Chai, Z.H. Mai, J.G. Zhao, Phys. Rev. B 50 Ž1994. 6119. w13x D.E. Savage, N. Schimke, Y.-H. Phang, M.G. Lagally, J. Appl. Phys. 71 Ž7. Ž1992. 3283. w14x Y. Setsuhara, T. Suzuki, Y. Makino, S. Miyaki, T. Sakata, H. Mori, Surf. Coat. Technol. 97 Ž1997. 254. w15x P.J. Kelly, R.D. Arnell, J. Vac. Sci. Technol. A 16 Ž1998. 2858. w16x G.G. Stoney, Proc. R. Soc. London Ser. A 82 Ž1909. 172. w17x C.J. Tavares, L. Rebouta, M. Andritschky, S. Ramos, J. Mater. Process. Technol. 92r93 Ž1999. 177.