Journal of Magnetism and Magnetic Materials 121 (1993) 201-204 North-Holland

Anr '

Annealing effects on the structure and magnetic properties of Ni /Ti multilayers

C. Sella a, M. M~aza b, M. K a a b o u c h i a, S. E1 M o n k a d e a, M. Mi loche a and H. Lassri c

a Laboratoire de Physique des Mat~riaux, C.N.R.S., 92125 Meudon, France b Laboratoire L~on Brillouin, C.E,4.-C.N.R.S., Bat. 563, Centre d'Etudes Nucl~aires de Saclay, 91191 Gif-sur-Yvette, France c Laboratoire de Magngtisme et Mat~riaux Magn~tiques, C.N.R.S., 92125 Meudon, France

Interfaces of Ni-Ti multilayers deposited by dc triode sputtering under high purity conditions (base pressure 10 -7 Tort plus getter-sputtering) and their thermal evolution are studied. This was achieved by combining high energy electron diffraction, secondary ion mass spectrometry and magnetometry. The study shows the existence of a non-magnetic interracial amorphous layer due to interdiffusion over very short distances during deposition. A marked asymmetry between the two sides of a given layer is observed, due to interfacial contamination of Ti by C and O acting as a diffusion barrier. Annealing treatments yield an increase in amorphization and finally crystallization of the interfacial amorphous phase to a NiTi stable phase.

1. Introduction

Ni-Ti multilayers are particularly interesting for applications as high reflectivity neutron optics (mirrors, polarizers, etc.) which require sharp in- terfaces, a high stack regularity and a good ther- mal stability. As shown recently [1,2] the mag- netic properties of such metallic multilayers are related to the structural characteristics of the films and particularly to the existence of an inter- facial amorphous layer due to interdiffusion over very short distances during the deposition pro- cess. In order to evaluate the annealing effects on the thickness of this interfacial layer and its mag- netic state, extensive structural and magnetic studies were carried out.

2. Experiments

The multilayers were deposited by dc triode sputtering under computer-controlled conditions. In this syostem [3], the layer thickness is controlled to + 1 A and the reproducibility is better than 1%. Water-cooled float-glass and NaC1 substrates were used. The base pressure was 10 -7 Torr and

Correspondence to: Prof. C. Sella, Laboratoire de Physique des Mat6riaux, C.N.R.S. 92125, Meudon, France. Fax (33) 1-45075845, Tel. (33) 1-45075163.

the sputtering pressure 7 x 10 -4 Tort using a large flow of high purity argon (N56). The multi- layers are deposited after getter-sputtering of the Ti target for one hour, the impurities coming from the pumping system or from the outgassing of the walls and components of the chamber being down to insignificant levels (comparable to UHV conditions). The substrate to target dis- tance was kept at 17 cm in order to minimize the interaction of the plasma with the surface. The estimated surface substrate temperature during deposition was 50°C. The rates of deposition of

o

Ni and Ti were 22 and 17 A / n m , respectively. Both metal layers had the same thickness and were varied in the range 10 to 300 A. The total number of bilayers was adjusted so as to get a total Ni thickness of about 600 .&.

Structural studies were made by transmission electron microscopy (TEM), high energy electron diffraction (HEED), grazing angle X-ray reflec- tometry (GAXR) to check the periodicity and the thickness of the interracial layer. Secondary ion mass spectroscopy (SIMS) profiling was per- formed with a Cameca IMS3F using an O~ or Cs + primary ion beam. The saturation magneti- zation and coercive force were measured by using a vibrating sample magnetometer. Annealing treatments were carried out at different tempera- tures during 3 h under a vacuum of 10 -6 Torr.

0304-8853/93/$06.00 © 1993 - Elsevier Science Publishers B.V. All rights reserved

202 C. Sella et al. / Annealing effects on N i / T i multilayers

3. Results and discussion

3.1. Structure and magnetic properties of as de- posited multilayers

First let us discuss the structure of single lay- ers. The structure of Ni was fcc with a grain size of the order of the layer thickness. The structure of Ti deopends strongly on its thickness. For t(Ti) < 100 A, one has an fcc structure, however for t(Ti) > 150 .A, both fcc and hcp are found to coexist. SIMS profiles indicate that the abnormal fcc structure in very thin Ti films seems to be stabilized by impurities (H, O, C). Similar abnor- mal fcc structures have been reported [4] for other transition metals such as Mo, W, Nb, Hf, Zr, etc.

In the case of multilayers, for metal layers o

thicker than 40 A, both layers are in fcc crys- talline form, with a (111) fiber structure. But for thinner layers, the diffraction rings of Ti disap- pear progressively and the width of Ni rings in- creases, resulting in an NiTi solid solution with an amorphous like structure, characterized by three halos whose diameters are very close to (111), (220) and (311-222) rings of Ni.

SIMS depth profiles performed using an O~- or Cs + primary ion beam and detecting the posi- tive secondary ions M + or cationized ions such as CsM + are particularly interesting for large modu- lated multilayers. As shown in fig. 1, a strong enhancement of Ti signal is observed in the inter- diffusion layer, while with O~ primary ions, an enhancement of Ni signal is observed in this same interfacial layer. It may be remarked that the origin of this enhancement is not completely un- derstood up to now. This is due to the difficulty of decoupling the sputtering process from the ionization process. Nevertheless this enhance- ment effect seems to be directly related to the presence of an amorphous interdiffusion layer. A similar effect has been observed recently by us in F e - T i multilayers. Practically, this effect can be used to characterize the interdiffusion layer at the successive interfaces.

As shown in fig. 1B, C, this enhancement effect exhibits a marked asymmetry between the two sides of a given layer. This is to be related to

10 6 2 J l 2 1 '

I l z C I A

10 ~'

10 z

10

z m !1

i w 4 .

i i le ,

I m l

i m m

u

I i e .

taw I I .

t o 4 .

Nt

1 2

I 10 20 30 ~0

\

C,O C,O , , ,

I m n

t o ! .

I D ! .

i i I

e

I n t e r f a c e i I n t e r f a c e 2 1

t C,O C,O

i u u L i , , ,

w l w I n i n I l l S O S l l 4 m "¢W

Erosion time [ran]

Fig. 1. SIMS depth profiles of a (Ni 200 ~k-Ti 200 .~) 4 multilayer using a Cs + primary ion beam (A and B) and an

O~- primary ion beam with 0 2 backfill (C).

the deposition process: interface labelled (1) is contaminated by reactive residual gases (H, O, C) t rapped on the surface of the Ti target and re- leased at the beginning of Ti sputtering. The Ni target is less reactive and no contamination is observed at interface (2). One can notice that even if deposition is carried out under high purity conditions, the extreme sensitivity of SIMS allows

C. Sella et al. / A n n e a l i n g effects on N i / T i multilayers 203

~5

3

2 /=SK {'/ I I

50 100 150

t(Ni) l~l

Fig. 2. Variation of the product Ms, t(Ni) versus t(Ni) at 295 and 5 K.

trace analysis at the interfaces down to ppb de- tection limits for light elements (H, O, C). This interfacial contamination layer acts as a diffusion barrier. Thus, interdiffusion at interface (2) is greater than at interface (1). Such interracial asymmetry was recently deduced from grazing angle X-ray reflectometry measurements [6]. For example in the case of an as-deposited (Ni 60 A-Ti 60 A) 10 multilayer, GAXR simulation indicated interface layer thicknesses of 9 and 15

for interfaces of type (1) and (2), respectively. The magnetization decreases strongly with a

decrease in Ni layer thickness. This could be explained in terms of a magnetically dead layer of Ni at each interface due to alloying effects. The thickness of such a dead layer can be estimated as shown in fig. 2, where we have plotted the product M-t(Ni) as a function of t(Ni) at both 295 and 5 K. The slope which corresponds to the

magnetization yields 489 and 500 emu cm -3 at 295 and 5 K, respectively. The extrapolated Ni dead layer thickness is 14 and 12 A at 295 and 5 K, respectively. The in-plane M-H loops are rectangular and no significant anisotropy was found in the film plane.

3.2. Annealing effects

Annealing treatments were carried out on samples with individual layer thicknesses of 60 ,~ (10 bilayers t(Ni) = t(Ti) = 60 ,~). The magnetiza- tion M s decreases slowly as the annealing tem- perature is increased beyond 150°C. Above 150°C, M s decreases rapidly and finally the samples be- come feebly magnetic (at 295 K) when annealed

3 6 0

2 7 0

1 8 0

90

o S

[.e*~]: Hc [oooo 1: Ms

28

i i

1 5 0 2 0 0

2 2

16

0 I ~ 1 0

0 5 0 1 0 0 2 5 0

r t~Cl

Fig. 3. Dependence of magnetization M s and coercivity H c on annealing temperature.

Fig. 4. Electron diffraction patterns of (Ni 60 ,~-Ti 60 A) 10 multilayers: (A) as deposited, (B) annealed for 3 h at 300°C,

(C) annealed for 3 h at 400°C.

204 c. Sella et al. /Annea l ing effects on N i / T i multilayers

above 180°C (fig. 3). Coercivity H c increases for annealing at 150°C, due to bet ter texturing, and decreases for higher annealing temperatures.

Structural studies (fig. 4) show that for anneal- ing temperatures beyond 100°C, no change oc- curs. However for annealing temperatures in the range 150-350°C, amorphization increases. Ti rings become more and more diffuse and disap- pear completely. Ni diffraction lines are still in- tense but the lattice constant increases indicating some dissolution of Ti in Ni (fig. 4B). The sample becomes feebly magnetic. Above 350°C, we ob- serve new weak diffraction lines due to the for- mation of a new crystalline compound which is identified as the NiTi phase reported in the liter- ature [5]. At 400°C (fig. 4C), the crystallization of the amorphous phase to a stable NiTi phase is complete. The remaining fcc Ni with an increased lattice constant is feebly magnetic.

Diffusion kinetics can be deduced from the evaluation of the magnetically dead layer thick- ness as a function of the annealing temperature. In the range 50-140°C, a diffusion coefficient of the order of 4 × 10 -18 cm2/s is obtained from the data of fig. 3 with an activation energy of 0.4 eV. Above 150°C, the diffusion coefficient in- creases due to more complex phenomena (release of hydrogen, recrystallization of Ni, increase in

amorphization and finally crystallization of the amorphous phase to a NiTi stable phase). These phenomena may influence the diffusion kinetics.

The interdiffusion layer thickness also is a function of the surface tempera ture of the film during deposition and of the duration of the condensation. Therefore it may be affected by the sputtering technique. A comparative study of the magnetization of as-deposited N i - T i multilayers prepared by rf magnetron, dc triode and rf diode sputtering with film surface temperatures esti- mated at 30, 50 and 100°C, respectively, also confirmed this influence.

References

[1] C. Sella, M. Kaabouchi, M. Miloche, M. M~aza and R. Krishnan, 1st Intern. Symp. on Atomically Controlled Sur- faces and Interfaces, Tokyo, 1991, Appl. Surf. Sci. 60-61 (1992) 781.

[2] M. Porte, H. Lassri, R. Krishnan, M. Kaabouchi, M. M~aza and C. Sella, 6th Intern. Conf. on Solid Films and Surfaces, Paris, 1992, to appear in Appl. Surf. Sci.

[3] C. Sella, K.B. Youn, R. Barchewitz and M. Arbaoui, Vacuum 36 (1986) 121.

[4] K.L. Chopra, M.R. Randlett and R.H. Duff., Philos. Mag. 16 (1967) 216.

[5] F.E. Wang, W. Buehler and S. Pickart, J. Appl. Phys. 36 (1965) 3232.

[6] M. M~aza, M. Kaabouchi, B. Pardo, C. Sella, H. Lassri and F. Bridou, submitted to J. Magn. Magn. Mater. (1992).