Materials Science and Engineering, A 168 (1993) 257-261 257
Influence of the deposition conditions on the optical, electrical and magnetic properties of sputtered Fe-A1203 and Ni-A1203 nanocermet films
M . G a d e n n e a n d P. G a d e n n e
Laboratoire d'Optique des Solides, Universit~ P. et M. Curie, 4 Place Jussieu, 75252 Paris Cedex 05 (France)
M . T. R a m d o u a n d J. P. S e g a u d
Direction des Matdriaux, ONERA, 91120 Palaiseau (France)
H . L a s s r i a n d R. K r i s h n a n
Laboratoire du Magndtisme et des Mat~riaux Magn(tiques, CNRS, 1 place A. Briand, 91195 Meudon Cedex (France)
C. Sella
Laboratoire de Physique des MatOriaux, CNRS, 1 place A. Briand, 92195 Meudon Cedex (France)
Abstract
We reveal the influence of the deposition conditions on the physical properties of cermet films made of nanometer sized magnetic metal (# ~/~0) and alumina grains. The films were prepared by r.f. cosputtering under two typical base pressures (10 -6 and 10 -7 Torr), using either an argon plasma or an argon+ 10% hydrogen plasma under 6 x 10 -3 Torr. They present a wide range of compositions and thicknesses. Electron microprobe analysis, transmission electron microscopy and high energy electron diffraction were used to study their crystallographic and granular structure. We report the electrical d.c. conductivity, optical transmittivity with respect to wavelength, magnetization and ferromagnetic resonance as a function of the deposition conditions. These physical properties are governed to a large extent by oxidation of the metal.
1. Introduction
It is now well known that, for a given cermet type, the percolation threshold mainly depends on the con- centration of metal [1]. We have shown that the threshold occurs for a critical thickness of film for a high enough metal concentration [2, 3]. Moreover, when comparing the physical properties in the litera- ture, one notices that they can be very different depending on the elaboration technique. For example, the d.c. conductivity and the optical properties vary over a wide range, even if the thickness of the films and the apparent concentration of metal are fixed. Studying a mixture of gold and chromium, it has been shown [4]
that the deposition conditions play a prominent part in the oxidation of this compound. The magnetic proper- ties of cermet made of magnetic metal inclusions in alumina may also be modified by partial oxidation of the magnetic metal [5]. Taking into account these pre- liminary remarks, we investigate here in a systematic
way the influence of the deposition conditions (base pressure, composition of the sputtering gas, etc.).
2. Preparation of the samples
Our samples were thin cermet films ( F e - A I 2 0 3 and N i - A I 2 0 3 ) , deposited by r.f. cosputtering onto several kinds of substrate. The target was an alumina disc (110 mm diameter) on which small nickel or iron plates (5 mm diameter) were regularly placed in hexagonal lattice. The number of small discs varied between 55 and 223, to prepare samples with very different metal concentrations. A rotating substrate holder, allowing in situ measurement of the d.c. resistance, provided samples with good homogeneity. Experiments were performed under various base pressures (from a few 10 -6 Torr to 10 -7 Torr) using either an argon plasma or an argon + 10% hydrogen plasma, under a working pressure of 6 x 10 - 3 Torr. The thicknesses of the films
0921-5093/93/$6.00 © 1993 - Elsevier Sequoia. All rights reserved
258 M. Gadenne et al. / Fe-Al203 and Ni-Al203 cermet films varied from a few tenths of nanometers to several
micrometers for all concentrations.
3. Deposition rate
The global deposition rate is mainly dependent on the target composition, as iron, nickel and alumina do not have the same sputtering rate. In Table 1 we com- pare the deposition rate measured for two target con- figurations under vacuum and plasma conditions. The total sputtering time was the same (100 min) for all measurements. Obviously the global rate is higher when the base pressure is lower. The addition of hydrogen slows down this rate, probably by modifying the quantity and nature of iron oxides.
5. Atomic composition of the samples
The composition was determined by electron micro- probe analysis (Table 2). The results show the number of atoms of metal (nickel or iron), aluminum and oxygen. The argon included cannot be detected. On the one hand, the number of oxygen atoms is always greater than the alumina stoichiometry suggests. This means that there is significant reorganization of atoms during sputtering. On the other hand, when present in the plasma, hydrogen increases the amount of oxygen atoms in a very remarkable way. These remarks do not confirm the commonly adopted explanation th~it there are molecules of H20 trapped in the amorphous alumina and the samples. However, it can reasonably be stated that there is a lack of oxygen in the deposited alumina.
4. D.c. conductivity
The variations of the d.c. resistance was measured in situ during sputtering. As an example we showthese variations in Fig. 1 for one target configuration (169 nickel plates on the alumina disc) and for four deposi- tion conditions. It is evident that the better the base pressure, the sooner the critical percolation thickness is reached. The same kind of dependence is found for iron cermets, and will influence strongly their IR optical properties (see Section 6). The correct explana- tion is suggested by observation of the electron diffrac- tion diagrams (Fig. 2). In fact the characteristic diffraction tings of NiO appear much clearer for the sample obtained under a base vacuum of approxi- mately 10 -6 Tort than for that obtained at approxi- mately 10-7 Torr. We note here that there is no ring corresponding to metallic aluminum. From this obser- vation we infer that oxidation is the main factor delay- ing the percolation threshold by lowering the conductivity.
T A B L E 1. Comparison of the global deposition rate vs. the plasma composition and base pressure, for two different target configurations
Plasma composition Base pressure Rate for Rate for
(Torr) 180 metal 79 metal plates plates (hmin-') (Amin -~)
Pure argon 4 x 10 -6 137.7 93.3
2.6 x 10 -6 131.5 94.0
Pure argon 2 × 10 -7 160.9 127.5
2 x 10 -7 180
A r g o n + 10%H 2 2.8 × 10 -6 96.2 71.4
6. Optical properties
We took optical measurements of the reflectance R and transmittance T in the range 0.35-2.7/zm, using a Varian V spectrophotometer, of Fe-AI203 and Ni-AI203 cermet thin films. In order to reveal the influence of the deposition conditions, we focus on Fe-A1203 cermet films, comparing the transmittance of two sets of three films. All these films had very similar thicknesses (100 nm < d < 120 nm). Each set of three films was produced with the same target configuration.
Throughout the wavelength spectrum, the optical properties depend essentially on the deposition condi- tions, i.e. the base vacuum pressure and the composi- tion of the plasma (pure argon or argon+10%
hydrogen). We divide the study into two ranges, above and below about 1.5 ~m.
o o
t U
t / ) i i i t r
1oooo
1oo
10
0.1 o ,~ 16o
DEPOSITED THICKNESS (nm.)
Fig. 1. Variation of the d.c. resistance vs. the deposited thickness for four deposition conditions with the same target configuration (169 nickel plates on alumina): • base pressure 4.6 x 10 -6 Torr, pure argon plasma; rn base pressure 2.6 x 10 -6 Torr, pure argon plasma; x base pressure 8 x 10 -7 Tort, pure argon plasma; x base pressure 2 x 10-6 Torr, argon + 10% hydrogen plasma.
M. Gadenne et aL / Fe-Al203 and Ni-Al:O.~ cermet filrns 259
(a) (b)
Fig. 2. Electron diffraction pattern of two nickel-alumina thin films (223 nickel plates), elaborated under different base pressures
(a) 10 -7 Torr, (b) 10 -6 Torr.
Above 1.5 /tm, we recognize the characteristic optical behavior of percolative samples. As already pointed out for the conductivity measurements (Section 4), although films of the same set are obtained from the same target configuration and have the same thickness, they behave differently, with regard to both electrical d.c. conductivity and optical properties (Fig.
3). As the d.c. resistance of the samples increases, the transmittance increases for a given wavelength and its slope vs. wavelength increases too. As already shown for granular gold [6] and gold alumina cermets [7], the transmittance of the samples close to the percolation threshold (sample d in Fig. 3) remains almost constant in that range of wavelength.
Below 1.5 /zm, the transmittance of the samples obtained under the best base pressure (approximately 10 -7 Torr) behaves monotonically vs. wavelength. In contrast, the transmittance of samples obtained either under a poor base pressure (a few 10-6 Torr) or with a plasma containing 10% hydrogen shows a dip. The wavelength at which this dip occurs increases with metal concentration (1.1--1.5 pm). We also took reflectance measurements and can deduce that the dip in transmittance is an absorption peak. The same kind of absorptance appears in the same range of wave- length in Ni-AI203 cermets deposited under the same
vacuum conditions, possibly owing to the well known grain resonance [8]. However, we need more experi- mental results (at least a study of spectra of the oxides) to confirm the explanation of this phenomena.
7. Magnetic properties
We measured the magnetization with a vibrating sample magnetometer (VSM) and observed the ferro- magnetic resonance (FMR) in the samples. The measurements were taken at room temperature. The magnetization M (emu cm -3) was expressed in terms of the volume of the sample and not of the metal content.
These results are quite complex to understand in detail although one can reveal some trends.
7.1. F e - A l 2 0 3 f i l m s
VSM measurements show that, as expected, M increases with the iron content and, for a given concen- tration of iron, M is slightly dependent on the deposi- tion conditions and particularly on the sputter gas composition. For instance, in the 180 series, M increases from 530 emu cm -3 to 617 emu cm -3 when the sputtering gas is changed from argon to argon+ 10% H2. The contributions to the magnetic
260 M. Gadenne et al. / Fe-A120 3 and Ni-A120 3 cermet films TABLE 2. Composition in percentage of atoms (microprobe
analysis) of cermets Fe-A1203 cermets
Number of Deposition Atomic concentration X iron disks conditions
Fe AI O Ar
180 Ar 49 13 36 4 102
A r + L N 2 49 13 35 4 101
Ar+ 10%H2 46 13 38 4 101
144 Ar 46 17 35 3 101
A r + L N 2 47 17 34 4 102
Ar + 10%H 2 44 13 40 3 100
79 Ar 30 22 45 4 101
A r + L N 2 31 24 38 4 98
Ar + 10%H 2 33 18 44 4 99
60 Ar 24 25 46 4 99
A r + L N 2 26 27 41 4 98
Ar + 10%H 2 24 18 54 5 98
Ar, pure argon plasma, base pressure approximately 10 -6 Torr;
Ar + LN 2 pure argon plasma, base pressure approximately 10-7 Torr; Ar + 10%H 2 argon + 10% hydrogen plasma, base pressure approximately 10- 6 Torr.
N i - A l 2 0 3 cermets
0.9 0.8
o.r . - ~
O.5 . . . _ . ~ "J
~ o.4 e " - . . . ~ ~
03
0.2 f /
o . ,
/
J
d
J
0 500 1000 1500 2000 2500 3000
WAVELENGTH (rim.)
Fig. 3. Transmittance of iron-alumina films vs. wavelength, for two sets of films. 73 iron plates on the alumina target: (a) thick- ness 114.2 nm, resistance 10.7 Mfl, 2 x 10 -7 Tort, pure argon plasma; (b) thickness 100 nm, resistance 2.1 MQ, 2 x 10 -6 Torr, argon+10% hydrogen plasma; (c) thickness 103.8 rim, resistance 0.68 MQ, 2.6 x 10 -6 Torr, pure argon plasma. 180 iron plates on the alumina target: (d) thickness 119.2 nm, resis- tance 750 ~, 2 x 10 -7 Tort, pure argon plasma; (e) thickness 105.7 nm, resistance 104 kf2, argon+ 10% hydrogen plasma; (f) thickness 115.2 nm, resistance 2.9 kfl, 2.6x 10 -6 Torr, pure argon plasma.
Number of Sample Atomic concentration Precision nickel disks
Ni AI O
180 1 43.8 22.5 33.7 +0.5
2 43.8 22.5 33.7 + 0.5
144 1 40.5 23.8 35.7 + 0.5
2 40.5 23.8 35.7 + 0.5
114 1 33.5 26.6 39.9 + 0.5
2 33.2 26.7 40.1 +0.5
79 1 24.8 30.1 45.1 +0.5
2 24.8 30.1 45.1 +0.5
42 1 13.4 34.6 52 +0.5
2 13.5 34.6 51.9 +0.5
m o m e n t in the samples arise f r o m various sources, such as metallic iron o r the f o r m a t i o n o f a spinel type ferrite, either simply the o r d i n a r y magnetite o r magnetite with A13 ÷ ions substituted in place of Fe 3 ÷ ions. T h e m o r e oxidation occurs, the lower will be the magnetization, because iron has the highest value o f M.
We can obtain m o r e information f r o m F M R studies.
For example, inducing r e s o n a n c e with the fields applied in the plane o f the film and p e r p e n d i c u l a r to it, it is possible to obtain the effective magnetization 4~rMef r which is equal to 4 z t M - H u , w h e r e Hu is the anisotropy field. Knowing 4 ~ M , o n e can calculate H~.
However, if it is assumed that the anisotropy is negligible, because the hysteresis loops show easy saturation in the plane, t h e n the F M R m e a s u r e m e n t s yield directly the value of magnetization M. T h e F M R spectra are quite c o m p l e x and show a greater sensitiv-
ity to the p r e p a r a t i o n conditions. Also, the presence of m o r e than o n e absorption m o d e of c o m p a r a b l e intensity is attributed to the signal from regions of different magnetization. However, the calculations were p e r f o r m e d for all modes, but only those which also give a g factor close to 2.2, which is the signature of iron, were retained for discussion. Using just argon as sputter gas leads to unreasonable values of g.
H o w e v e r for a r g o n + 1 0 % H2 the spectrum gives c o r r e c t g values. T h e value o f M f r o m F M R is higher, indicating that the V S M values were underestimated as a result of uncertainty in the volume of the magnetic species. Also, the line width is very low for the second case, which indicates better quality and h o m o g e n e i t y of the sample.
7.2. Ni-A120~ films
In the case of nickel, o n e could expect a simpler situation c o m p a r e d with iron, because in Ni-A1203 the only magnetic species is metallic nickel. Neither its oxide n o r nickel aluminate spinel is magnetic.
T h e magnetization obtained by V S M increases with the nickel content f r o m 48 e m u cm -3 for 79 discs on the target to 125 e m u cm -3 for 144 discs. C o m p a r e d with the magnetization of 4 5 0 emu cm -3 for bulk nickel, these values are considerably lower, indicating some oxidation a n d / o r mixing with A1203. Again the exact v o l u m e o f magnetic nickel being not known, the M values m a y be an underestimate. E x a m i n a t i o n of the results f r o m F M R will give m o r e information. Figure 4
M. Gadenne et al. / Fe-ALO3 and Ni-Al20 ~ cermet films 261
dl I dH (arbitrzzry units)
N i - A I 2 0 3 180 d i s k s
6 0 O 0
/
X (r~-*us~,)
Fig. 4. FerroMagnetic resonance (FMR) spectrum of a nickel- alumina film (180 nickel plates).
shows the F M R spectrum for the sample of 180 nickel plates on the alumina target prepared under the best base pressure. The external field was applied normally to the film plane. The spectrum is relatively simple con- sisting of an intense mode, and another mode about one tenth as intense, at a higher field. This rules out the possibility of considering them as spin wave modes. A simple interpretation would be for the two modes to arise from two regions with different magnetic proper- ties. The volume corresponding to the stronger mode is about ten times that for the weaker mode. Once more, the calculations were done here assuming the absence of any anisotropy. The g factors (2.00 and 2.30) are reasonably close to that of metallic nickel (2.18). How- ever, it is seen that the magnetization (260 and 150 emu cm -3) is still lower than that of bulk nickel. The two values of M indicate that there are two regions with different magnetic properties. The region correspond- ing to mode 1 shows a smaller line width, indicating that this region is more homogeneous. It is not possible to draw any more conclusions from these measurements.
The magnetic properties show a strong dependence on the deposition conditions but are not well under- stood at present. We need to carry out other studies such as the temperature dependence of M to determine the Curie temperature and Mrssbauer studies to obtain a more complete picture. In the case of nickel, the reduction in magnetization can be clearly attributed to the oxidation or mixing of nickel with alumina.
8. Conclusion
When studying cermets, the dispersion of the experi- mental results corresponds mainly to the multiplicity of the deposition conditions. The influence of metal oxidation is of prime importance. Curiously the presence of hydrogen in the argon plasma does not prevent oxidation, but the good quality of the base pressure is very effective. The composition, the size of the clusters, and consequently the electrical d.c. con- ductivity, the critical percolation thickness, the optical transmission in the visible and near IR are very sensi- tive to the deposition conditions of the films. It is thus necessary to be very cautious before planning a volumic metal concentration and a value of conductiv- ity or transmittance.
However, with carefully characterized samples, one can confirm the strong correlation between the com- position in conducting inclusions and the electrical and optical properties above and below the percolation threshold. As the metallic inclusions are magnetic, they induce a magnetization in the samples more strongly if they are less oxidized. The nickel and iron cermets show very similar behavior, but the explanation of the mechanisms is much more complicated for iron cermet films, because there are three different iron oxides with different magnetic behaviors.
References
1 B. Abeles, P. Sheng, M. D. Couts and Y. Arie, Adv. Phys., 24 (1975)407.
B. Abeles, in R. Wolfe (ed.), Applied Solid State Science, Vol.
6, Academic Press, New York, 1976, p. 2.
2 M. Gadenne and P. Gadenne, PhysicaA, 157(1989) 344.
3 M. Gadenne, P. Gadenne, C. SeUa and J. C. Martin, Thin Solid Films, 221 (1992) 183.
4 P. Gadenne, C. Sella, M. Gasgnier and A. Benhamou, Thin Solid Films, 165 (1988) 29.
5 J. L. Dormann, P. Gibbart, G. Suran and C. Sella, Physica B, 86-88 (1977) 1431.
C. Sella, A. Kaba, S. Berthier and J. Lafait, Sol. Energy Mater., 16 (1987) 143.
6 P. Gadenne, Thesis, Universite Pierre et Marie Curie, Paris,
•986.
P. Gadenne, A. Beghdadi and J. Lafait, Opt. Commun., 65 ( 1 ) (1988) 17.
7 M. Gadenne; Thesis, Universite Pierre et Marie Curie, Paris, 1987.
M. Gadenne, J. Lafait and P. Gadenne, Opt. Commun., 71 (5) (1989) 273.
8 S. Berthier, Ann. Phys. Fr., 13 (1988) 503.
C. G. Granqvist and O. Hunderi, Phys. Rev. B, 16 (8) (1977) 3513.
