Study of Natural Oxidation of Ultra-Thin Aluminum Layers with In-Situ Resistance Measurement
- PDF / 237,942 Bytes
- 6 Pages / 420.48 x 639 pts Page_size
- 16 Downloads / 149 Views
ABSTRACT Exposure to oxygen (1 Torr) at room temperature of thin Al films deposited by UHV ion beam sputtering has been studied using an in-situ resistance measurement set-up. Two lock-in amplifiers allow low noise data acquisition. By monitoring the conductance during deposition and oxidation we can deduce the consumed Al thickness as a function of exposure time (t). It is found that the Al/vacuum interface is diffuse for electron scattering. A two-stage mechanism for natural oxidation is revealed: fast growth (for t 30k we found after 5 minutes 12k < °xtAi < 13k. In contrast,
the consumed Al is 6k for °tAl = 9k.
186
70 60 E 0
50
C>
0 40 U-
Cu
30 20 101--
--
20
40 60 Al Thickness tAl (A)
80
Figure 2: The conductance monitored during Al deposition (plain line) is fitted with the FuchsSondheimer model. The bulk resistivity, mean free path product is po4 - 7.2 10-6 pQi cm 2 . The scattering factor p = 0.0 indicates the electron scattering at the Al/vacuum surface is diffuse.
0 C') m-
-1
CL 0~ L.
-2
C..) C:
-3
----
0~
01
4-,
-4 C: 0
C-
-5 0
75 25 50 Initial Al thickness OtAl (A)
Figure 3: Conductance drop (conductance after oxidation minus conductance before) for the set of bilayer Ta/Al with various Al thickness.
187
1210 -
02
pressure change
S t) Soo
Cn
6E
2n
tl
0
I-
--
4+
experiment fit
0
2 0
+ + +++
0.001
0.01
0.1
1
10
Oxidation time t (min)
Figure 4: The consumed Al is obtained as sketched on Figure 1.The fit is made with Equation (4) and the parameters in (5). The ripple on the experimental curve is due to the pressure change as 02 is introduced.
MODEL We have seen from Figure 4 that the second stage of oxidation is a logarithmic growth. Therefore, the Mott-Cabrera mechanism may not be operative, nor the diffusion [1]. Following Eley and Wilkinson (place exchange model) [11], we describe the oxide growth with x, the oxide thickness and p, a constant dependent on the film structure: dt
exp(-k
(1)
Now we consider the oxidant species concentration in the bulk of the gas (C*), at the vacuum/oxide interface (C.), and at the oxide/metal interface (C(x)). We can write the oxidant species flux from gas bulk to vacuum/oxide interface (FA) and from oxide to oxide/metal interface (FB) as follows [12]:
FA = h(C*- Co), FB = koC(x),
(2)
where the coefficients h and k 0 depict the transport rate from gas bulk to vacuum/oxide interface and the reaction rate of oxidant with metal at the oxide/metal interface. Because of Equation (1), we can write: C(x) = Coexp(-ltx),
188
(3)
where CO is also a function of x. For the steady-state solution, the fluxes are equal and proportional to the oxide growth rate. This leads to t = x (Cn)+ 1(C n)(e
-)
,
(4)
where n is the density of oxidant species in the oxide. For long oxidation time, Equation (4) shows a logarithmic dependence in time of x. Thus the total oxidation rate is governed by place exchange rate (p) and oxidant/metal reaction rate (ko). While for short oxidation time, a linear form is found: the total oxidation
Data Loading...
