Determination of native Al oxide thickness by XPS

       The thickness of native oxide on aluminum can be estimated by a simple XPS (sometimes also referred to as ESCA) method.^1-6 This method can also be applied to the measurement of thin (i.e. < -100 A) oxide films on other metals (i.e. MgO/Mg, SiO₂/Si, TiO₂/Ti, etc.).⁴

       When the oxide layer on aluminum is relatively thin (i.e. <75~85 A), peaks of aluminum oxide and the underlying metal can be distinctively observed in the A1 2p XPS spectrum, an example of which is shown in Figure 1. The relative intensities of the oxidic and metallic A1 2p peaks are related to the oxide thicknesses according to a relationship developed by Carlson³ as:

$\frac{I_{m}}{I_{o}}=\frac{N_{m}}{N_{o}}\frac{\lambda _{m}}{\lambda _{o}}\frac{\textup{exp}[-(d/\lambda _{o}\textup{sin}\theta )]}{1-\textup{exp}[-(d/\lambda _{o}\textup{sin}\theta )]}$                                        (1)

where I is the intensity of the photoelectron peak (i.e. peak areas); N is the volume density of metal atoms; λ is the inelastic mean free path (IMFP) of photoelectrons; d is the oxide thickness; and θ is the electron take-off angle with respect to the sample surface. The subscripts m and o stand for metal and oxide, respectively. This expression can be further simplified to give the oxide thickness (d) as:

$d=\lambda _{o}\textup{sin}\theta \textup{ln}[\frac{N_{m}}{N_{o}}\frac{\lambda _{m}}{\lambda _{o}}\frac{I_{o}}{I_{m}}+1]$                                                    (2)

The IMFPs and the volume densities of metal atoms for the metal and oxide must be known to determine the oxide film thickness from Eqn. 2. Fortunately we can find such data from literature.

The most recent IMFPs for Al for electron energies from 50 eV to 30 keV has been calculated by Tanuma et al.⁷ from experimental optical data using the full Penn algorithm. The IMFPs for Al₂O₃ have also been calculated by the same authors.⁸ When using Al Ka x-ray radiation, the kinetic energy of Al 2p photoelectrons is 1415~1418 eV.The results of Tanuma et al.^{7, 8} indicate that, for electrons of this energy, the IMFPs in aluminum oxide and aluminum metal are both ~28 A.

The volume density of metal atoms in the oxide layer will vary, depending on the presence of other elements and the amount of hydration. Strohmeier has reported an N_m/N_o ratio of 1.3~1.5 from estimated densities for γ-Al₂O₃.⁴ In this study, we take an N_m/N_o ratio of 1.4.

We obtain our XPS spectra with a Kratos AXIS Ultra photoelectron spectrometer, which uses a mono Al Ka x-ray source. All binding energies were referenced to the main C 1s line at 284.5 eV. XPS peak areas were determined by curve fitting using standard Kratos software. In all of the computations, the spectral background was assumed to be linear over the peak widths. Peak areas were measured with a precision of +/- 5%. The measurement was taken at an electron take-off angle of 90° (i.e. sin θ=1). The measured sample is a thermally evaporated Al film that has been exposed to the typical laboratory environment of Singapore for two weeks. The measured Al 2p XPS spectrum is shown in Figure 1.

Figure 1 Al 2p peak envelop, showing respective oxide and metal components

         This measurement suggests that for this specific sample, the ratio of intensities of oxide and metal I_o/I_m is ~2.0, which corresponds to an oxide film thickness of ~37.4 A. Considering the extremely large humidity in Singapore, this oxide film thickness is reasonable.

       The accuracy of this measurement directly depends on the approximate values chosen for λ_o, λ_m, N_o and N_m. The values chosen for λ_o and λ_m are among the most recently reported values, and the uncertainty for λ_ois estimated to be +/-5%, while for λ_m is to be +/-15%.^{7, 8} The atomic density ratio No/Nm is estimated to be 1.4 +/-0.1. The error of the measured intensity ratio I_o/I_m is around +/-5%. Hence, the error for the oxide thickness obtained using Eqn. 2 is estimated to be 3.7 A.

        The approximate total sampling depth of XPS is on the order of 3λ, which represents the depth from which ~95% of the total photoelectron signal arises. Therefore, the maximum sampling depth of A1 2p electrons through an aluminum oxide film when using an Al x-ray source is ~84 A.

       The precise measurement of native Al oxide thickness is meaningful for our TDTR (Time Domain Thermoreflectance) measurement, which is very sensitive to the Al transducer film thickness. Native Al oxide is transparent and does not absorb energy from the heating source. However, Al oxide film contributes heat capacitance and affects the measurement. After we have measured the native Al oxide thickness precisely and accurately, we know how much we should add to the Al film thickness to take into account the effect of the native Al oxide film.

References:

1.         T.A. Carlson and G. E. McGuire, Journal of Electron Spectroscopy and RelatedPhenomena 1 (2), 161-168 (1972).

2.         W. J. Carter, G. K. Schweitzer and T.A. Carlson, Journal of Electron Spectroscopy and Related Phenomena 5 (1), 827-835 (1974).

3.         T. A. Carlson, Surf Interface Anal 4 (4), 125-134 (1982).

4.         B. R. Strohmeier, Surf Interface Anal 15 (1), 51-56 (1990).

5.         I. Olefjord and A. Nylund, SurfInterface Anal 21 (5), 290-297(1994).

6.         M. R. Alexander, G. E. Thompson, X.Zhou, G. Beamson and N. Fairley, Surf Interface Anal 34 (1), 485-489 (2002).

7.         S. Tanuma, C. J. Powell and D. R. Penn,Surf Interface Anal 43 (3), 689-713(2011).

8.         S. Tanuma, C. J. Powell and D. R. Penn,Surf Interface Anal 20 (1), 77-89(1993).