Asymmetric interface broadening in epitaxial Mo/W (001) superlattices
grown by reactive magnetron sputtering on MgO (001) substrates

E. B. Svedberg, J. Birch, I. Ivanov, E. P. Munger and J.-E. Sundgren

Thin Film Physics Division, Department of Physics
Linkoping University S-581 83 Linkoping, Sweden

Asymmetric interfacial broadening in sputtered epitaxial Mo/W
superlattices (SL) using both Ar and Kr discharges is reported. The
different atomic masses of Mo and W, implies that the conditions
during sputtering of W on Mo can be expected to be different than the
conditions prevalent during sputtering of Mo on W. Possible effects on
the sputtering process should also be different in Ar and Kr
discharges, since the fluxes and energies of the backscattered
neutrals strongly depend on the gas-to-target mass ratio. Special
interest is taken in the effect of the interfaces due to their
importance in phenomena such as giant magneto resistance. 

100 periods composed of alternating 44 A Mo and 54 A W layers were
deposited at ~700 C in a dual magnetron HV sputtering system, first in
Ar and then in Kr. MgO (001) substrates were cleaned and annealed
prior to these depositions. Conventional theta-2theta  x-ray
diffraction (XRD) was carried out in the low angle region (1-10 deg.
2theta) and high angle region around the Mo/W 002 Bragg reflection
(50-66 deg. 2theta).  Single crystal Mo and W SL were observed with
more than 10 superlattice peaks indicating that the SL had well
defined and ordered layers. 

In order to deduce the interface broadening and possible asymmetries
in the broadening of W on Mo compared to Mo on W, either XRD or high
resolution electron microscopy can be used. However, both techniques
demand extensive simulations with several fitting parameters and the
latter also requires tedious sample preparation. With XRD it is
difficult to achieve a unique fit unless the films have a very high
crystalline quality allowing high resolution rocking curves or
reciprocal space maps to be recorded to provide additional
information. Consequently in metallic SL, the fitting procedure of
low-angle diffractograms does not lead to an unambiguous determination
of the interfacial structure. An alternative approach to deduce
interface broadening during sputter deposition is to simulate the
growth process and use the simulated compositional profiles as input
parameters for the x-ray simulations. 

To simulate the growth process, the TRIM90 code was initially used to
calculate the number of Ar and Kr atoms reflected from the targets.
The results from these calculations were then used to determine the
average energy of these particles impinging at the substrates, taking
into account that the atoms have to cross the sputtering gas to reach
the sample surface. At the substrate, the TRIDYN code was then
employed to dynamically calculate the intermixing effects occurring
during growth. The bombardment of the growing W on Mo interface by
energetic argon species reflected from the W target caused significant
elemental intermixing with a typical width of ~1.2 nm. For comparison,
the Mo on W interface width is only 0.5 nm. Finally, simulations of
the low-angle diffractograms were made with the Philips GIXA software  
in order to validate the interfacial mixing from the XRD data. The
compositional profiles used for the asymmetric interfaces were taken
directly from the TRIDYN calculated profiles. The low angle x-ray
diffractograms from SL with asymmetrically broadened interfaces were
compared with the measured x-ray diffractograms. The good overall fit
of the simulations to the measured diffractograms is a indicator that
the interface broadening is indeed asymmetric.