48
A. Łapicki et al. / Journal of Magnetism and Magnetic Materials 240 (2002) 47–49
ultimate focusing. The focusability is also the factor
determining the selection of possibly lowest controllable
ion current (2–4 pA). Such conditions are important if a
high resolution of deposited patterns is desired. Using
argon or other noble gas ion beams might be advisable if
gallium contamination becomes critical. Any pattern
(although limited in size) of deposited particles can be
fabricated if properly designed and processed through
the vector scan rastering scheme supported by software
provided with the apparatus.
backscattering than cobalt. In addition, those carbonyls
that do get trapped in surface layers still have a chance
to desorb due to intensive thermal activation. The
explosive character of decomposition certainly contri-
butes to a lower ultimate resolution of deposited
patterns in comparison with the FIB focusing limit.
Secondary processes such as thermally activated surface
and bulk diffusion (resulting either in enhanced segrega-
tion or mixing), possible defect cascades within the
substrate lattice and formation of complex defects, all
work against the sharply defined patterns. Charging
effects and mechanical vibrations are also important
factors but can be minimized. Problems of local over-
heating (promoting some undesirable secondary effects)
can be minimized when the fabrication procedure is
performed as a series of rather quick scanning cycles
over a larger pattern instead of sequential (serial)
deposition of individual dots. The satisfactory results
The deposition occurs at a region of overlapping
molecular and ion beams. It has been suggested that
mechanisms of such IBICVD processes [10] are better
described in terms of synergetic effects of several
phenomena than as a simple model. So far nobody has
been able to quantitatively model even individual
aspects of the problem as, for example, the deposition
rate. The first step of the process is energy exchange
between highly energetic gallium ions and almost
stationary (thermal energy distribution) carbonyl com-
plex molecules. The energy of 30 keV is rather excessive
in terms of thermodynamics so it is rational to look for
other mechanisms than those modeling thermal decom-
position. Among several energy exchange pathways like
electromagnetic radiation, electronic excitation and
atomic collisions, the latter one should be dominating
in the experimental conditions under considerations.
For the same reason (high energy, heavy species), it may
be sufficiently adequate to describe the collision inter-
actions between the gallium ions and precursor mole-
cules in terms of a hard sphere model. Dynamics of such
approximation can be presented as an instant and
complete momentum transfer from the ion moving with
a velocity of about 300 000 m/s to a heavier molecule.
After the collision, the gallium ion/atom is furnished
with only thermal-scale velocity oriented in a random
direction. The precursor molecule will gain a significant
momentum towards the surface and an explosive
decomposition will be initiated. It is rather hard to
predict the exact time sequence and velocity distribution
among cobalt atoms and CO molecules (some may also
undergo decomposition). The recoiled species will
probably need several nanoseconds to reach the surface
which is significantly a longer time than the periodicity
of intramolecular vibrations, and therefore, it is rather
safe to assume that the decomposition process of the
precursor molecule is quite advanced at the moment of
the second collision, this time with the surface. However,
the presence of even unbound carbonyl radicals still
moving almost together with cobalt atoms has probably
a cushioning effect and subsurface penetration of cobalt
is rather limited. The impact energy is being dissipated
over neighboring atoms through enhanced lattice vibra-
tions (local heating) and this will certainly finalize the
dissociation process. The lighter carbonyl species as well
as carbon or oxygen atoms have a higher probability of
were obtained by
a combination of 5-point dot
definition, 128 ms dwell time and 60 000 cycles for an
array of 100 dots. The total time of such an experiment
was about 40 min. The deposited particles can be as
small as about 150–200 nm when deposited on a smooth
surface like MgO(1 0 0) substrate (it has been confirmed
for the deposition of tungsten), but were larger in the
presented case of Collodion membranes (1.5–2.5 mm).
3. Results and discussion
General topographic appearance of deposited islands
was verified in situ by ion beam microscopy (secondary
electron imaging, SEM) as shown in Fig. 1. Magnetic
Fig. 1. Ion beam microscopy (SEM) of cobalt dots deposited
by IBICVD onto Collodion membrane. The inset shows
remanent states after saturated magnetization in two opposite
directions parallel to the surface plane, measured by MFM.
Magnetic field strength applied is 2000 Oe.
Lapicki, Adam
