Microscale Testing and Characterization Techniques for Benchmarking Crystal. . .
99
important [60]. By contrast, finer milling processes use a lower current and a more
sharply focused beam to achieve precision in the final geometry. Combining them
in a two-step process provides the most efficient path for machining microtensile
samples.
Another technique that can lead to more efficient FIB machining is to design
the machining path based on the shape of the ion beam. Even when the focus of
the beam is corrected as much as possible, there is still a possibility of beam tails.
These regions of the beam can be problematic because material can be removed in
undesired regions, which can cause damage or undesired taper, and the sputtering
rate in this region can be higher or lower than the rest of the beam [61]. Although
it seems like minimizing the tail would be the best method, it can actually be
more effective to take advantage of the shape of the beam with the tail in order
to simultaneously use a higher current and achieve faster milling rates without
sacrificing machining quality. In addition, by calibrating the beam to have the tail
occur only on one edge of the beam and having a clean cut with the other edge, it is
actually possible to have the beam cut even better than simply minimizing the effect
of the tail for a symmetric beam.
In the current study, an automated procedure for machining microtensile samples
with the FIB was developed using Python scripts to control a Tescan Lyra dual beam
system for both imaging and milling. For each thin foil sample to be machined, a
circular fiducial marker was first milled within 200 μm of the edge of the foil to
provide a reference of where the top sample grip section was to be located. Using
image processing through Python, the fiducial marker was identified and used for
automated realignment of the beam as the sample was rotated for cutting. The
sample geometry was discretized into a set of cutting steps, with corresponding
rotation angles and positions relative to the fiducial marker and beam size, for each
cut that was made. The sample stage was rotated to the appropriate angle, where
the image was centered on the fiducial marker, and then subsequently moved to
where the cut was to be made and the cutting process performed. This procedure
was repeated for each step of the sample geometry to complete one pass of the
geometry. Each subsequent pass was performed to cut a geometry that approached
the final shape and used less current to minimize damage in the final sample. Using
a beam current of 3–5 nA for initial bulk passes and then reducing to 500 pA was
effective for minimizing surface damage. Images of a sample after the first cut of
this machining process and the final sample are shown in Fig. 1.
This combination of using bulk and fine milling, optimizing the beam shape, and
taking advantage of the asymmetric shape of the beam rather than working around it
provide an effective and automatable method for producing the complex geometries
found in microtensile samples, as well as in micropillars and microcantilevers.
Despite the precision and widespread use of FIB milling, it does have limitations.
One of the major limitations is the scale at which samples can be fabricated.
Slow material removal rates, typically on the order of 1 μm 3 /nAs or less, make
it impractical to FIB-machine samples with dimensions larger than 50 μm in a
reasonable amount of time [59, 61–63]. In addition, machining of larger samples
requires more care, as machining across longer distances makes it difficult to
99
important [60]. By contrast, finer milling processes use a lower current and a more
sharply focused beam to achieve precision in the final geometry. Combining them
in a two-step process provides the most efficient path for machining microtensile
samples.
Another technique that can lead to more efficient FIB machining is to design
the machining path based on the shape of the ion beam. Even when the focus of
the beam is corrected as much as possible, there is still a possibility of beam tails.
These regions of the beam can be problematic because material can be removed in
undesired regions, which can cause damage or undesired taper, and the sputtering
rate in this region can be higher or lower than the rest of the beam [61]. Although
it seems like minimizing the tail would be the best method, it can actually be
more effective to take advantage of the shape of the beam with the tail in order
to simultaneously use a higher current and achieve faster milling rates without
sacrificing machining quality. In addition, by calibrating the beam to have the tail
occur only on one edge of the beam and having a clean cut with the other edge, it is
actually possible to have the beam cut even better than simply minimizing the effect
of the tail for a symmetric beam.
In the current study, an automated procedure for machining microtensile samples
with the FIB was developed using Python scripts to control a Tescan Lyra dual beam
system for both imaging and milling. For each thin foil sample to be machined, a
circular fiducial marker was first milled within 200 μm of the edge of the foil to
provide a reference of where the top sample grip section was to be located. Using
image processing through Python, the fiducial marker was identified and used for
automated realignment of the beam as the sample was rotated for cutting. The
sample geometry was discretized into a set of cutting steps, with corresponding
rotation angles and positions relative to the fiducial marker and beam size, for each
cut that was made. The sample stage was rotated to the appropriate angle, where
the image was centered on the fiducial marker, and then subsequently moved to
where the cut was to be made and the cutting process performed. This procedure
was repeated for each step of the sample geometry to complete one pass of the
geometry. Each subsequent pass was performed to cut a geometry that approached
the final shape and used less current to minimize damage in the final sample. Using
a beam current of 3–5 nA for initial bulk passes and then reducing to 500 pA was
effective for minimizing surface damage. Images of a sample after the first cut of
this machining process and the final sample are shown in Fig. 1.
This combination of using bulk and fine milling, optimizing the beam shape, and
taking advantage of the asymmetric shape of the beam rather than working around it
provide an effective and automatable method for producing the complex geometries
found in microtensile samples, as well as in micropillars and microcantilevers.
Despite the precision and widespread use of FIB milling, it does have limitations.
One of the major limitations is the scale at which samples can be fabricated.
Slow material removal rates, typically on the order of 1 μm 3 /nAs or less, make
it impractical to FIB-machine samples with dimensions larger than 50 μm in a
reasonable amount of time [59, 61–63]. In addition, machining of larger samples
requires more care, as machining across longer distances makes it difficult to