IEEE Robotics & Automation Magazine - June 2013 - 66

0 ms

3 ms

6 ms

9 ms

12 ms

15 ms

18 ms

Y Position (nm)

2,800

Center
Front
Back

2,300
1,800
1,300
800
800

Velocity (mm/s)

2,800
X Position (nm)

500
400
300
200
100
0

V_Front
V_Back
V_Center

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
Time (ms)

Figure 2. The captured motions during 2-mm dash task and their tracked trajectory plots of MagPieR. The current imposed value was 1.2 A
(the scale bar is 500 n m).

magnesium niobate-lead titanate (PMN-PT)] for out-ofplane vibration. The microrobot is intended to overcome surface friction. It can move
on the horizontal plane of
a planar capacitor whose
We chose the MagPieR
bottom electrode is the
arena substrate itself while
with the thread crossing
the top electrode is an
the longitudinal axis on the optically transparent conductive glass (tin-doped
indium oxide, ITO).
ferromagnetic nickel layer
Piezo PMN-PT material is by far superior to
to show the better-guided
the classical lead zirconate titanate (PZT)
linear propulsion.
ceramics due to the very
high longitudinal d 33
piezoelectric coefficient (3,100 pC/N when compared
with 590 pC/N). The major strain capability of the new
PMN-PT materials made this type of actuation possible
(which we did not observe for PZTs).

Figure 3. Scanning electron microscope (SEM) images of several
MagPieR agents: Type I images are the square ones featuring two
trenches while Type II images are the rectangular ones with a single
microtrench. (Photo courtesy of FEMTO-ST Institute.)

IEEE ROBOTICS & AUTOMATION MAGAZINE

june 2013

The microrobot fabrication process consisted of cutting the PMN-PT substrate into centimeter-size plates,
sputtering a Cu-Cr primer layer at the top side, sputtering
a Ti-Cr layer at the bottom side, protecting the resin at the
bottom side, Ni electroplating on the top side, saw dicing
into small square or rectangular samples, and saw trenching of the top Ni layer.
The final structures were 224- n m high, of which 200 n m
was the PMN-PT layer and 24 n m was the Ni layer, as shown
in Figures 3 and 4(a). We used two different geometries,
square and rectangular:
3
● Type I, 388 # 388 # 224 n m with two 50- n m-wide
threads
3
● Type II, 388 # 300 # 224 n m with one 50- n m-wide
thread.
The densities of PMN-PT and Ni were almost identical, but the former material was more mechanically compliant and fragile. The mass of the Type II model, which
competed in the challenge, was 0.21 mg, of which less
than 9% consisted of the ferromagnetic core.
Saw trenching was intended to allow for a faster and
more oscillatory-free orientation along the magnetic field
lines. Similarly, we tested bulk samples (without trenches)
that naturally showed faster speeds due to the increased
amount of Ni, but they performed, as expected, in a more
unstable manner.
The microrobot was positioned into a self-fabricated
3.5 # 2.0 mm2 arena, whose dimensions were imposed
by the competition organizers [Figure 1(a)]. To visualize
the scene perpendicularly, the ITO-conductive glass was
considered for the top electrode. The microrobot and the
arena were packaged into a 6.5 # 6.5 mm2 assembly sealed
with two diagonal flanges, as shown in Figure 5, and was

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - June 2013