DIELECTROPHORETIC ARCHITECTURES - Bio-Inspired and Nanoscale Integrated Computing - page 163

response resistivity of 3.8 10 3 O cm (6.0-M O resistance), consistent with a B
doping ratio of 2000:1, and an estimated barrier potential of 2.0 V. The barrier
potential was higher than the 0.34 V measured in evaporated Au/p-Si junctions
[29], and is attributed to incomplete contact of the nanowire with the electrodes.
Further confirmation that dielectrophoretically trapped nanowires acted as
interconnects was provided by substituting ethanol as a solvent and permitting the
substrate to dry after trapping. A pair of nanowires thus trapped, as shown in
Figure 5.7e, appeared to rest on both electrode faces but initial voltage sweeps
yielded a 30-M O resistance, as shown in Figure 5.7f. After several sweeps, however,
a sharp current turn-on was observed at 6.8-7.8V, as shown in Figure 5.7g, which
may indicate electrostatically induced bending of the nanowires to better contact
the electrodes. Above the turn-on bias, the nanowires exhibited a 1.7-M O
combined resistance, which is compatible with the solvent-based result.

5.3.3. Reconfiguring Nanowire Interconnects
The reported method for trapping nanowire interconnects furthermore enabled
reconfiguration, since nanowires were maximally polarized when aligned between a
pair of electrode tips. Reconfiguration of a nanowire bundle was achieved using
''source'' and ''drain'' electrodes with opposite phase and a ''latch'' electrode with
variable phase. Several nanowires were independently trapped between the source
and latch electrodes with 100-ms-period bursts, as shown in Figure 5.8a, and then
bundled with 250-ms-period bursts, as shown in Figure 5.8b. The phase of the latch
electrode was then inverted, with the same burst period, causing the nanowire
bundle to experience a dielectrophoretic force toward the drain electrode.
Because the electrode tips were arranged in an isosceles right triangle formation,
the torque about the latch electrode was higher than that about the source electrode,
and the nanowire remained in contact with the latch electrode during the motion, as
shown in Figure 5.8c. The reconfiguration was completed 0.25-0.75 s after the phase
inversion, as shown in Figure 5.8d, and could be reversed by restoring the original
phase of the latch electrode, as shown in Figure 5.8e.
Similarly, parallel reconfiguration of a pair of nanowire interconnects was
achieved with four electrodes in a 10- m m square, as shown in Figure 5.9. To
accomplish this, the relative phases between diagonally opposite electrodes were
fixed and the relative phase of the diagonal pairs was inverted. Dielectric break-
down of some nanowires occurred and short fragments were visible around the
connecting nanowires.
For theoretical comparison, we now calculate how rapidly a nanowire might
be dielectrophoretically reconfigured. The dielectrophoretic force per unit length
on a cylindrical nanowire with length l wire and diameter d wire is given by

n o

1

8 e solv p d wire Re *

*

ð * 2

rð * 2

F dep ¼ ðÞr Þ Þ

is the solvent permittivity and *

where e solv ðÞ is the frequency-dependent

Clausius-Mossotti factor [6].