ONR/ARPA Progress Report
April 1, 1995 to August 31, 1995
Grant Number: N00014-95-1-0986
1. Overview
Research in the Artificial Muscle Group at the Artificial Intelligence
Laboratory is currently focused on three areas. First, we are
developing the mechanics and subsystems necessary for a practical
polymer gel based actuator. This includes the polymer, stimulation
system, packaging, force transmission mechanics, energy storage,
dynamic model, and control system. Numerous subprojects in this area
have included (1) a compliant actuator sheath allowing linear,
anisotropic expansion/contraction, (2) load balancing transmission
mechanics coupling multiple gel fibers to a single inert cable, and
(3) compliant, interstitial fluid irrigation system for acid/base gel
stimulation.
Second, we have emphasized the development of a
realistic, yet practical actuator model and control system. The
objective of this research is to develop a lumped parameter model of a
polymer gel based actuator which is simple enough to be implemented in
the real-time control system, yet thermodynamically correct to
describe correct inter-domain energy coupling behavior.
Finally, we are chemically modifying the poly(acrylonitrile) (PAN) gel
fibers in order to eliminate the undesired hysteretic phenomenon while
retaining its superior mechanical properties. We have used our
knowledge of the principles of polyelectrolyte networks and polymer
structure-property relations to design-in the required properties for
the hydrogel actuator. We have devised several approaches to polymer
modification and functionalization of the pre-oxidated, saponified PAN
fibers to obtain gel response within narrow pH ranges. These include:
(1) conversion of carboxylic acid groups to amino groups and (2)
convert the basic pyridine groups to charged species.
2. Polymer gel actuator model and control system
2.1 Overview
The objective of this research is to develop a lumped parameter model
of a polymer gel based actuator which is simple enough to be
implemented in the real-time control system, yet thermodynamically
correct to describe correct inter-domain energy coupling behavior.
The research consists of three parts: (1) model development, (2) model
verification on a actuator prototype, and (3) application of the
dynamic model using a real time control system. The lumped parameter
model based on the bond graph formalism has been developed for both
the non-electrolyte and electrolyte gel. The model unifies the gel's
equilibrium swelling behavior with the ion transport kinetic model
under a single formalism. In addition, the model verification test
station is developed, and we are currently ready to begin collecting
data.
2.2 Model Development
2.3 Model Verification
A testing apparatus was designed and built over the summer to test force
characteristics of artifical muscle. The testbed was designed to
hold one fibre of PAN gel or a small sheet of PVA/PAA gel. The apparatus
was build so that the acid or base which would cause the gel to
contract or expand would fill the container of the testbed and immerse
the fibre so that the rising fluid created as little turbulence
as possible. A laser diode and photodetector detects when the
fluid level reaches the gel so that a time mark can be set on the
force measurements on the gel.
The research work done this summer involved the construction
of a testbed to the determine the forces and contraction
rates produced by a single bundle of PAN fibers or PVA gel
strips, when submerged in a caustic solution. Emphasis on
the design of the testbed included the reduction of noise
resulting from a turbulent flow from an inlet valve and a
way of determining the exact moment that the gel strip made
contact with its surrounding fluid medium.
With this in mind, we came up with the following design, figure 1.
To reduce the noise problem, rather than
lowering the fiber into the fluid medium and worrying about
splashing or back-wash eddie currents, we simply decided to
let the fluid reach the fiber in a uniform manner. Taking
this into account that the inflow coming from the inlet
valve is turbulent, we thought of adding a device which
would make the rising fluid more laminar. To do so, we
constructed a porous slab made of acrylic to slow down the
incoming flow before reaching the fiber. We set it at a
height of approximately 1 in. above the inlet valve. (We
have not conducted any tests to determine an optimal height
or design of the acrylic slab to produce a steady, yet quick
rise of the fluid).
Finding a mechanism to determine the moment the gel-strip
made contact with its surrounding fluid, in order to
activate a force vs. time plot for the fiber was
difficult. We resolved the issue (we think) by implementing
a laser-photosiode setup. At exactly the same height as the
fiber, but parralel to it, we set up a laser and photodiode
at opposite ends of the testbed. The object of this , was to
let the fluid itself split the laser beam, which in turn
would activate a timer and start the plot in the computer.
Other important details that need mentioning, is the fiber
termination and pulley attachments. On one end of the
testbed, a small hole was drilled with the intent of
slipping one end of the nylon string embedded in epoxy (the
nylon string is attached to both ends of the fiber which is
in the middle of the testbed), to insure a safe attachment
to one of the testbed walls. On the other side, the pulleys
were added to maintain the motion of the fiber to be as
linear as possible.
Finally, a teflon sheet was added just below the fiber to
prevent any excessive bendind of the fiber. The extension of
the testbed was designed to accomodate both the force-sensor
and laser setup so that the unit functions as one system
rather than discrete parts.

Figure 1. Polymer gel actuator testbed.
3. Polymer Gel Functionalization and Modification
3.1 Initial Approach
Artificial muscle systems present a new and challenging application
for polymer hydrogels, in which mechanical strength and amphoteric
response must be optimized. The current commercially available PAN gel
fibers have excellent mechanical characteristics, but exhibit
significant hysteresis that limits the overall efficiency of the gel
actuator system. We will use our knowledge of the principles of
polyelectrolyte networks and polymer structure-property relations to
design-in the required properties for the hydrogel actuator. We have
devised several approaches to polymer modification and
functionalization of the pre-oxidated, saponified PAN fibers to obtain
gel response within narrow pH ranges.
The high level of hysteresis observed in the current PAN gel fibers is
due to the presence of both highly basic pyridine and carboxylic acid
groups. There is a large pH range for which the pyridine groups are
protonated but the carboxylic acid moieties remain anionic. There are
a few approaches to this problem, each of which involves either
deactivation of the basic pyridine group and/or conversion of the
carboxylic acid group to a more basic functionality.
- 1) Conversion of carboxylic acid groups to amino groups: Two methods are:
- a) Amidization of carboxylic acid with dimethylamino ethylamine
to obtain a dimethylamino group as the electrolytic group. The
pKa of such a functionality woold be about 10 or 11. In the
presence of the pyridine groups, the pH range shoold be
considerably narrower, and is expected to
fall between approximately 9 and 11.
- b) Conversion of carboxylic acid to primary amine groups using
standard reagents such as i) NaN3, H2SO4, followed by NaOH
(Curtius rearrangement); ii) aminoalcohol, followed by thionyl
chloride and heat (Lossen reaction), or other common
methods. The primary amine is also expected
to be a strong base, and to respond in a range close to that
of the pyridine groups.
- 2) Convert the basic pyridine groups to charged species: This
approach eliminates the effects of modolation of the gel behavior
between two different electrolytic groups. (In this case, the polymer
gel fiber may exhibit an amphoteric response moderated or affected in
part by the presence of permanently charged species.) Two methods are:
- a) Alkylation of the pyridine groups with an alkyl bromide, thus
forming a positively charged pyridinium bromide group. This approach
woold work best in conjunction with conversion to the carboxylic acid
groups to amino groups, as the charges incorporated woold be the same.
- b) Substitution reaction of pyridine with soltone to produce
pyridinium alkyl solfonate groups in which the positively charged
pyridinium group is covalently attached to a negatively charged
solfonate group; the net charge of the moiety woold be effectively
neutral. This woold work in conjunction with either carboxylic acid
groups or amino groups.
Each of the above methods could be used alone, or in conjunction with
each other, to determine the amphoteric response of the gel. Other gel
systems, including systems based on changes in ionic strength rather
than pH, are possibilities for further polymer design.
3.2 Current Approach
One current approach to modifying the polymer gel involves the
conversion of the carboxylic acid group to an amino group, a more
basic functionality. In theory, this should cause the
expansion/contraction behavior of the gel to reverse, i.e., the gel
should swell in acid and remain neutral in base. The modification
reaction involves the addition of solid sodium azide and concentrated
sulfuric acid to a sample of the saponified gel fiber. The mechanism
is as follows:
R-COOH + NaN3 + 2H2SO4 --> R-NH3+HSO4- + NaHSO4 + N2 + CO2
R-NH3+HSO4- + 2NaOH --> R-NH2 + H2O + Na2SO4
A second approach that is being tested is the addition of a
large excess of ethylene diamine to a sample of the saponified
gel. The amino groups in the ethylene diamine should react with the
carboxylic acid groups to form a secondary amide, which is essentially
hydrophilic, and thus should have the same behavior in acid and base
as an amino group.
4. Future Development
Beyond the actuator modelling and polymer gel modification, our next
major goal is to construct a completely compliant gel "muscle." This
includes a compliant covering, irriation system, tendon terminations,
and supply lines. We will also concentrate our efforts on minimizing
extraneous support mechanisms, thus maximizing the effective power per
unit volume. Our intention, at the end of the next year, is to develop
a multiple actuator system, suitable for multiple degree-of-freedom
robotics or anthropomorphic prosthetics.