6034

J. Phys. Chem. 1993,97, 6034-6041

Electrochemical Applications of the Bending Beam Method. 2. Electroshrinking and Slow

Relaxation in Polypyrrole
Qibing Pei'and Olle Inganiis
Laboratory of Applied Physics, Department of Physics, Linktiping University, S-581 83, Linkijping, Sweden
Received: January 20, 1993

Cyclic voltammetry of chloride-doped polypyrrole (PPy(C1)) in an aqueous solution of tetraethylammonium

chloride and dodecyl benzenesulfonate-doped polypyrrole (PPy(DBS)) in an aqueous solution of sodium
dodecylbenzenesulfonate is studied. Emphasis is put on the effect of holding the polymers at the negative or
positive potential limit on the subsequent cyclic voltammograms. The bending beam method, where an active
polypyrrole layer and a substrate polyethylene layer are bound together to sense volume changes in the active
layer, is used to detect the volume changes in PPy(C1) and PPy(DBS) during doping and undoping under
constant potentials and while the polymers are held a t the undoped or doped state. Hysteresis of the first peak
on the subsequent cyclic voltammogram has been observed after holding PPy(C1) at the undoped state and after
holding PPy(DBS) at the doped state. Holding PPy(C1) at the doped state or PPy(DBS) at the undoped state
does not induce this effect. Slow volume shrinking occurs by -0.92% in PPy(C1) held at the undoped state for
10 h and by -0.45% in PPy(DBS) held at the doped state for 10 h. The underlying nature is slow relaxation
of the polymer matrix: the free spaces left in the matrix after the undoping of PPy(C1) due to the ejection of
dopant anions or after the doping of PPy(DBS) due to ejection of cations from the matrix are gradually filled
in by the adjacent polypyrrole chain (in doped PPy(DBS) also by the nearby alkyl chains of DBS-). The volume
of PPy(C1) shrinks by -0.20% whereas that of PPy(DBS) swells by +1.71% during undoping. Re-doping
reverses these volume changes. Such electroshrinking and electroswelling are induced by the transport of
charge compensating ions in the course of the redox.

Introduction
Electrically conductiveorganic polymers such as polythiophene
(PT), polyaniline (PANi), and polypyrrole have, as electrode
materials, received rather extensive study in the past decade.'
They promise several important applications such as light-weight
rechargeable batteries2and electrochromicwindows and displays.3
The common mechanism behind these applicationsare the dopingundoping (redox), charge compensation, and the associated
conformational change, which still remains largely elusive.
In part 1 of this series," we have introduced the bending beam
method to the study of the electrochemistry of tosylate-doped
polypyrrole (PPy(TsO)), where a bipolymer strip made of a PPy(TsO) layer and a polyethylene layer bound together responds,
by bending, to volume changes in the PPy layer during its
electrochemicaldoping and undoping. Correlation between the
local linear strain and the curvature change of the strip has been
developed. It has been determined that cations insert into PPy(TsO) from the solution for charge compensation during the
undoping of this polymer, and the salt (lithium tosylate)
accumulated in the polymer drains out of the undoped PPyO.
Slow phase relaxation in the PPyoafter salt draining is proposed.
In this second paper, two PPy's doped with chloride (PPy(C1))
and dodecyl benzenesulfonate (PPy(DBS)) are investigated.
Emphasis is put on the slow relaxation process.
Slow relaxation has been observed in different redox polymer
systems during cyclicvoltammetry: the "break-in" phenomenon,5
that for a freshly-deposited polymer film requires several cyclic
potential sweeps for the cyclic voltammograms to become fully
overlapping, and the "hysteresis"5b~~~6
of the first voltammogram
of the polymer film after being held in its neutral state. The
hysteresis or "memory" effect is of primary interest, because an
elucidationmay shed light on the insulator-to-conductortransition
in electroactive polymers, especially in electrically conductive
polymer~.~J
Several interpretations have been proposed for the
hysteresis effect, e.g., ejection of solvent or ion pairs from the
polymer matrix; a relaxation of the polymer chains from a planar

to a twisted form; and morphological changes. Besides the

electrochemical techniques, in-situ spectroscopy,sb scanning
electron microscopy,sc quartz crystal microbalance,Sd in-situ
ESR,6esr and millimetric electrodes and ultramicroelectrodes68
have also been used to study the origin of the hysteresis effect.
However, a thorough understanding has not been achieved yet.8
As a striking fact, all those hysteresis effects above were
observed when the polymers were held in the neutral undoped
(discharged) state. For the redox polymer poly(tetracyanoquinodimethane adipate) (PTCNQ),Sfit was held at the high potential
limit of the cyclic voltammogram and the hysteresis was observed
in the following reduction peak (n-doping). For the conducting
polymers PANi and PPy and poly(3-methylthiophene) (PMTh),
they were held at the negative potential limit and hysteresis
appeared in the following oxidation peak (p-doping). Holding
at the charged or doped state seemed to cause no hysteresis effect
in both PTCNQ and the conductive polymers. On the basis of
this, Odin and Nechtschein suggested that the hysteresis originated
from the evacuation of residual electronic charges when the
polymer is held in the neutral state.6e A charge free, highly
insulating area, grew in the polymer from the polymer-substrate
electrode interface outward. Re-doping of the polymer film
needed charge transport over the highly insulating area, and thus
it slows down.
Here we report the presence of hysteresis effect in PPy(C1)
held in the neutral state and in PPy(DBS) held in the charged
state. General volume shrinking in both PPy's during the holding
are observed by the bending beam method. These cannot be
interpreted according to the mechanisms given above. Therefore,
we suggest that this slow relaxation during holding is attributed
to a mechanical process: the voids or open channels left behind
after the ejection of dopants or co-ions are gradually filled in by
the adjacent polymer chains.

Experimental Section
Pyrrole (Aldrich) was distilled under decreased pressure and
stored in the dark below -10 OC. Tetraethylammonium chloride

0022-3654/93/2091-6034%04.00/0 0 1993 American Chemical Society

Electroshrinking and Slow Relaxation in Polypyrrole

(EtdNCl), sodium dodecyl benzenesulfonate (DBS-Na), and all
the other electrolytes were reagent grade and were used without
further purification. Water was distilled and de-ionized. Polyethylene (LDPE, Neste DFDS-6430) film of 150-pm thickness
was coated with 0.2-pm-thickness gold layer by vacuum evaporation. A very thin chromium interlayer was used for better
adhesion between the gold layer and the polyethylene substrate.
This Au/PE film was used, as the working electrode, to
electrochemically polymerize pyrrole in an aqueous solution
containing 0.1 M pyrrole and 0.1 M Et4NCl or DBS-Na. A
standard one-compartment three-electrode cell was used with a
Ag/AgCl reference electrode and a Pt wire counter electrode. A
potential of 0.70 V was applied, and pyrrole was oxidatively
polymerized onto the gold surface to form a laminate (PPy/
Au/PE or PPy/PE). The thickness of the PPy(C1) layer formed
was ca. 10 pm, and that of the PPy(DBS) layer was ca. 15 pm
determined by the charge used for the polymerizationId and
confirmed using a Sloan Dektak 3030 surface profiler.
Strips of the size 2.5 cm long and 0.22 cm wide were cut out
from the PPy/PE laminate. Bending tests were conducted in a
shallow cell equipped with a Ag/AgCl reference and a Pt foil
counter electrode. The PPy/PE strip was positioned in the cell
as the working electrode, one end being clamped and the other
end free to move. A video camera was fixed above the cell to
record the deflection of the free end of the strip during an
electrochemical redox. The data were collected from a monitor
screen to calculate the curvature change (cm-I) of the strip.4This
curvature change is directly related to the volume change in the
active PPy layera4A curvature increase refers to volume swelling
in thePPylayer andstripbending tothesubstratePEside,forming
a convex curve.
Cyclic voltammetry was conducted in the standard onecompartment three-electrodecell with a Pt wire counter electrode
and a Ag/AgCl reference electrode. The working electrode, PPy/
Pt, was prepared by polymerizing pyrrole onto a Pt rod tip
(diameter: 1.7 mm) in the same way as preparing the PPy/PE
strips. A charge of 4.6 mC was consumed for the polymerization.
All the aqueous electrolyte solutions were prepared by dissolving
the appropriate electrolyte, e.g., Et4NCl, in purified water and
were purged with argon gas prior to use at room temperature. An
EG&G PAR Model 173 potentiostat/galvanostat in connection
with a PAR Model 175 universal programmer and a Watanabe
WX4403 X-Y recorder was employed for both PPy growth and
its doping/undoping. All the experiments were conducted in
laboratory air a t room temperature.

Results and Discussion

1. Relaxation in PPy(C1). The first cyclicvoltammograms of
PPy(C1) in an aqueous solution of 0.1 M Et4NC1, after being held
at the negative potential limit (-0.85 V) for a certain time (wait
time, T), are shown in Figure la. The anodic peak current
increases and the peak position shifts to more positive potentials
with increasing wait time, similar to the hysteresis effects observed
in other conducting polymers like PANPdand PMTh6Cand in the
redox polymer PTCNQ.5bscThe subsequent voltammograms after
the first (hysteresis) one develop into the steady-state voltammogram (T = 0 s) either immediately when the wait time is short
(T < 1000 s) or in a few cycles when the wait time is long (7 <
1000 s).

The peak currents and potentials of the first voltammograms

after various wait times are drawn in Figure lb. Neither the
currents nor the potentials follow a simple logarithmic law with
the wait time as in the case of PANi and PMTh.6 Taking into
consideration the time needed for the scanning potential to reach
from the holding point to the point at which PPy begins to be
oxidized does not change the general characteristic of Figure 1b.
We understand that it is reasonable to use the peak current and
potential to define qualitatively the hysteresis effect. However,

The Journal of Physical Chemistry, Vol. 97, No. 22, 1993 6035

4.3

-0.6

-0.3

0.0

E, V us Ag/AgCl
200

'

' " " I

'"8

"

A Peak current

"

Peak potential (V)

'

"

' ' ,

(u)

50,
1

0.50

, , ,

10

, , , '

1M)

, , #

loo0

0.0

loo00

Wait Time (s)

Figure 1. (a) First cyclic voltammograms in an aqueous solution of 0.1
M Et4NC1 of PPy(C1) after being held at the negative potential limit
(-0.85 V) for a certain wait time, T . The voltammogram at r = 0 s is
the steady-state CV after continuous cyclic potential scanning. Scanning
rate: 100 mV s-I. (b) Potential and current of the anodic peak in (a)
as a function of the wait time.

the redox peaks of the cyclic voltammograms of a conducting

polymer are usually complicated with several unclarified factorsag
The peak potential can be taken as the potential at which the
neutral polymer can be oxidatively doped only when the potential
sweeps very slowly. During such a slow potential sweeping, the
memory effect might have been erased before the current peak
appears. Furthermore, conducting polymers can usually be
characterized to be multistep redox systems with different
potentials.I0 Therefore, other parameters having concrete physical
meaning need to be found to define quantitatively the hysteresis
kinetics.
Figure 2 shows the first cyclic voltammograms of PPy(C1) in
an aqueous solution of 0.1 M Et4NCl after being held at the
positive potential limit (+0.3 V) for various wait times T. Only
a small hysteresis effect is observed after a long wait time. No
significant destruction of the polymer has occurred during the
holding, because the subsequent voltammograms tend to overlap
with the steady-state one a t T = 0 s. There is a great difference
between the hysteresis effects due to holding PPy(C1) in the neutral
state and in the doped state. In neutral PPy(Cl), the polymer
chains bear no charge and the polymer matrix contains none or
few ions (Cl- or Et4N+), while, in doped PPy(Cl), the polymer
chains are positively charged with a large amount of CI- around
the polymer chains for charge neutrality:

We propose that during the doping of PPyO, the dopant anions

insert into the polymer and the polymer matrix is dilated to leave
free space to accommodate the dopants which are likely solvated
to some extent, leading to a volume swelling in the polymer matrix.

6036 The Journal of Physical Chemistry, Vol. 97, No. 22, 1993

----.-.-.

.......

Pei and Inganas

= 10 s

2 = 100 8

-.-.-z

= 100 s

-..-..

z=1,ooos

.............7 = 10,000 s
+0.3

-0.3

0.0

-0.6

-0.9

+0.3

E,V us AdAgCl

Upon undoping or discharging, the dopants are ejected out of the

polymer matrix, and the free spaces (voids or channels) left behind
will be filled in by the adjacent polymer chains, leading tovolume
shrinking. This filling of free spaces by polymer chains is a slow
process. During the continuous cyclic potential sweeping, the
time interval is too short to let the neutral polymer chains fill in
the free spaces. The voltammograms overlap well. The free
space-filling occurs while the polymer is held in the neutral state,
so that the subsequent re-doping of the polymer and the insertion
of dopants becomes more difficult, and we see the hysteresis of
the first voltammogram after the holding. Holding PPy(C1) at
the doped state does not have this effect, and there is no subsequent
hysteresis.
During studies of the above cyclic voltammetry, we have chosen
an electrolyte, EtdNCl, with a bulky cation to avoid the insertion
of cations into rather than dopant anion ejection out of PPy(C1)
for charge compensation during its discharging. It has been
observed that the discharging of PPy usually involves the insertion
of cations to form co-ions in the discharged PPY.~."The co-ions
will drain out of the polymer unless the anions are too bulky to
move.'* We observed that the cyclicvoltammograms of PPy(C1)
in aqueous solutions of LiCl, NaC1, and KCl all contain the
contribution from cation insertion in the potential range -0.3 to
-0.7 V, and this contribution decreases with increasing cation
size from Li+, Na+, to K+. In the solution of Et4NCl, the
contribution of cation insertion is negligible so that we see a sole
anion transport process.
2. Relaxation in PPy(DBS). Now, we look at the behavior
of PPy(DBS) where the dopant is very bulky, being sluggish to
diffuse in the polymer matrix.12 The redox of PPy(DBS) relies
on cation transport:
-Na+, -e-

-0.3

-0.6

-0.9

E, V us AglAgCl

Figure 2. First cyclic voltammograms in an aqueous solution of 0.1 M

EtdNCl of PPy(CI) after being held at the positive potential limit (+0.30
V) for certain wait time, T . The voltammogram at T = 0 s is the steadystate one after continuous cyclic potential scanning. Scanning rate: 100
mV s-'.

PPyo(DBS-Na+)

0.0

PPy+(DBS-)

(2)

+Na+. +e-

The first cyclic voltammograms of PPy(DBS) in an aqueous

solution of 0.1 M DBS-Na+ after being held a t the negative
potential limit (-0.85 V) for various wait times T are shown in
Figure 3a. Only slight hysteresis effects are observed, in sharp
contrast to the behavior of PPy(C1). The potential of the first
charging peak at around -0.4 V shifts little after the waiting,
whereas the peak current first increases a little with increasing
wait time, then turns to decrease. Thesubsequent voltammograms
immediately stabilize (overlapping each other), but the peak
currents in the steady-state voltammograms have been decreased
when the wait time is very long. Figure 3b displays the relative

0.7
0

loo00200003000040000

Wait time (s)

Figure 3. (a) First cyclic voltammograms in an aqueous solution of 0.1
M DBS-Na of PPy(DBS) after being held at the negative potential limit
(-0.85 V) for a certain wait time, 7. The voltammogram at T = 0 s is
the steady-state CV after continuous cyclic potential scanning. Scanning
rate: 100 mV s-l. (b) Relative redox charges of the PPy(DBS) after
being held for time T .

redox capacity in the steady-state voltammograms of the PPy(DBS) as a function of the time of waiting in the neutral state.
The capacity decreases almost linearly with wait time, through
the decrease is very slow. Almost 80% of the capacity is still
retained after 10 h of holding.
Figure 4a shows the first cyclic voltammograms of PPy(DBS)
in an aqueous solution of 0.1 M DBS-Na+ after being held at the
positive potential limit (+0.3 V) for a certain wait time. Strikingly,
a rather strong hysteresis effect appears. The cathodic peak
current increases and the peak position shifts to more negative
potentials with increasing wait time, similar to the hysteresis
effect observed in PPy(C1) held in the neutral state. The
subsequent voltammograms after the first hysteresis one develop
into the steady-state voltammogram (T = 0 s) either immediately
when the wait time is short (T < 1000 s) or in a few cycles when
the wait time is long (T > 1000 s). Within T 5 10 000 s, the
subsequent voltammograms lose little capacity; therefore overoxidation or destruction of PPy(DBS) held at +0.3 V can be
excluded.
The peak currents and potentials of the first voltammograms
after varying wait times are drawn in Figure 4b. Neither the
currents nor the potentials follow a simple logarithmic law with
the wait time, similar to the case of PPy(C1) held in the neutral
state. The peak potential shifts less than 0.1 V after 10 000 s of
holding, while, in the case of PPy(Cl), the potential shifts by 0.27
V.

To interpret the hysteresis effect in PPy(DBS) after being

held in the doped state but no hysteresis after being held in the
neutral state, we follow the argument in the case of PPy(C1).
During the discharging of PPy(DBS), Na+ cations from the
solution insert into the polymer and the polymer matrix is dilated
to leave free spaces (voids or channels) to accommodate the

Electroshrinking and Slow Relaxation in Polypyrrole

-..-..
1

4.3

I :

7 = 1,000s

............z = 10,000s

0.0

The Journal of Physical Chemistry, Vol. 97, No. 22, 1993 6031

)i!

.\....:

-0.3

-0.9

-0.6

E,v us AglAgc1

i
-0.52

,Cn

q:

r
Peak current (p.4)

-0.68

10

loo

loo0

loo00

Wait Time (s)

Figure 4. (a) First cyclic voltammograms in an aqueous solution of 0.1
M DBS-Na of PPy(DBS) after being held at the positive potential limit
(+0.30V) for a certain wait time, T . The voltammogram at 7 = 0 s is
the steady-state CV after continuous cyclic potential scanning. Scanning
rate: 100 mV SKI. (b) Potential and current of the anodic peak in (a)
as a function of the wait time.

cations, leading tovolume swelling in the polymer matrix. Upon

doping or charging, the Na+ cations are ejected out of the polymer
matrix, and the free spaces left behind will gradually be filled in
by either the adjacent charged PPy chains or the nearby DBSanions which have long flexible alkyl tails, or by both, leading to
volume shrinking. This ion transport and free space-filling mode
is similar to the case of PPy(Cl), but in the opposite direction.
That is, we see the hysteresis of the first voltammogram after
holding PPy(DBS) in the doped state, as compared to holding
PPy(C1) in the neutral state. No hysteresis occurs for holding
PPy(DBS) in the neutral state or holding PPy(C1) in the doped
state.
It is interesting to observe the cyclic voltammogramsof freshlydeposited PPy(DBS), because there should be none or few free
spaces in this doped polymer. This state can be taken as the case
with infinite wait time ( T = m) in the doped state. The cyclic
voltammograms of freshly-deposited PPy(DBS) in an aqueous
solution of 0.1 M DBS-Na+ are displayed in Figure 5 . Indeed
we see a very strong cathodic peak in the first cycle, with peak
potential even more negative than that for the PPy(DBS) after
being held in the doped state for a long time (Figure 4a, 7 =
10 000 s). Several cycles are needed for the voltammograms to
become overlapping. This break-in phenomenon is similar to
that observed in the redox polymer PTCNQ.S Wedid not observe
significant break-in in PPy(C1) where the first voltammogram of
the freshly-depositedPPy(C1) is similar to the subsequent ones
except for a little bit higher cathodic peak current. This absence
of break-in is, we think, because of the same reason that accounts
for the absence of hysteresis in the first voltammogram of PPy(Cl) after being held in the doped state.
During running the above cyclic voltammetry of PPy(DBS),
we have chosen an electrolyte, DBS-Na+, with a bulky anion to

-0.3

0.0

-0.6

-8

.9

E,V us AgIAgc1

Peak potential (V)

50

4.3

Figure 5. Cyclic voltammograms in an aqueous solution of 0.1 M DBSNa of PPy(DBS) freshly prepared. Scanning rate: 100 mV SKI. The
sequence of cycles is indicated in the figure.

ensure a sole cation transport for charge compensation during

the redox. In the case where we had used an electrolyte with a
small anion, there might have occurred the insertion of anions
into PPy(DBS) along with, or instead of, cation ejection for charge
compensation during its doping:
PPyo(DBS-Na+)

+X-, -e-

PPy+(DBS- X-Na)

(3)

-X-, +e-

where x- denotes the small anion.

3. Volume Changes in PFy(C1). In the following we will use

the bending beam method to study the volume changes during
the redox of PPy(C1) and check whether holding PPy(C1) in its
neutral state causes volume shrinking.
The PPy(C1) layer of a PPy(Cl)/PE bipolymer strip immersed
in an aqueous solution of 0.1 M Et4NCl can be repeatedlyundoped
at -0.8 V vs AglAgCl and doped at 0.3 V. Except for a slight
difference in the first undoping4oping cycle, the chronoamperometric curves overlap well in the following cycles. Figure 6a
shows one chronoamperometric curve during an undoping. The
simultaneous bending of the bipolymer strip has been observed
and displayed also in Figure 6a. The undoping current is initially
very large, decays rapidly, and turns to approach zero gradually
after about 15 s. The curvature of the strip responds synchronously, with an initial rapid decreaseand subsequent slow decrease.
The volume of the PPy(C1) layer shrinks in linear proportion to
the curvature change^.^ To give an idea of the order of magnitude,
taking the ratio of the moduli of elasticity of the PPy and PE
layers being 2.5, with the ratio of thickness of the PPy and PE
layers being 1/ 15 (the PPy layer is 10pm thick), we can calculate4
the volume change ratio in the PPy layer to be -0.2% after 1 min
of undoping. The undoping seems to be completed within 1 min,
but the slow volume shrinking continues beyond 10 h. This will
be further addressed later.
The chronoamperometric curve and corresponding curvature
change of the PPy(Cl)/PE strip during a doping are shown in
Figure 6b. The doping current is initially large and increases
further. It turns to decrease rapidly after reaching a peak at
about 8 s and approacheszero after 1 min of doping. This unusual
initial current increase may be accounted for by the moving
boundary concept:4 the doping of a neutral PPyO proceeds outward
from the inner surface of the polymer. As the doping progresses,
the frontier of the doped regions moves outward toward the

6038 The Journal of Physical Chemistry, Vol. 97, No. ;2, 1993
0

r!

-0.01

Pei and Inganas

0
h

-0.05

-0.1

m -0.15

-lor

"

"

""""'
8 0 1

" '

"

"

-0.2

-0.04
0

Time (h)

Time ( 8 )
,

]0.04

'

0.04~'"'"

'

"

'

'

'

"

"

4
0'01
0

Time ( 8 )
Figure 6. Chronoamperometric response and simultaneous curvature
change of a PPy(CI)/PE strip during undoping a t -0.80 V (a) and doping
at +0.30 V (b) in an aqueous solution of 0.1 M EkNCI. A few previous
undoping-doping cycles were performed to obtain overlapping chronoamperometric curves.

polymer-solution interface, and the diffusion distance of the

charge compensating ions which controls the rate of the doping
decreases, so that faster doping is expected. The curvature of the
bipolymer strip increases accordingly, with initial slow increase
followed by rapid increase and turning to gradual changing when
the doping current approaches zero. The volume of the PPy(C1)
layer on the strip swells in linear proportion to the curvature
changes. Taking the data given earlier, we calculate the volume
changes ratio to be +0.18% after 2 min of doping. The doping
is completed within 2 min.
These volume changes in PPy(C1) during the brief doingundoping can be readily interpreted as due to the insertion into
and ejection out of the polymer of the charge compensating anions
C1-. Doping induces C1- insertion into the polymer which is thus
dilated, while undoping induces the ejection of the C1- dopants
out of the polymer which thus contracts. However, the volume
change ratios during these doping-undoping (around 0.2%) are
too small when considering that the weight percent of the dopant
anions in the doped polymer is around 14%(assuming doped PPy
contains 0.3 chloride per pyrrole unit"), without considering the
solvation of the dopant anions. Thus it is likely that after a brief
undoping of PPy(Cl), there are many free spaces left behind
after the original dopants have been ejected. These free spaces
will gradually be filled in by the adjacent neutral polymer chains,
and the polymer matrix contracts.
Figure 7a shows the curvature change of the PPy(Cl)/PEstrip
in the long run after undoping at -0.8 V, the potential being
maintained to avoid influenceof the solution species on the neutral
polymer. Theinitial rapiddecreaseof thecurvatureoccursduring
the undoping of PPy(C1). After that, the curvature turns to
decrease slowly, reaching from -0.028 cm-1 at the beginning to
-0.1 57 cm-I after 10h, and continues further. Thevolume change
ratio (in linear proportion with the curvature change) is -0.9%
in 10 h after the initial undoping. This slow volume shrinking
is related to the slow relaxation in neutral PPy(C1) being held at
the negative potential limit as discussed in section 1, indicative
of a mechanical origin for the relaxation. Though the kinetics
of the volume shrinking during the relaxation does not follow a
simple logarithmic law, which implies a complex relaxation

Time (h)
Figure 7. Curvature change of a PPy(CI)/PE strip maintained at -0.80
V (a) and +0.30 V (b) in an aqueous solution of 0.1 M Et4NCI (the
extension of Figure 6).

probably including more than one step, the mechanical origin is,
we think, still the major cause of the slow relaxation in neutral
PPy(C1).
The long time curvature change of the PPy(Cl)/PE strip after
doping at +0.3 V of doping with the potential being maintained
is displayed in Figure 7b. No significant cPange occurs in 10 h,
in accord with the observation in section 1 of little hysteresis
effect by holding PPy(C1) at the doped state. The volume
shrinking ratio in 10 h after the initial doping is only +0.06%.
PPy's doped with other small anions, like perchlorate and
nitrate, have also been investigated and behave in a similar way
to PPy(C1).
4. VolumeChangesinPPy(DBS).Accordingto theargument
above, those PPy doped with bulky anions shall behave in the
opposite way: they swell upon undoping due to cation insertion
from the electrolytesolutionfor charge compensation and contract
upon doping due to the ejection of these cations. As an example,
PPy(DBS) will be investigated.
A PPy(DBS)/PE bipolymer strip immersed in an aqueous
solution of 0.1 M DBS-Na+ was repeatedly undoped at -0.85 V
vs Ag/AgCl and doped at 0.3 V. After several undoping-doping
cycles, the chronoamperometric curves in the following cycles
turn to overlap well. Figure 8a shows one overlapping chronoamperometriccurve during undoping. The simultaneousbending
behavior of the bipolymer strip is also displayed in Figure 8a. The
undoping current is initially very large, decays rapidly, and turns
to approach zero gradually after about 13 s. The curvature of
the strip responds synchronously, with an initial rapid increase
and subsequent slow increase. To give an idea of the order of
magnitude of the volume change of the PPy(DBS) layer, taking
the ratio of the moduli of elasticity of the PPy and PE layers
being 2.5, with the ratio of thickness of the PPy and PE layers
being 1/10 (the PPy(DBS) layer is 15 pm thick), we calculate
the volume change ratio in the PPy layer to be + 1.7%after a half
minute of undoping. The undoping seems to be completed within
a half-minute. Note that PPy(DBS) swells whereas PPy(C1)
shrinks during an undoping, because, in PPy(DBS), the dopant
anions are bulky and insertion into the polymer of cations from
the solution (Na+)dominatesthe diffusion of charge compensating
ions (scheme 2).

Electroshrinking and Slow Relaxation in Polypyrrole

0

'

"

'

'

'

'

'

0.24
~

The Journal of Physical Chemistry, Vol. 97, No. 22, 1993 6039
0

Q
..
f

-0.05

-0.1

-OI5

.
.

20

40

Time (s)

'

"

"

' ' I , ,

-0.08

-0.12

4
V

40

20

Time

"

",

, ' , ,

,,,, ,, , , ,

.,

-.

-0.16

10

Time (h)
0.2 7 "
' ,
' ,
'
' ,

(8)

Figure 8. Chronoamperometric response and simultaneous curvature

change of a PPy(DBS)/PE strip during undoping at 4.85 V (a) and
doping at +0.30V (b) in an aqueous solution of 0.1 M DBS-Na. Several
previous undoping-doping cycles were performed to obtain overlapping
chronoamperometric curves.

The chronoamperometricresponse and correspondingcurvature

change of the PPy((DBS)/PE strip during the doping at +0.3 V
are shown in Figure 8b. The doping current is initially very
large, decays rapidly, and approaches zero after 10 s of doping.
The initial very large current in contrast to the case with neutral
PPy(C1) is because a different mode of diffusion of the charge
compensating ions is concerned: in the doping of neutral PPy(DBS) (actually PPyO(DBS-Na+)),transport of sodium cations
out ofthe polymer matrixinto thesolutiondominates thediffusion.
The concentration difference of the cations in the polymer and
the solution is large, around 2.5 M (assuming doped PPy contains
0.3 DBS- per pyrrole unit and its density is 1.5 g/cm3 19,a high
diffusion rate is thus obtained (the solution concentration is 0.1
M). The curvature of the PPy(DBS)/PE strip decreases, first
rapidly and then gradually. The volume change ratio is -1.5%
after a half-minute of doping. The doping is completed within
this period.
These volume change ratios in PPy(DBS) during the dopingundoping are still smaller than the weight percent of the
compensating sodium cations in the undoped polymer, around
4% (solvation of ions has not been considered). Thus it is
reasonable that after a brief doping of the neutral PPyO(DBS-Na+), there are many free spaces left behind after the
cations have been ejected out. These free spaces shall be filled
in gradually by the adjacent doped PPy chains or DBS- anions,
and the polymer matrix contracts.
Figure 9a shows the curvature change of the PPy(DBS)/PE
strip in the long run after doping, with +0.3 V of doping potential
being maintained to avoid influence of the solution species on the
doped polymer. The initial rapid decreaseof the curvature occurs
during the doping of the neutral PPyo(DBS-Na+). After that,
the curvature turns to decrease slowly, reaching from -0.130
cm-' at the beginning to -0.166 cm-l after 2.5 h, as indicated in
the inset of Figure 9a. The curvature decrease become very
gradual beyond 3 h. The volume change ratio is 0.4% in 2.5 h
as the curvature decreases by -0.036 cm-I. This slow volume
shrinking should also be related to a slow relaxation in the doped
PPy(DBS) and is related to the hysteresis effect obtained by
holding PPy(DBS) in the doped state, as observed in section 2.

"
.
.
,
.
-0.1

Time (h)
Figure 9. Curvature change of a PPy(DBS)/PE strip maintained at
4.85 V (a) and +0.30 V (b) in an aqueous solution of 0.1 M DBS-Na
(the extension of Figure 8).

A comparison of both the extent and kinetics of the slow

relaxations in doped PPy(DBS) and neutral PPy(C1) is interesting.
Both relaxations involve volume shrinkingof the polymer matrices.
It is noticed that the relaxation in neutral PPy(C1) is more
significant than that in doped PPy(DBS), reflected both in the
shift of peak potentials and the increase of peak currents of the
subsequentcyclic voltammogramsof these polymers after a certain
wait time (Figures l a and 4a), and in the volume shrinking ratios
in the polymers (-0.9% and -0.45%, respectively, in 10 h). In
addition, the relaxation kinetics in neutral PPy(C1) seems to be
slower than that in doped PPy(DBS), indicated by a comparison
of Figure 7a and Figure 9a. It has been mentioned that the
weight percent of chloride in doped PPy(C1) is around 14%,while
brief undoping induces only a volume shrinking by -0.2%.
Therefore the free spaces found in the newly undoped PPy(C1)
must constitutea highvolume ratioof the polymer matrix, leading
to a significant relaxation afterward. In neutral PPyo(DBS-Na+),
the cations constitute a weight percent of only 4% of the polymer
matrix. After a brief doping, the polymer contracts by lS%,so
that the volume ratio of free space found in the freshly doped
PPy(DBS) is not high, and the subsequent relaxation is not
significant. Doped PPy(DBS) contains DBS- anions which have
long flexible alkyl chain tails and should be more mobile than the
stiff conjugated PPy main chains. Neutral PPy(C1) contains
only the stiff conjugated PPy main chains, and the filling of these
polymer chains in the nearby voids or channels for relaxation is
therefore slower than in doped PPy(DBS) where the relatively
mobile DBS- can fill in the voids or channels more easily.
It has been demonstrated that bulky anions like DBS- are
trapped in PPy. However, it has been shown that they are not
trapped there permanently. In the neutral PPP(DBS-Na+), the
DBS-Na+ salt or ion pairs can diffuse out of the polymer matrix
albeit very slowly. This salt draining is reflected in the large
volume shrinking of the polymer matrix after being undoped as
depicted in Figure 9b. The curvature (and the volume as well)
of the PPy(DBS)/PE strip after undoping decreases linearly with
time. We have seen in Figure 3b that the capacity of PPy(DBS)
is lost almost linearly with the wait time holding the polymer at
the neutral state. Both the contraction and capacity loss are
related to the salt draining of DBS-Na+out of the polymer matrix

6040 The Journal of Physical Chemistry, Vol. 97, No. 22, 1993

TABLE I: Hysteresis of the First Cyclic Voltammograms of

PPy Thin Films after Being Held at Their Neutral or Doped
State and Bending Behavior of PPy/PE Bipolymer Strips
during Doping-Undoping, Both in Aqueous Solutions of 0.1 M
Et&l (for PPy(cI)) and of 0.1 M DBS-Na (for PPy(DBS))
exoerimental conditions
held at -0.85 V for 100 s

parameters"
shift of Ep,)(mV)
increase of Ip,,(%)
held at -0.85 V for 10 000 s
shift of E,,., (mV)
increase of Ip,,(9%)
held at +0.3 V for 100 s
shift of E , (mV)
increase of I, (5%)
held at +0.3 V for 10 000 s
shift of E , (mV)
increase of I, (%)
brief undopinghof a PPy/PE
time to complete (s)
curvature change (cm- I )
strip
volume change (%)
held at the undoping potential curvature change (cm ' )
for 10 h after brief undoping volume change (%)
of the PPy/PE strip
time to complete (s)
brief doping at +0.3 V of a
curvature change (cm-I)
PPy/PE strip
volume change (%)
held at the doping potential for curvature change (cm- I )
10 h after brief doping of the volume change (%)
PPy/PE strip

PPy(CI)
+50
+28
+290
+181
-20
+12
-40
+4
=60
-0.028
4.20
-0.129
-0.92

PPy(DBS)
<+lo
+10
+30
-5
-40
+46
-7 5
+94
=30
+0.145
+1.71
-0.220
-2.59

=120 =30
+0.025 -0.127
+0.18 -1.50
+0.008 -0.038
+0.06 4 . 4 5

I? Epa,
I,,, E,, and IF are the peak potentials and currents of the first
anodic and cathodic peaks on the cyclic voltammograms, respectively.
At a constant potential of -0.80 V for PPy(CI)/PE and of -0.85 V for
PPy( DBS)/PE.

into the solution in contact, though the linearity cannot be fitted

into a simple diffusion model.
5. Summary of the Results and Discussion. The experimental
results presented above are summarized in Table I. It clearly
shows the hysteresis of the subsequentfirst cyclic voltammograms
of PPy(C1) after being held at the neutral state and of PPy(DBS)
after being held at the doped state. The latter provides an example
of hysteresisdue to holding an electroactivepolymer at the doped
state. No or little hysteresis has been observed after holding
PPy(C1) at the doped state or PPy(DBS) at the neutral state.
This oppositebehavior in the two P w s indicates that the previous
interpretations for hysteresis as due to solvation/de-solvation (Le.,
doped and neutral PPy are solvated to different extents), due to
a relaxation of the polymer chains from a planar to a twisted
form or vice versa, or due to evacuation of residual electronic
charges when holding the polymer in the neutral state cannot be
applied here. These interpretations would imply that hysteresis
occurs after holding PPy(DBS) at the neutral state, not at the
doped state. The interpretation of hysteresis as due to ejection
of ion pairs from the polymer matrix cannot be applied either
because of the same reason, and, furthermore, in the case of
PPy(tosylate), the slow relaxation continues well after the salt
draining (ejection of ion pairs) has been ~ompleted.~
In PPy(DBS) held at the neutral state, this salt draining results in the
loss of redox capacity of the polymer in the solutionof DBS-Na+,
but no hysteresis appears on the subsequent cyclicvoltammograms.
The volume of PPy(C1) shrinks whereas that of PPy(DBS)
swells during undoping. Re-doping reverses thesevolume changes.
The underlying nature is depicted in schemes 1 and 2. Both
PPy(C1) being held at the neutral state and PPy(DBS) being
held at the doped state contract after the charge compensating
ions have been ejected from the polymer matrix. The volume
shrinkage of the polymer matrix during the ejection is lower than
the volume these ions occupied in the matrix before the redox.
Thus free spaces are left in the matrix after the ejection, which
can be voids or open channels, depending on how these ions are
distributed in the matrix.15 The slow relaxation in neutral PPy(Cl) and doped PPy(DBS) during the holding can subsequently
be attributed to a mechanical origin: PPy chains (and the alkyl
chains of DBS- in PPy(DBS)) move toward the adjacent free
spaces and fill them. Re-doping of PPy(C1) after a long period

Pei and Inganls

of holding at the neutral state and undoping of PPy(DBS) after
a long period of holding at the doped state need to re-open free
spaces in the polymer matrices to accommodate the inserting
ions. This causes hysteresis.

Conclusion
Hysteresis of the first peak on the subsequent cyclic voltammogram has been observed after holding PPy(C1) at the neutral
state and holding PPy(DBS) at the doped state. Holding PPy(Cl) at the doped state or PPy(DBS) at the undoped state does
not induce this effect. Slow volume shrinking occurs by -0.92%
in PPy(C1) held at the undoped state for 10 h and by -0.45% in
PPy(DBS) held at the doped state for 10 h, indicated by the
bending of the bipolymer strips of PPy/PE. The underlying nature
is slow relaxation of the polymer matrix: the free spaces left in
the matrix after the undoping of PPy(C1) due to the ejection of
dopant anions, or after the doping of PPy(DBS) due to ejection
of cations from the matrix, are gradually filled in by the adjacent
PPy chain (in doped PPy(DBS) also by the nearby alkyl chains
of DBS-). Scanningforce microscopy focusing on the local change
of micromorphology during the slow relaxation is to be carried
out to observe free spaces being closed up.
The electrochemicallyinduced volume shrinking and swelling
in PPy, electroshrinking and electroswelling,by around 1%, are
the primary interest in the development of these polymers as
smart materials.16

References and Notes

(1) (a) Skotheim, T. A., Ed. Handbook of Conducting Polymers;
Dekker: New York, 1986; Vols. 1 and 2. (b) Diaz, A. F.; Robinson, J. F.;
Mark, H. B., Jr. Ado. Polym. Sci. 1988,84,113. (c) Evans, G. P. In Aduances
in Electrochemical Science and Engineering, Gerischer, H., Tobias, C. W.,
Eds.; Wiley: New York, 1991; Vol. 1. (d) Pei, Q.; Qian, R . Electrochim.
Acta 1992, 37, 1075. (e) International Conferences on Synthetic Metals,
Santa Fe, 1988, Tubingen, 1990, and Gbteborg, 1992; for the proceedings,
see: Synth. Met. 1988, 27-29; 1991, 4143.
(2) (a) Nigrey, P. J.; MacInnes, D.; Nairns, D. P., MacDiarmid, A. G.;
Heeger, A. J. J. Electrochem. SOC.1981,128, 1651. (b) Shacklette, L. W.;
Maxfield, M.; Gould, S.; Wolf, J. F.; Jow, T. R.; Baughmann, R. H. Synth.
Met. 1987, 18, 611.
(3) (a) Garnier, F.;Tourillon,G.; Gazard, M.; Dubois, J. C. J.Electroanal.
Chem. 1983, 148, 299. (b) Kobayashi, T.; Yoneyama, H.; Tamura, H. J .
Electroanal. Chem. 1984, 161, 419. (c) LaCroix, J. C.; Kanazawa, K. K.;
Diaz, A. F. J. Electrochem. Soc. 1989, 136, 1308.
(4) Pei, Q.; Inganls, 0. J . Phys. Chem. 1992, 96, 10507.
(5) (a) Kaufman, F. B.; Schroeder, A. H.; Engler, E. M.; Kramer, S. R.;
Chambers, J. Q. J . Am. Chem. SOC.1980,102,483. (b) Inzelt, G.; Day, R.
W.; Kinstle, J. F.; Chambers, J. Q. J . Phys. Chem. 1983, 87, 4592. (c)
Dautartas, M. F.; Bowden, E. F.; Evans, J. F. J . Electroanal. Chem. 1987,
219,71. (d) Varineau, P. T.; Buttry, D. A. J . Phys. Chem. 1987,91, 1292.
(e) Karimi, H.; Chambers, J. Q. J . Electroanal. Chem. 1987, 217, 313. (f)
Inzelt, G. Electrochim. Acta 1989, 34, 83.
(6) (a) Tourillon, G.; Garnier, F. J. Electroanal. Chem. 1984,161, 51.
(b) Heinze, J.; Bilger, R.; Meerholz, K. Eer. Bunsen-Ges. Phys. Chem. 1988,
92, 1266. (c) Baudoin, J. L.; Chao, F. J. Chim. Phys. (Paris) 1989,86, 181.
(d) Villeret, B.; Nechtschein, M. Phys. Rev. Lett. 1989,63, 1285. (e) Odin,
C.;Nechtschein, M.Phys.Reu.Lett.1991,67,1114. (f)Odin,C.;Nechtschein,
M. Synth. Met. 1991, 44, 177. (g) Odin, C.; Nechtschein, M. Synth. Met.
1992,47,329. (h) Goldenberg, L.M.;Aeiyach, S.; Lacaze, P. C. Synrh. Mer.,
in press.
(7) (a) Chiang, C. K.; Druy, M. A.; Gau, S. C.; Heeger, A. J.; Louis,
E. J.; MacDiarmid, A. G.; Park, Y. W.; Shirakawa, H. J . Am. Chem. SOC.
1978,100, 1013. (b) Bredas, J. L.;Street, G. B. Ace. Chem. Res. 1985,18,
308. (c) Penner, R. M.; Martin, C. R. J . Phys. Chem. 1989, 93, 984. (d)
Mao, H.; Pickup, P. G. J. Phys. Chem. 1989, 93, 6480.
(8) Odin, C.; Nechtschein, M. ICSM'92. Synth. Met., in press.
(9) (a) Kanazawa, K. K.; Diaz, A. F., Geiss; R. F.; Gill, W. D.; Kwak,
J. F.; Logan, J. A.; Rabolt, J. F.;Street, G. B.J . Chem. SOC.,Chem. Commun.
1979,854. (b) Bull, R. A.; Fan, F. F.; Bard, A. J. J . Electrochem. SOC.1982,
129, 1009. (c) Feldberg, S . W. J . Chem. Soc. 1984, 106, 4671. (d) Penner,
R. M.; Van Dyke, L. S.; Martin, C. R. J . Phys. Chem. 1988, 92, 5274. (e)
Tezuka, Y.; Aoki, K. J . Electroanal. Chem. 1989,273, 161. (f) Krishna, V.;
Ho, Y. H.; Basak, S.; Rajeshwar, K. J . Am. Chem. SOC.1991,113,3325. (8)
Cai, Z.; Martin, C. R. Synlh. Met. 1992, 46, 165. (h) Doblhofer, K. J .
Electroanal. Chem. 1992, 331, 1015.

Electroshrinking and Slow Relaxation in Polypyrrole

The Journal of Physical Chemistry, Vol. 97, No. 22, 1993 6041

(10) (a) Genies, E. M.; Pernault, J.-M. J. Electroanal. Chem. 1985,191,

1 1 1. (b) HBrhold, H.-H.; Melbig, M.; Raabe., D.; Opfermann, J.; Scherf, U.;
Stockmann, R.; Weiss, D. Z . Chem. 1987,27,126. (c)Tanguy, J.; Mermilliod,
N.; Hoclet, M. Synth. Met. 1987, 18, 7.
(1 1) (a) Chao, F.; Baudoin, J. L.; Costa, M.; Lang, P. Makromol. Chem.
Makromol. Symp. 1987,8, 173. (b) Zhou, Q.-X.; Kolaskle, C. J.; Miller, L.
L. J. Electroanal. Chem. 1987, 223, 283.
(12) (a) Fan, F. F.; Bard, A. J. J. Electrochem.Soc. 1986,133,301. (b)
Iyoda, T.; Ohtani, A.; Shimidzu, T.; Honda, K. Chem. Lett. 1986, 687. (c)
Shimidzu, T.; Ohtani, A.; Iyoda, T.; Honda, K. J. Electroanal. Chem. 1987,
224, 123. (d) Wernet, W.; Wegner, G. Makromol. Chem. 1987,188, 1465.

1990,35,1145. (f) Li, F.; Albery, W. J. J. Chem. Soc., Faraday Trans. 1991,
87,2949. (g) Panero, S.; Prosperi, P.;Scrosati, B. Electrochim. Acta 1992,
37, 419.
(1 3) Pei, Q.; Qian, R. C-MRS-International'90 Proceedings, 1990;
Elsevier: Amsterdan, 1991; Vol. 3, p 195.
(14) Qian, R.; Qiu, J. Polym. J. (Tokyo) 1987, 19, 157.
(15) (a) Yang, R.; Evans, D. F.;Christensen, L.; Hendrickson, W. A. J.
Phys. Chem. 1990, 94,6117. (b) Yang, R.; Naoi, K.; Evans, D. F.; Smyrl,
W. H.; Hendrickson, W. A. Langmuir 1991, 7, 556.
(16) (a) Pei, Q.; Inganis,O. Ado. Muter. 1992,4,277. (b) Pei, Q.; Inganas,

(e) De Paoli, M. A,; Panero, S.; Prosperi, P.; Scrosati, B. Electrochim. Acta

0. Synrh. Met., in press.