Electrochemical Investigations of Polyethylene Soggy Sand Electrolye

www.afm-journal.de

FULL P

APER

www.MaterialsViews.com

Anna Jarosik , Christian Pfaffenhuber , Armin Bunde , and Joachim Maier *

Electrochemical Investigations of Polyethylene Glycol-Based “Soggy Sand” Electrolytes – From the Local Mechanism to the Overall Conduction

Using the example of SiO 2 dispersions in LiClO 4 /polyethylene glycol electro-lytes, the conduction mechanism of “soggy sand” electrolytes is discussed. The study is essentially based on zeta potential, impedance and transfer-ence number measurements as well as on modeling. All the results can be explained by anion adsorption by the oxide particles and increased concen-tration of free Li + in the double layer. The initially colloidal dispersion quickly forms fractal networks by cluster–cluster aggregation. Once they percolate, an interfacially dominated Li + conductance is observed. The subsequent coarsening of the network is self-decelerating leading to a steady state con-ductivity that is, for low volume fractions, enhanced compared to SiO 2 free electrolytes. At higher values, blocking and inhomogeneity effects (e.g., salt trapping) lead to decreased values of the overall conductivity.

1. Introduction

On the basis of the concept of “heterogeneous doping”, [ 1 , 2 ] var-ious designed solid–solid composites have been prepared and investigated in order to achieve synergistic conductivity prop-erties. Examples are solid–solid composites of Ag halides with Al 2 O 3 in which the cation adsorbing power of the insulating oxide leads to increased cation vacancy concentrations along the interfaces. [ 1 , 3–5 ] The anion analogue has been achieved with fl uoride ion conducting CaF 2 or PbF 2 heterogeneously doped with the rather acidic SiO 2 adsorbing F − ions, thus generating F − vacancies in the space charge zones adjacent to the oxide particles. [ 6 ] Liquid–solid composites have also been investigated. Addition of the surface-acidic oxides TiO 2 or sulfated ZrO 2 to liquid imidazole resulted in adsorption of the imidazolate anion accompanied by generation of free protons in the space charge zones. [ 7 ] In all these cases, the undissociated ground state (regular ions or undissociated molecule) is broken up. Intriguing from an application point of view are salt containing polymers [ 8 , 9 ] or even liquid solvents (of typically low dielectric constants) to which electrically insulating oxides are admixed,

© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, WeinheimAdv. Funct. Mater. 2011, 21, 3961–3966

DOI: 10.1002/adfm.201100351

Dr. A. Jarosik , C. Pfaffenhuber , Prof. J. Maier Max Planck Institute for Solid State Research Heisenbergstr. 1, 70569 Stuttgart, Germany E-mail: [email protected] Prof. A. Bunde Justus-Liebig-Universität Gießen Heinrich-Buff-Ring 16, 35392 Giessen, Germany

frequently achieving enhanced conductivi-ties. [ 10–12 ] If a similar mechanism applies, the ground state that must be broken is the ion pair. While in the case of poly-mers, mobility issues may and do play a role, in the case of the soggy sand electro-lytes, it is obvious that the concentration effect is crucial: anions can be adsorbed on the SiO 2 surfaces resulting in a higher concentration of free Li + in the boundary zones (see Figure 1 ). While this local mechanism in the liquid–solid compos-ites is comparable to the solid–solid case, the percolation behavior is very different.

Network formation may be sensitive to small variations of the parameters. Unlike the solid–solid composites where the two-

phase morphology is frozen, the SiO 2 networks in the liquids are not spatially fi xed. They will also coarsen and may, if not kinetically prevented, eventually sediment.

These issues explain many of the diffi culties in exactly repro-ducing the conductivity data. Even if the size and volume frac-tion of the silica particles stay the same, severe variations of the conductivity results can occur. This is typically due to a lack of control of the surface chemistry of the oxide but also due to the fl uidity of the system. To give an example: recent experiments performed by us on e.g., THF-containing electrolytes showed a maximum at lower volume fraction, corresponding to an ear-lier onset of percolation but also earlier decrease in conductivity than in previous experiments, [ 12 ] even though the silica particles are nominally very similar. As the initial values extrapolated to higher volume fractions are in reasonable agreement with the earlier data on the same system, it can be concluded that it is the network morphology that is different. A similar lack of exact reproducibility occurs even in the Monte Carlo simulations. It is clear that diffi culties in reproducibility stem not only from the detailed surface chemistry of the oxide but also from shape, size, volume fraction, and since network formation is a kinetic phenomenon, from pretreatment (e.g., homogeneity of par-ticle wetting). In recent experiments on polyethylene glycol for which we reach a certain control over the network, the results are much more reproducible even though the scatter from experiment to experiment is still perceptible. [ 13 ]

In this study we chose liquid polyethylene glycol as solvent, LiClO 4 as salt and SiO 2 as additive (particle size: 7 nm, fumed − SiO 2 , 10 nm − SiO 2 ). This system offers the possibility to de-convolute transient effects from stable effects. [ 13 ] Combination of zeta potential (surface potential), dc polarization (transport

3961wileyonlinelibrary.com

FULL

PAPER

3962

www.afm-journal.dewww.MaterialsViews.com

(pair)

+Li

µ

µ

µ

~

~

~

distance from particle

0

Li+ (free)

elec

troc

hem

ical

pot

enti

al

0

Li+

Figure 1 . Electrochemical potential ( ̃: ) of Li + in the solvent together with its standard values ( ̃: 0 ) in the ion pair ground state and as a free ion in solution. The distance between ̃: and :̃ 0 are inverse logarithmic meas-ures of the respective concentrations.

number) and ac conductivity experiments allows us to draw conclusions on local and long range transport behavior. Rheo-logical and Monte Carlo investigations accompanying electroki-netic investigations will be only discussed in as much as they are of direct relevance for the mechanistic interpretation. They are to be published in greater detail elsewhere. [ 13 , 14 ]

2. Experimental Section

The sample preparation and the utilized materials are described elsewhere. [ 13 ] Here it suffi ces to state that lithium perchlorate (LiClO 4 , Sigma Aldrich) was dissolved in poly(ethylene glycol)dimethyl ether (PEG-150, Fluka and PEG-350, Sigma Aldrich) and the composite electrolytes were prepared by dispersing SiO 2 (10 nm and 7 nm fumed, Sigma Aldrich) with different volume fractions ( n ) in the LiClO 4 /PEG solution using a Vortex shaking device.

An acoustic and electroacoustic DT-1200 spectrometer (Dispersion Technology, Inc., Quantachrome) was used to determinate the Zeta potential ( ζ ). The electroacoustic sensor

© 2011 WILEY-VCH Verlag Gwileyonlinelibrary.com

volume fraction ϕ

t0

δσ

σ

t0 >> τ

a)

Figure 2 . a) Schematic plot of ionic conductivity versus SiO 2 volume fractiothat the SiO 2 network formation happens fast between t = 0 and t = t 0 . Posisients of the ionic conductivity of the soggy sand electrolytes after preparat

measures magnitude and phase of the colloid vibration current (CVI) at 3 MHz. The measurements were performed at room temperature in ambient atmosphere.

The ionic conductivity of the electrolytes was measured using ac -impedance spectroscopy in the frequency range from 10 6 Hz to 10 − 1 Hz with a voltage amplitude of 0.1 V (Solartron 1260 frequency analyser). The samples were placed between two parallel gold electrodes of a custom-made cell. This cell was immersed in a stainless steel container (fl ushed by dry argon gas) and placed in RC6CP Lauda thermostat.

The Li + ion transport number of the composite electrolytes was determined using a special variant of Hebb-Wagner-Yokota polarization technique, [ 15 , 16 ] according to the approach by Evans et al. [ 17 ] In this method a dc experiment is performed using the cell: Li | Li + X − composite electrolyte | Li.

In the steady state the anions are blocked at the lithium metal electrodes owing to the non-reversibility of these electrodes for the anions. The blocking occurs even more effectively as a con-sequence of a selective Li + -conductive passivation layer formed at the lithium/electrolyte interface.

The blocking of the anions leads to the fact that in the steady state only the cationic current fl ows ( I (t = ∞) = I+ (t = ∞) ). The comparison with the initial state yields the cationic trans-port number t+ = F+ /(F+ + F−) provided correction with respect to time dependent overvoltages is done as described previously. [ 17 ] The measurements were performed at a constant potential of 10 mV. The steady state was typically obtained after several hours I∞

∧= I (( t = 10 h)) . This evaluation ignores the mobility of ion pairs that are in

equilibrium with the free ions which would decrease the polar-ization even if the Li + conductivities are negligible compared with the anion conductivity. [ 18 , 19 ]

3. Results and Discussion

3.1. Time and volume fraction dependence of ac conductivity: long range transport and percolation along the oxide network

A typical behavior of composite liquid polymer electrolytes concerning the dependence of the conductivity on the SiO 2 volume fraction is sketched in Figure 2 a. While the upper curve

mbH & Co. KGaA, Weinheim Adv. Funct. Mater. 2011, 21, 3961–3966

0 200 400 600 800 1000

1.7x10-3

1.8x10-3

1.9x10-3

2.0x10-3

2.1x10-3

2.2x10-3

σσ σσ m /

S.cm

-1

time, t /min

b)

n for freshly prepared material ( ∼ t 0 ) and for the stationary state ( t ∞ ). Note tion and height of the maximum is strongly parameter dependent. b) Tran-ion. System: 1 M LiClO 4 /PEG-150: SiO 2 (7 nm-fumed), ϕ = 0.02

FULL P

APER


Figure 3 . Fractal networks formed from the starting confi guration (left column, t = 0) via hit-and-stick (center column, fi nal situation of the hit-and-stick process t = t 0 ). The networks are then allowed to coarsen (right column, fi nal situation of the coarsening process t = t ∞ ) as detailed previously. [ 14 ]

corresponds to freshly prepared electrolytes, the values of the lower curve correspond to steady state values obtained after a waiting time of several days. We recognize a conductivity max-imum at low volume fractions. After the maximum, the con-ductivity drops, often quite sharply to values that can even be below Fm (n = 0) ≡ F∞ ( ∞ refers to the bulk).

A typical time behavior between preparation and steady state is sketched in Figure 2 b. The experimental results conform to the extensive modeling discussed previously [ 14 ] (see Figure 3 ) and indicate the following picture.

The SiO 2 particles (negatively charged because of anion adsorp-tion) are initially separated by double layer repulsion. If the repul-sive barrier is overcome by thermal events, the SiO 2 particles stick together via van der Waals bonds or even covalent bonds as a result of cluster–cluster aggregation. In almost all cases of relevance for our context, the network formation is too quick to be observed experimentally. Only recently, we succeeded to follow the kinetics of network formation in the case of extremely small volume frac-tions. This will be reported separately. [ 20 ] For high enough ϕ a fractal percolation cluster is formed, [ 21 ] and from this ϕ -value enhanced conductivities via surface contribution have the chance to be observed in a macroscopic experiment. The percolation threshold appears at ϕ -values which are the lower the lower the particle size ( R ). Note that the mean particle distance scales as R/ ϕ 1/3 and, for nanosized particles, leads to a low mean distance even at very low n . It is also simple to prove that the percolation threshold ( nc ) for cluster–cluster aggregation tends to zero for a vanishing par-ticles size, as the volume fraction of a single particle increases faster with particle size than the number of particles in the cluster decreases. [ 22 ] Network formation is also confi rmed by confocal fl uorescent microscopy, [ 20 ] and rheological measurements. [ 13 ] At even higher volume fractions, the conductivity is depressed. This depression can lead to values smaller than those of the SiO 2 free electrolyte. This may be explained by partial collapse of the net-work and/or blocking effects by insuffi ciently conducting network parts (severe agglomeration, insuffi cient salt concentration and/or trapping of salt inside inactive cluster). [ 14 ]

While at concentrations higher than nc the time dependen-cies are diffi cult to interpret quantitatively, the time dependen-cies in the regime of comparatively low volume fractions are revealing in terms of network behavior.

Figure 2 b shows a typical time dependent conductivity measurement. Immediately after the preparation, we fi nd a

© 2011 WILEY-VCH Verlag GAdv. Funct. Mater. 2011, 21, 3961–3966

substantial σ -enhancement (compared to the fi ller free electro-lyte), indicating rapid network formation. The decay hereafter can be consistently interpreted in terms of coarsening. This coarsening leads to a partial loss of conductivity and hence of percolation effi ciency. Obviously the (apparent) steady state conductivity values are still enhanced, pointing towards self-deceleration of the process. The fact that the time constants ( τ ) but not the conductivity variations are fairly insensitive towards volume fraction suggests that only weakly bound SiO 2 particles (e.g., with only one or two neighbors) quickly rear-range, while denser confi gurations (e.g., SiO 2 with three neigh-bors) are locked-in and do not coarsen in the time window of measurement. [ 14 ]

The fact that the time constants are also not very dependent on the polymerization degree (i.e., viscosity) of PEG shows that τ is determined by inner network rearrangement characterized by SiO 2 surface diffusion rather than by SiO 2 diffusion in the solvent that is clearly too quick to be observed. This justifi es the decoupled mod-eling in Ref. 14. More detailed modeling should not only allow for coupling but should also allow for chain segment mobility (perhaps the reason for the very sharp decay), internal collapse and ageing effects (another reason for slowing down of the coarsening effect).

3.2. Zeta Potential and Transference Number: The Local Mechanism

While the total conductivity gives information on the long range transport, measurements of zeta potential and of the transfer-ence number give additional insight into the local mechanism. As far as zeta potential measurements of the liquid–solid composites are concerned, the information is only of qualita-tive nature. This is due to morphological and transport inho-mogeneities that are not adequately included in the evaluation. Yet, the very worthwhile information concerning the sign of the surface charge and its order of magnitude in dependence on salt and oxide concentration and the sign of the variation with salt and oxide concentrations is reliable. With the excep-tion of LiCl as salt, we always found a negative space charge potential ( Figure 4 ) which corroborates the mechanistic picture (Figure 1 ) that the SiO 2 particles preferentially adsorb the anions owing to the acidic surface character (the special role of LiCl leading to an anion adsorption is well-known in literature). [ 23 , 24 ]

mbH & Co. KGaA, Weinheim 3963wileyonlinelibrary.com

FULL

PAPER

3964


Figure 4 . a) Effective zeta potential ( .e f f ) as a function of LiClO 4 concentration in LiClO 4 /PEG-150 containing 1.5 wt.% SiO 2 (7 nm fumed) or SiO 2 (10 nm). Salt concentrations range from 0.01 M to 1 M . Measurements were carried out at room temperature. b) Effective zeta potential ( .e f f ) as a function of weight fraction of SiO 2 (7 nm fumed) or SiO 2 (10 nm) in LiClO 4 /PEG-150. Salt concentrations range from 0.01 M to 1 M . Measure-ments were carried out at room temperature.

1 2 3 4 5 6 7 80

-5

-10

-15

-20

-25

-30

-35

-40

-45

ζζ ζζ eff-p

oten

tial

/ m

V

1M LiClO4/PEG-150 : SiO2(10nm)

1M LiClO4/PEG-150 : SiO2(7nm-fumed)

0.1M LiClO4/PEG-150 : SiO2(10nm)

0.1M LiClO4/PEG-150 : SiO2(7nm-fumed)

0.01M LiClO4/PEG-150 : SiO2(10nm)

0.01M LiClO4/PEG-150 : SiO2(7nm-fumed)

weight percentage of SiO2

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.10

-5

-10

-15

-20

-25

-30

-35

-40

-45ζζ ζζ ef

f-pot

enti

al /

mV

LiClO4 concentration in solvent

SiO2-7nm-fumed

SiO2-10nm

LiClO4/PEG-150 : SiO

2 (1.5 wt.%)

The fact that the zeta potential decreases with increasing ϕ (Figure 4 b) and decreasing Li-salt concentration (Figure 4 a) is in agreement with the expected lowered surface charge density.

A local conductivity increase as a consequence of anion adsorp-tion conforms with the concept of heterogeneous doping [ 1 ] and its application to soggy sand electrolytes as set out previously. [ 10 ] Given the fact that the dielectric constant of poly(ethylene glycol)dimethyl ether ( ε r = 23.9 at T = 25 ° C) and poly(ethylene glycol)methyl ether ( ε r = 25.2 at T = 25 ° C) is too low to dissociate all ion pairs, we expect an increased cation conductivity and a depressed anion con-ductivity in the vicinity of the oxide particles. As these space charge zones are essentially parallel to the electrolyte bulk, the space charge zones contribute to or even dominate the overall cation con-ductivity, while the depression of anion concentration is irrelevant for the overall anion conductivity as long as the bulk concentration is still substantial. The mean local conductivity in the space charge zones ̄F8 is given by the cation contribution ( ̄F8+ ) according to:

F̄8 ≈ F̄8+ ∝ √c0+ ∝ � (1)

where Σ is the surface charge and c 0 + the concentration of the free Li-ions in the solution at the site adjacent to the oxide’s sur-face. [ 1 ] The lower index λ refers to the space charge zones, and the lower index ∞ , used in the following, refers to the bulk.

If β ( $∞ , $8 ) measures the fraction of the local pathways (bulk, space charge zones) contributing to the overall conduc-tivity σ m , then to a good approximation:

Fm = $∞ (1 − n)F∞ + $8nF̄≈ $∞ (1 − n) (F∞+ + F∞−) + $8nF̄8+

8

(2)

ϕ : The symbol denotes the volume fraction of SiO 2 , the bulk conductivity of which can be safely neglected. For volume frac-tions that are not too high (below the maximum) all liquid parts can be assumed to percolate, i.e., $∞ ≈ 1 . Unlike solid–solid composites where $8 may in specifi c cases and for good reasons be independent of ϕ , [ 1 ] in the cases addressed here $8 will always be a sensitive function of ϕ . As mentioned above, the problem is characterized by cluster–cluster aggregation and

© 2011 WILEY-VCH Verlag Gwileyonlinelibrary.com

a fractal percolation behavior that can only be tackled numeri-cally. [ 14 ] Nevertheless for cases where ϕ , particle size and mor-phology stay constant Equation 2 works well (see below).

The described method of heterogeneous doping leading to increased Li + conductivity via space charge contributions is also corroborated by the transference number measurements. As long as transport contributions from ion pairs can be excluded and additional resistances are corrected, the measured conduc-tivity in the polarized steady state is

Fpol = F+ (3)

Here we refer to n = 0 . If ion pairs are present and have a non-zero diffusivity D , we must write instead ( s ≡ Dc F 2/RT , F : Far-aday constant, R : gas constant, c : molar volume concentration)

Fpol = F+ +

sF−s + F−

< F+ + F− = F

(4)

The second term sF− / (s + F−) refers to the possibility of effectively transporting Li + in the steady state by counter motion of X − and the ion pair. [ 19 ] For s = 0 Equation 3 is recovered (the presence of immobile ion pairs, however, infl uences the effec-tive diffusion coeffi cient and hence, is important for the tran-sient effects). [ 19 ] We will assume this case for liquid electrolytes with n = 0 . On the same basis, it is straightforward to show that now for the soggy sand electrolytes ( n > 0 ) [ 19 ]

Fm, pol = Fm+ (5)

results with Fm+ defi ned analogously to Fm in Equation 2 . (Note that the index m indicates spatial average.)

Hence, the measured cation transference number is given by

tm+ =Fm, pol

Fm=

$∞ (1 −n)F∞+ + $8nF̄8+

$∞ (1 − n)F∞ + $8nF̄8+,

(6)

while for the effective anion transport number one fi nds

1 − tm+ ≈ $∞ (1 −n)F∞−

Fm (7)

as ̄F8 ≈ F̄8+ .

mbH & Co. KGaA, Weinheim Adv. Funct. Mater. 2011, 21, 3961–3966

FULL P

APER


0,000 0,005 0,010 0,015 0,020 0,025 0,0302,0x10

-4

3,0x10-4

4,0x10-4

5,0x10-4

6,0x10-4

7,0x10-4

8,0x10-4

9,0x10-4

1,0x10-3

1,1x10-3

t m+σσ σσ

m / S

cm-1

volume fraction of SiO (7nm-fumed)2

0,000 0,005 0,010 0,015 0,020 0,025 0,0302,0x10

-43,0x10

-44,0x10

-45,0x10

-46,0x10

-47,0x10

-48,0x10

-49,0x10

-41,0x10

-31,1x10

-31,2x10

-31,3x10

-31,4x10

-31,5x10

-3

volume fraction of SiO (10nm)2

t m+σ m

/ S

cm-1

Figure 5 . a) The effective cation conductivity tm+Fm = Fm+ as a function of n for the system 1 M LiClO 4 /PEG-150/SiO 2 (7 nm-fumed). b) The effective cation conductivity tm+Fm = Fm+ as a function of n for the system 1 M LiClO 4 /PEG-150/SiO 2 (10 nm)

Figure 5 shows the course of the overall Li + conductivity tm+Fm = Fm+ as a function of n . Obviously it roughly parallels the Fm (n) function, [ 14 ] indicating that it is primarily the Li + con-ductivity which is positively impacted by SiO 2 addition. Unlike the solid–solid composites for which the space charge contribu-tions can exceed bulk conductivities by orders of magnitudes and hence dominate the overall conductivity (Fm) , here a sub-stantial part of Fm is formed by the unaffected bulk ( F∞ ). Hence the comparatively modest transference number increase is not in contradiction but rather agrees with locally greatly enhanced Li + conductivities (and conforms with the rather modest σ -increase �Fm /F∞ = (Fm − F∞) /F∞ ). The following exemplary estimate shows this. Combining Equation 2 and Equation 6 one obtains for tm+ as a function of the total conductivity enhance-ment for j = 1 :

tm+ =

�Fm /F∞ + t+

1 + �Fm /F∞ (8)

with t+ = tm+ (n = 0). Equation 8 assumes a pure Li + conduc-tivity in the space charge zones. We recognize that an increase in the cation transference number from t+ to tm+ from typi-cally 0.33 to 0.5 corresponds indeed to a �Fm /F∞ of about 1.8. Table 1 compiles experimental tm+ values and shows that there is reasonable consistency. The fact that the experimental values

© 2011 WILEY-VCH Verlag GmAdv. Funct. Mater. 2011, 21, 3961–3966

Table 1. Comparison of t m + (calculated) with t m + (experimental) and �FmF∞ for composite electrolytes consisting of 1 M LiClO 4 /PEG-150/SiO 2 .

Sample name �FmF∞

t m + (calculated)

t m + (experimental)

1M LiClO 4 /PEG-150 fi ller free electrolyte 0 0.33 0.33

ϕ SiO2(7nm-fumed) = 0.005 0.9 0.6 0.6

ϕ SiO2(7nm-fumed) = 0.01 0.8 0.6 0.5

ϕ SiO2(7nm-fumed) = 0.02 1.1 0.7 0.6

ϕ SiO2(7nm-fumed) = 0.03 0.9 0.6 0.3

ϕ SiO2(10nm) = 0.005 0.9 0.6 0.6

ϕ SiO2(10nm) = 0.01 0.6 0.6 0.5

ϕ SiO2(10nm) = 0.02 0.5 0.6 0.8

ϕ SiO2(10nm) = 0.03 0.9 0.6 0.5

are somewhat smaller than the calculated ones is expected as Equation 8 represents the maximum effect. At n > nc where dFm/dn < 0 , Equation 8 will even fail grossly. But it is striking that experimental t -values do still exceed t+ . This indicates that the local mechanism is still active, but it is the percolation problem that obscures the picture.

In short: the semi-quantitative agreement corroborates the assumption of a pure Li conductivity at the surface leading to tm+ values substantially exceeding the values of the SiO 2 -free electrolytes. The fact that they are smaller than 100% is due to the signifi cant contribution of the conductive bulk. Blocking of pathways at higher n leads to loss in the overall conductivity but not necessarily to a lower transference number.

4. Conclusion and Outlook

The mechanism of the conductivity of the soggy sand electrolytes based on the results on PEG can be summarized as follows:

1. The acidic SiO 2 particles adsorb the anions of the Li-salt, free Li + ions out of the ion pairs leading to an increased Li + con-ductivity in the double layer

2. The anion conductivity is accordingly lowered leading locally to almost single ion transport.

3. The concentration effect appears to be dominant. Enhanced cation mobility because of immobilized anions is nonethe-less possible.

4. The local picture is very similar to ion exchange electrolytes, such as Nafi on [ 25 ] where the anion is immobilized by covalent bonding, rather than by adsorption forces as here. This com-parison is given in Figure 6 .

• The initial colloidal particles agglomerate on thermal en-counter leading to fractal percolating networks.

• Owing to the substantial conductivity of the SiO 2 free elec-trolyte, the overall transport is still infl uenced by the unper-turbed bulk.

• The networks coarsen as this leads to variation of the sur-face energy. This results in disconnection and lowered overall conductivity.

• Under favorable conditions the coarsening kinetics is suffi ciently self-decelerating to come to a steady state. In unfavorable cases, sedimentation occurs.

bH & Co. KGaA, Weinheim 3965wileyonlinelibrary.com

FULL

PAPER

396


Figure 6 . Analogy of local ion conduction mechanism between a) ion exchange electrolytes (e.g., Nafi on) and b) soggy sand electrolytes.

• The steady state conductivity typically shows a dome-shaped behavior. The decreased conductivity at high n can be ascribed to pathway interruption or salt trapping.

The soggy sand electrolytes are a highly promising class of electrolytes as they comprise a variety of advantages: improved mechanical behavior (partial shapeability), enhanced ion con-ductivity, high Li transference numbers, improved safety and compatibility with nanotechnology. The fi eld is in its infancy and the manifold of control parameters (nature of liquid and solid phase, surface density, size, shape, volume fraction, nature and concentration of salt etc.) lead to various degrees of freedom for further improvement. Further studies should also be devoted to stability issues. One major challenge is to come up with morphologies that are stable (possibly highly packed) but still show suffi cient conductivities. The other important issue is a better control of the surface chemistry of the fi ller particles and hence a better reproducibility with respect to local

© 2011 WILEY-VCH Verlag 6 wileyonlinelibrary.com

[ 1 ] J. Maier , Prog. Solid State Chem. 1995 , 23 , 171 . [ 2 ] C. C. Liang , J. Electrochem. Soc. 1973 , 120 , 1289 . [ 3 ] K. Shahi , J. B. Wagner , App. Phys. Lett. 1980 , 37 , 757 . [ 4 ] J. Maier , Ber. Bunsenges. Phys. Chem. 1986 , 90 , 26 . [ 5 ] H. Yamada , A. J. Bhattacharyya , J. Maier , Adv. Funct. Mater. 2006 , 16 ,

525 . [ 6 ] K. Hariharan , J. Maier , J. Electrochem. Soc. 1995 , 142 , 3469 . [ 7 ] S. Beyazyildirim , K. D. Kreuer , M. Schuster , A. J. Bhattacharyya ,

J. Maier , Adv. Mater. 2008 , 20 , 1274 . [ 8 ] F. Croce , R. Curini , A. Martinelli , L. Persi , F. Ronci , B. Scrosati ,

R. Caminiti , J. Phys. Chem. B 1999 , 103 , 10632 . [ 9 ] W. Wieczorek , D. Raducha , A. Zalewska , J. R. Stevens , J. Phys. Chem.

B 1998 , 102 , 8725 . [ 10 ] A. J. Bhattacharyya , J. Maier , R. Bock , F. F. Lange , Solid State Ionics

2006 , 177 , 2565 . [ 11 ] S. K. Das , A. J. Bhattacharyya , J. Phys. Chem. B 2010 , 114 , 6830 . [ 12 ] A. J. Bhattacharyya , J. Maier , Adv. Mater. 2004 , 16 , 811 . [ 13 ] A. Jarosik , S. Hore , N. Kaskhedikar , C. Pfaffenhuber , J. Maier , Elec-

trochim. Acta , submitted. [ 14 ] A. Jarosik , U. Traub , A. Bunde , J. Maier , Phys. Chem. Chem. Phys.

2011 , 13, 2663. [ 15 ] M. H. Hebb , J. Chem. Phys. 1952 , 20 , 185 . [ 16 ] I. Yokota , J. Phys. Soc. Jpn. 1961 , 16 , 2213 . [ 17 ] J. Evans , C. A. Vincent , P. G. Bruce , Polymer 1987 , 28 , 2324 . [ 18 ] N. Kaskhedikar , J. Paulsdorf , A. Burjanadze , Y. Karatas , B. Roling ,

H. D. Wiemhofer , Solid State Ionics 2006 , 177 , 2699 . [ 19 ] J. Maier , Adv. Func. Mater. 2011 , 21 , 1448. [ 20 ] C. Pfaffenhuber , S. Beyazyildirim Sörgel , K. Weichert , T. Mundinger ,

M. Bele , D. Samuelis , J. Am. Chem. Soc., submitted. [ 21 ] A. Bunde , S. Havlin , Springer-Verlag Berlin/Heidelberg , 1991 . [ 22 ] W. C. K. Poon , M. D. Haw , Adv. Colloid Interface Sci. 1997 , 73 , 71 . [ 23 ] M. Kosmulski , J. Colloid Interface Sci. 1998 , 208 , 543 . [ 24 ] G. V. Franks , J. Colloid Interface Sci. 2002 , 249 , 44 . [ 25 ] A. Eisenberg , H. L. Yeager , Perfl uorinated Ionomer Membranes , The

American Chemical Society , Washington DC , 1982 .

mechanism and network morphology. Even though the self-deceleration of coarsening is a very encouraging fi nding for generating soggy sand electrolytes of practical importance, the full potential will only be exploited, if dense SiO 2 networks with high volume fraction will be formed (in the limit density packed structures) that are morphologically stable and still allow for suffi cient percolation of the liquid phase.

Received: February 14, 2011 Published online: August 24, 2011

GmbH & Co. KGaA, Weinheim Adv. Funct. Mater. 2011, 21, 3961–3966

Electrochemical Investigations of Polyethylene Soggy Sand Electrolye

Documents