This action might not be possible to undo. Are you sure you want to continue?

Welcome to Scribd. Sign in or start your free trial to enjoy unlimited e-books, audiobooks & documents.Find out more

Download

Standard view

Full view

of .

Look up keyword

Like this

Share on social networks

1Activity

×

0 of .

Results for: No results containing your search query

P. 1

Steric Effects in the Dynamics of Electrolytes at Large Applied Voltages. I. Double-Layer ChargingRatings:

(0)|Views: 20|Likes: 0

Published by michelle_lil

See more

See less

https://www.scribd.com/doc/109408635/Steric-Effects-in-the-Dynamics-of-Electrolytes-at-Large-Applied-Voltages-I-Double-Layer-Charging

10/08/2012

text

original

Steric effects in the dynamics of electrolytes at large applied voltages. I. Double-layer charging

Mustafa Sabri Kilic and Martin Z. Bazant

Department of Mathematics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA

Armand Ajdari

Laboratoire de Physico-Chimie Theorique, UMR ESPCI-CNRS 7083, 10 Rue Vauquelin, F-75005 Paris, France

͑

Received 2 November 2006; published 16 February 2007

͒

The classical Poisson-Boltzmann

͑

PB

͒

theory of electrolytes assumes a dilute solution of point charges withmean-ﬁeld electrostatic forces. Even for very dilute solutions, however, it predicts absurdly large ion concen-trations

͑

exceeding close packing

͒

for surface potentials of only a few tenths of a volt, which are oftenexceeded, e.g., in microﬂuidic pumps and electrochemical sensors. Since the 1950s, several modiﬁcations of the PB equation have been proposed to account for the ﬁnite size of ions in equilibrium, but in this two-partseries, we consider steric effects on diffuse charge dynamics

͑

in the absence of electro-osmotic ﬂow

͒

. In thisﬁrst part, we review the literature and analyze two simple models for the charging of a thin double layer, whichmust form a condensed layer of close-packed ions near the surface at high voltage. A surprising prediction isthat the differential capacitance typically varies nonmonotonically with the applied voltage, and thus so doesthe response time of an electrolytic system. In PB theory, the differential capacitance blows up exponentiallywith voltage, but steric effects actually cause it to decrease while remaining positive above a threshold voltagewhere ions become crowded near the surface. Other nonlinear effects in PB theory are also strongly suppressedby steric effects: The net salt adsorption by the double layers in response to the applied voltage is greatlyreduced, and so is the tangential “surface conduction” in the diffuse layer, to the point that it can often beneglected compared to bulk conduction

͑

small Dukhin number

͒

͑

s

͒

: 82.45.Gj, 61.20.Qg

I. INTRODUCTION

In this two-part series, we develop a simple analyticaltheory for the dynamics of electrolytes, taking into accountsteric effects of ﬁnite ion size. Motivated by recent experi-ments in microﬂuidics, microbatteries, and electrochemicalsensors, our motivation is to describe the response of anelectrolyte to an applied voltage of several volts, which islarge enough to cause crowding of ions near a surface, evenif the bulk solution is very dilute and in the absence of sur-face reactions. The ions thus lose their classical Poisson-Boltzmann distribution, which has major implications fortheir dynamics.As a guide to the reader, we summarize the main results.The present “Part I” begins in this section with a historicalreview of dilute solution theory, its limitations at large volt-ages, and attempts to account for steric hindrance, speciﬁcinteractions, and many-body electrostatics.As a ﬁrst approxi-mation, we focus only on steric effects and analyze the dy-namical response of a thin diffuse layer to a large appliedvoltage, using two simple continuum models

͑

Sec. II

͒

. Thekey results, common to both steric models, are

͑

i

͒

the diffuselayer’s differential capacitance is bounded and decreases atlarge voltage

͑

Sec. III

͒

, and

͑

ii

͒

it cannot easily engulf enough ions to perturb the bulk concentration or to conductsigniﬁcant tangential currents

͑

Sec. IV

͒

. These predictionsare completely opposite to those of dilute solution theory

͑

based on the Gouy-Chapman model of the diffuse layer

͒

. Inthe companion paper, “Part II”

͓

͔

, we propose general, time-dependent equations with steric constraints and revisit theparallel-plate charging problem in more detail.

A. Dilute solution theory

For the past century, dilute solution theory has providedthe standard model of electro-diffusion of ions

͔

andelectrokinetic phenomena

͓

͔

. The fundamental assumptionis that the chemical potential of a pointlike ion

i

in a dilutesolution has the simple form

i

=

kT

ln

c

i

+

z

i

e

,

͑

1

͒

where

z

i

e

is the charge,

c

i

the concentration, and

the elec-trostatic potential, determined in a mean-ﬁeld approximationby Poisson’s equation,− ·

͑

͒

=

=

͚

i

z

i

ec

i

,

͑

2

͒

typically with a constant permittivity

. The form

͒

issometimes called the “ideal” component of the chemical po-tential

͓

͔

, to which various “excess” components at ﬁniteconcentration can be added

͑

see below

͒

.In many situations, it is assumed that the ions are in qua-sithermal equilibrium with a Boltzmann distribution,

c

i

=

c

i

0

e

−

z

i

e

/

kT

͑

3

͒

with a reference concentration

c

i

0

, in which case Eq.

͑

͒

re-duces to the Poisson-Boltzmann equation

͑

PB

͒

. For example,to analyze a thin double layer, it is natural to choose

c

i

0

to bethe concentration of species

i

in the nearby neutral bulk so-lution. In most situations, the PB equation is hard enough tosolve that the Debye-Hückel linearization for small poten-tials,

͉

͉

T

, is required for analytical progress, where

T

=

kT

/

z

i

e

is the thermal voltage

͑Ϸ

25 mV for monovalentions at room temperature

͒

.

PHYSICAL REVIEW E

75

, 021502

͑

2007

͒

1539-3755/2007/75

͑

2

͒

/021502

͑

16

͒

©2007 The American Physical Society021502-1

The well-known exception, which admits a simple solu-tion, is the case of a symmetric binary

z

:

z

electrolyte in aone-dimensional

͑

1D

͒

geometry, where the PB equation inthe form

2

x

2

= 2

zec

0

sinh

ͩ

ze

kT

ͪ

͑

4

͒

was solved analytically by Gouy

͓

͔

and Chapman

͓

͔ ͑

GC

͒

for a semi-inﬁnite electrolyte of bulk concentration

c

0

=

c

+0

=

c

−0

near a ﬂat charged surface

͔

.

͑

It is less well-knownthat Gouy

͓

also solved the case of an asymmetric 2

z

:

z

electrolyte.

͒

Although this solution may seem quite special, itdescribes the most common situation where diffuse charge isconﬁned to a thin capacitorlike “double layer” near a solidsurface. The width of the diffuse layer is the Debye

͓

͔ ͑

or,more properly, Gouy

͔͒

screening length,

D

=

ͱ

kT

2

z

2

e

2

c

0

͑

5

͒

as can be seen by scaling Eq.

͒

with

x

˜

=

x

/

D

and

˜

=

/

T

to obtain the dimensionless form,

˜

Љ

=sinh

˜

. Thescreening length

͑

D

Ϸ

1–100 nm in aqueous electrolytes

͒

istypically much smaller than any geometrical length, so bulksolution remains quasineutral with diffuse charge conﬁned tothin, locally ﬂat, quasiequilibrium double layers.Due to its analytical simplicity and wide applicability, theGouy-Chapman solution has become the standard model forhighly charged double layers. It is the basis for the classicaltheory of tangential “surface conduction” through the diffuselayer

͔

, as well as the recently noted phenomenon of salt “adsorption”

͑

or “uptake”

͒

from the neutral bulk by thediffuse layer in response to a large applied voltage

͓

͔

.Such predictions of the GC model have major implicationsfor electrokinetic phenomena of the ﬁrst kind, such aselectro-osmosis, electrophoresis, streaming potential, anddiffusiophoresis, at large surface potentials

͓

͔

.Dilute solution theory has also been used in nearly everymodel of diffuse charge during the passage of current. Nearequilibrium, the ﬂux density of ion

i

is proportional to thegradient of its chemical potential

͑

͒

,

F

i

= −

b

i

c

i

i

= −

D

i

ͩ

c

i

+

z

i

ekT c

i

ͪ

,

͑

6

͒

where Einstein’s relation,

D

i

=

b

i

/

kT

, relates the ion’s mobil-ity

b

i

to its diffusivity

D

i

. For a system in quasisteady state,·

F

i

=0, the nonzero current,

J

=

͚

i

z

i

eF

i

, only slightly per-turbs the Boltzmann distribution of the ions. As a result, theGC solution is also widely used to describe diffuse-layereffects on electrode reaction kinetics

͔ ͑

the Frumkincorrection

͔͒

, up to Nernst’s limiting current, where thebulk concentration of an electroactive species vanishes. Atlarger “superlimiting” currents, the PB equation loses valid-ity, but dilute-solution theory is still used to describe diffusecharge, which loses its Boltzmann distribution

͔

and ex-tends into the bulk as “space charge”

͓

͔

, while retaining aninner boundary layer at the screening length

͔

. The dilute-solution space-charge model is the basis for theories of elec-trokinetic phenomena of the second kind, such as superfastelectrophoresis

͓

͔

and hydrodynamic instability at a limit-ing current

͓

͔

.Dilute solution theory has also been used to describe thedynamics of electrolytes, subject to time-dependent appliedvoltages. The classical description comes from the Poisson-Nernst-Planck equations

͑

PNP

͒

, which consist of Eq.

͑

͒

andmass conservation laws for the ions,

c

i

t

= − ·

F

i

=

D

i

ͫ

2

c

i

+

z

i

ekT

·

͑

c

i

͒

ͬ

,

͑

7

͒

where

D

i

ϭ

constant is normally assumed. Again, general so-lutions are only possible upon linearization, where the poten-tial satisﬁes the Debye-Falkenhagen equation

͔

. In thenonlinear regime, for thin double layers, the GC solutionagain holds since the diffuse-charge remains in quasiequilib-rium, and the diffuse-layer acts like a voltage-dependent dif-ferential capacitance in series with a bulk resistor

͓

. Suchequivalent circuit models can be derived systematically fromthe PNP equations by asymptotic analysis, which also revealscorrections at large voltages, due to bulk diffusion in re-sponse to salt adsorption

͓

͔

and surface conduction

͓

͔

atlarge applied voltages. Such nonlinear diffuse-layer effectscould be important in interpretting impedance spectra inelectrochemical sensing

͔

, or in understanding high-rate thin-ﬁlm rechargeable batteries

͓

͔

.Another current motivation to study nonlinear diffuse-charge dynamics comes from “induced-charge” electroki-netic phenomena

͓

͔

. The preceding models of double-layerrelaxation have been used extensively in theories of acpumping of liquids over electrode arrays

͓

͔

, induced-charge electro-osmotic ﬂows around metallic colloids

and microstructures

͓

, and dielectrophoresis

and induced-charge electrophoresis

͓

͔

of polarizable particles in electrolytes, although a nonlinearanalysis based on the PNP equations has not yet been at-tempted for any of these situations. This may be a goodthing, however, since we will show that dilute solutiontheory generally breaks down in the regime of experimentalinterest, where the applied voltage is much larger than thethermal voltage,

V

Ͼ

T

.

B. Validity of the nonlinear model for dilute solutions

Dilute solution theory provides a natural starting point tounderstand nonlinear effects in electrolytes, and the GCmodel is used in all of the examples above to model thediffuse layer at large applied potentials. In spite of its math-ematical appeal, however, nonlinear dilute solution theoryhas limited applicability, due to the exponential sensitivity of the counterion concentration to voltage. For example, in thetypical case of quasiequilibrium, the Boltzmann factor

͑

͒

brings nonlinearity to the PB equation

͒

when

Ͼ

T

, butthis dependence is so strong that the surface potential cannotget much larger without invalidating the assumption of adilute solution.To see this, note that there must be a maximum concen-tration,

c

max

=

a

−3

, of counterions packed with typical spacing

a

near a highly charged surface. This effective ion size is

KILIC, BAZANT, AND AJDARI PHYSICAL REVIEW E

75

, 021502

͑

2007

͒

021502-2

clearly no smaller than the ionic radius

͑

typically

Ϸ

1 Å

͒

,although it could be considerably larger

͑

several nm

͒

, takinginto account hydration effects and ion-ion correlations, espe-cially in a large electric ﬁeld. For the sake of argument,consider a very dilute bulk solution,

c

0

=10

−5

M, with

a

=3 Å and

z

=1. The nonlinear regime begins at a diffuse-layer potential

⌿

D

Ϸ

kT

/

ze

=25 mV, but the steric limit isreached in dilute solution theory at

⌿

c

Ϸ

13

kT

/

ze

=330 mV,where

⌿

c

= −

kT ze

ln

͑

a

3

c

0

͒

=

kT ze

ln

ͩ

c

max

c

0

ͪ

.

͑

8

͒

Since the solution ceases to be “dilute” well below the stericlimit, there is only a narrow window of surface potentials

͑Ϸ

25–200 mV

͒

, sometimes called “weakly nonlinear re-gime”

͓

, where nonlinearity arises and the fundamentalassumption

͑

remains valid.Unfortunately, the most interesting predictions of the non-linear PB theory tend to be in the “strongly nonlinear re-gime,” where the dilute approximation fails dramatically. Forexample, the Dukhin number, which controls the relative im-portance of tangential conductivity in a thin diffuse layer,

s

,compared to bulk conductivity,

, at a geometrical scale

L

,has the following form

͓

͔

,Du=

s

L

= 4

D

L

sinh

2

ͩ

ze

⌿

D

4

kT

ͪ

,

͑

9

͒

assuming the Gouy-Chapman model. This is also the dimen-sionless ratio,

⌫

s

/

c

0

L

, of the excess surface concentration of ions,

⌫

s

, relative to the bulk concentration, so it also governsthe

͑

positive

͒

adsorption of neutral salt from the bulk in re-sponse to an applied voltage

͓

͔

. The general derivationof Eq.

͒

assumes a thin double layer

͔

, but in that case

͑

D

L

͒

a large Dukhin number corresponds to situationswhere the steric constraint is signiﬁcantly violated,

⌿

D

Ͼ⌿

c

, rendering Eq.

͑

͒

inappropriate. Similar concerns ap-ply to other nonlinear effects in Gouy-Chapman theory, suchas the rapid increase of the differential capacitance

͑

deﬁnedbelow

͒

with surface potential,

C

D

=

D

cosh

ͩ

ze

⌿

D

2

kT

ͪ

,

͑

10

͒

which would have important implications for electrochemi-cal relaxation around conductors

͔

and for acelectro-osmosis

͓

͔

.

C. Beyond the Poisson-Boltzmann picture

We are certainly not the ﬁrst to recognize the limitationsof dilute solution theory. Historically, concerns about the un-bounded capacitance of a thin diffuse layer in the GC solu-tion

͑

ﬁrst motivated Stern

͓

͔

to hypothesize that theremust also be a compact layer of adsorbed ions on the surface,as originally envisioned by Helmholtz

͓

. The Stern layeris where electrochemical reactions, such as ion dissociation

͑

setting the equilibrium charge on a dielectric

͒

and/or redoxcouples

͑

setting the Faradaic current at an electrode

͒

, arebelieved to occur, within a molecular distance of the solidsurface

͓

͔

. The Stern layer capacitance

͑

see below

͒

helpsto relieve the overcharging of the diffuse layer in Gouy-Chapman theory, but, due to steric constraints, it too cannotconceivably withstand a voltage much larger than

⌿

c

. In-stead, at larger voltages, the region of ion accumulation mustinevitably extend away from the surface into the solution,where ions undergo hindered transport in a concentrated so-lution without having speciﬁc interactions with the solid.The most basic aspect of a concentrated solution is theﬁnite molecular length scale,

a

0

. Over half a century ago,Wicke and Eigen

͔

made perhaps the ﬁrst attempts toextend dilute solution theory to account excluded volumeeffects in a simple statistical mechanical treatment. Thetheory was developed further in the past decade by Iglic andKral-Iglic

͓

͔

and Borukhov, Andelman, and Orland

. These authors proposed free energy functionals,based on continuum

͑

mean-ﬁeld

͒

approximations of the en-tropy of equal-sized ions and solvent molecules, which theyminimized to derive modiﬁed Poisson-Boltzmann equations

͑

MPB

͒

. The motivation for this work was mainly to addressthe effect of large ions, whose sizes may be comparable tothe screening length, in an equilibrium diffuse layer with

⌿

D

Ϸ

T

. Our main point here is that crowding effects canalso be very important for small ions near a polarizable sur-face, when subjected to a “large” voltage exceeding thethreshold

⌿

c

. No matter how dilute is the bulk solution, asufﬁciently large electric ﬁeld can always draw enough ionsto the surface to create a concentrated solution in the diffuselayer.There have also been attempts to go beyond the mean-ﬁeld approximation of steric effects, by treating speciﬁc

͑

andin some cases also Coulombic

͒

ion-ion and ion-wall interac-tions. Simple MPB equations have been proposed, whichmodify the charge density to account for the spatial correla-tion function between an ion in solution and a ﬂat wall

͑

viaan effective external potential

͒

based on molecular dynamicssimulations

͓

͔

, but such models are not easily extendedto any other geometry, such as a rough wall

͔

. Correctionsof this type can also be obtained from a general probabilisticmodel of interacting ions, whose dynamics are given by non-linearly coupled Langevin equations. Using this approach,Schuss, Nadler, and Eisenberg rigorously derived “condi-tional PNP”

͑

and PB

͒

equations where each ion concentra-tion

c

i

͑

r

͒

in the mean-ﬁeld Poisson’s equation

͒

is replacedby the conditional probability density of ﬁnding an ion at acertain position, given the positions of the other ions

, although a simple closure of the model requiresfurther assumptions about statistical correlations.There are a variety of general statistical mechanical ap-proaches from liquid state theory

͓

, which have been ap-plied to electrolytes, taking into account not only steric ef-fects, but also many-body electrostatic correlations

͓

.Since the 1970s, the modest, but challenging, goal has beento accurately predict the equilibrium distribution of ions inMonte Carlo simulations of the “primitive model” of chargedhard spheres in a homogeneous dielectric continuumbounded by a hard, charged wall. Typically, the model isstudied in the limit of “small” surface potentials

͑

⌿

D

Ϸ

kT

/

e

͒

relevant for equilibrium surfaces. For example, aplethora of MPB equations

͑

such as “MPB4,” “MPB5,”

…

͒

STERIC EFFECTS IN THE DYNAMICS

…

. I. DOUBLE-LAYER

…

PHYSICAL REVIEW E

75

, 021502

͑

2007

͒

021502-3

- Read and print without ads
- Download to keep your version
- Edit, email or read offline

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

CANCEL

OK

You've been reading!

NO, THANKS

OK

scribd