A Model of Electrocodeposition on a Rotating Cylinder Electrode

and

Published 19 December 2006 • © 2006 ECS - The Electrochemical Society
, , Citation Jaesung Lee and Jan B. Talbot 2007 J. Electrochem. Soc. 154 D70DOI 10.1149/1.2398819

1945-7111/154/2/D70

Abstract

A model of particle electrocodeposition on a rotating cylinder electrode was developed to predict the incorporated particle volume fraction in a composite film. Asymptotic solutions to the transport equations for ions within the diffusion boundary layer showed that diffusion of the reactive species in a supporting electrolyte solution is the dominant process to carry current at the cathodic surface. The ratio of the total volume of particles to that of the deposited metal was expressed as a function of two transport parameters, current density, a probability to determine the particle surface concentration, and two experimental fitting parameters.

Export citation and abstractBibTeXRIS

Metallic matrix composite films can be produced by incorporation of particles during electroplating; these films have been extensively used to enhance resistance to wear and corrosion. The advantages of electrocodeposition are uniformity of deposition, high deposition rate, and a low temperature and pressure process. Recently, Stojak et al.1 have reviewed the general process mechanisms and various models proposed for electrocodeposition.

The rotating cylinder electrode (RCE) has been used to study mass transfer and also electrocodeposition.2 The main advantages of an RCE system are a constant shear rate and a uniform potential and current density at the cylindrical electrode surface, which ultimately yields uniformly electrodeposited films. The flow pattern, even at a low rotation speed, is turbulent, which enhances the mass transfer to an electrode.3, 4 Correlation of mass transfer rates for an RCE have been extensively investigated by Eisenberg et al.5, 6 The Chilton-Colburn analogy was used to determine the diffusion boundary layer thickness as a function of inner rotating cylinder diameter, rotational speed, and kinematic viscosity and diffusivity of the electrolyte at the limiting current density. Gabe and Robinson7 also investigated mass transfer in turbulent flow for an RCE system in which the Prandtl–von Kármán model was used to derive the apparent turbulent diffusion coefficient in both the viscous sublayer and fully developed turbulent flow. Stojak and Talbot2 studied electrocodeposition using an RCE to control hydrodynamics as a function of other process variables, such as current density and bath composition, to produce copper-alumina nanocomposites.

To model electrocodeposition, it is important to understand the behavior of suspensions of particles. Jeffrey and Acrivos8 have reviewed the rheological properties of suspensions of neutrally buoyant rigid particles. The factors influencing the properties of suspensions were considered, especially the effective viscosity as a function of the particle volume fraction. The mass transfer of suspensions of inert particles has been studied with both a rotating disk electrode9 and an RCE,10 where the effective diffusivity and viscosity were expressed in terms of the volume fraction of particles. When microspheres of Equation or symbol description not available5–80μm diameter, which is much larger than the diffusion boundary layer thickness of a reactive ion, were used, small-scale microconvection induced by the rotation of particles near an electrode enhanced mass transfer of the reactive ion, and thus increased the limiting current density due to a particle-free wall layer. However, when colloidal particles (Equation or symbol description not available<1μm diameter) were used, the particle-free wall layer was not formed due to nonhydrodynamic forces acting on the particles in suspension, and some particles in the neighborhood of the electrode were incorporated into the growing deposited film. Stojak and Talbot11 used an RCE system to experimentally investigate the limiting current density and the polarization behavior with and without particles of Equation or symbol description not available50nm diameter as a function of electrode rotational rate and particle loading. For the mass transfer limited region, there was no enhancement of mass transfer due to particles at and less than loadings of Equation or symbol description not available120g∕L (Equation or symbol description not available3.5vol% in solution). When the particle loading was increased to Equation or symbol description not available150g∕L (Equation or symbol description not available4.6vol% in solution), the limiting current densities increased by 14, 32, and 19% at rotational speeds of 500, 1000, and Equation or symbol description not available1500rpm, respectively.

A surface force boundary layer approximation (SFBLA) has been used to explain the role of surface forces in particle deposition.12–16 This approximation assumes that very near the electrode surface particles are transported by diffusion in the presence of a surface interaction potential that includes both attractive and repulsive forces. Typically in the formulation of an SFBLA, the mass flux of particles to the surface is expressed by a rate constant times the surface concentration, where the rate constant is given by the integration of an interaction potential function.12 This particle mass flux is used as a boundary condition for the equation of convective diffusion of particles to obtain the distribution of the particles in a thin boundary layer near the electrode surface. However, for an electrodeposition process with a supporting electrolyte (of high ionic strength), the effect of the electrical double layer is negligible such that the van der Waals attraction force between a particle and the electrode is the only effective force.12 In the absence of strong electrical-double layer interactions, there does not exist a maximum in the interaction potential so that the integration for the rate constant does not converge to a fixed value.14, 16 The surface concentration then cannot be predicted by an SFBLA approach. Hence, another approach must be used.

Recently, Searson and colleagues17, 18 proposed a kinetic model for electrocodeposition of nanocomposite films for a rotating disk electrode (RDE), which was used to study the deposition of Equation or symbol description not availableNi∕Al2O3 composites. This model took into account the convective diffusion of particles to the growing surface, adsorption of particles, and the effect of gravitational forces on the particle flux. The volume fraction of particles was represented as a function of current density and transport parameters. The gravitational force was found to be important for the electrocodeposition of Equation or symbol description not available300nm particles, but not for Equation or symbol description not available50nm particles.

Based on the experimental results from electrocodeposition of Equation or symbol description not available50nm alumina in copper with an RCE,2 a model is proposed to predict volume fraction of particles in a composite by adapting the models proposed by Shao et al.18 and Celis et al.19 The RDE model of Shao et al. was modified by changing the transport parameters to model an RCE system and representing the flux of particles to an electrode surface from the concepts presented by Celis et al.19 In order to assess the assumptions of the codeposition model, the contributions of the electric field and diffusion within a supporting electrolyte solution were investigated by applying a perturbation method to the transport equation for ions in the diffusion boundary layer.

Model

We consider the motion of a particle within the hydrodynamic boundary layer Equation or symbol description not available(δ0) of an RCE, as shown in Fig. 1. The inner cylinder rotates with the angular velocity ω while the outer cylinder (not shown) is stationary. Figure 1 is divided into four regions: the diffusion boundary layer for particles Equation or symbol description not available(0<y<δp), the diffusion boundary layer for ions Equation or symbol description not available(0<y<δ), the viscous sublayer Equation or symbol description not available(0<y<δo), and the fully developed turbulent region Equation or symbol description not available(yδo), where Equation or symbol description not availabley=r−r1 in the radial direction Equation or symbol description not availabler, and Equation or symbol description not availabler1 is the radius of the inner rotating cylinder. The ionic concentrations at the surface and in the bulk solution are Equation or symbol description not availablecis and Equation or symbol description not availablecib, and Equation or symbol description not availablecs and Equation or symbol description not availablecb are particle concentrations at the surface and in the bulk solution, respectively. In this paper, the variation of concentration of particles is assumed to be confined within the narrow region near the inner rotating cylinder and beyond Equation or symbol description not availableδp the concentration of particles is constant. The boundary layer thickness for particles are discussed later in the section on mass transfer for an RCE.

Figure 1. Refer to the following caption and surrounding text.

Figure 1. Schematic of the concentration profiles for the RCE and a force balance on a particle within the hydrodynamic boundary layer Equation or symbol description not availableδ0. (In reality, the particle is much smaller than Equation or symbol description not availableδp.)

Stojak and Talbot2 investigated the electrocodeposition of inert Equation or symbol description not availableAl2O3 particles in plated copper from an acidic copper sulfate solution of Equation or symbol description not available0.1M Equation or symbol description not availableCuSO4 and Equation or symbol description not available1.2M of Equation or symbol description not availableH2SO4. In the electrolytic solution, the molecules of Equation or symbol description not availableCuSO4 and Equation or symbol description not availableH2SO4 dissociate predominately into three ionic species, Equation or symbol description not availableCu2+, Equation or symbol description not availableH+, and Equation or symbol description not availableHSO4−; only a very small fraction of the bisulfate ions dissociates, so that the concentration of Equation or symbol description not availableSO4− is neglected.20–22 The concentration of Equation or symbol description not availableOH− is also negligible due to the low pH. Thus, the concentrations of these three species, Equation or symbol description not availableCu2+, Equation or symbol description not availableH+, and Equation or symbol description not availableHSO4−, are considered. In this study, particle-particle interactions are not considered as the suspension of colloidal particles is dilute Equation or symbol description not available(<4.6vol%) and inertial forces are negligible due to small particle size of Equation or symbol description not available50nm. Electrokinetic effects were negligible as the zeta potential of the colloidal particles was zero in this highly conductive bath.

Mass transfer in an RCE system

Eisenberg et al.5, 6 extensively investigated mass transfer with an RCE and obtained the correlation of mass transfer rates in the form

Equation (1)Sh=0.0791Re0.7Sc0.356

where the Sherwood number Equation or symbol description not availableSh(=kdi∕DM) is a function of the Reynolds number Equation or symbol description not availableRe(=Vdi∕ν), which is based on the diameter Equation or symbol description not available(di=2r1) of the inner cylinder and the peripheral velocity Equation or symbol description not availableV(=πdiω) of the inner cylinder, and the Schmidt number Equation or symbol description not availableSc(=ν∕DM), which is a ratio of kinematic viscosity ν of the electrolyte to the diffusivity Equation or symbol description not availableDM of the reactive ion (i.e., Equation or symbol description not availableCu2+). The convective mass transfer coefficient Equation or symbol description not availablek(=DM∕δ) can be expressed as

Equation (2)k=iLzMFcMb

where Equation or symbol description not availableiL is the limiting current density, Equation or symbol description not availablez is the charge number, Equation or symbol description not availableF is the Faraday constant, and Equation or symbol description not availablecMb is the bulk concentration of the reactive ion. The diffusion layer thickness at an RCE using Equation or symbol description not availableiL=zMFDMcMb∕δ is calculated as

Equation (3)δ=12.64di0.3ν0.344DM0.356V−0.7

with units of the parameters as given in Table I.

Table I. List of values for physical properties.

Physical propertyValue at Equation or symbol description not available298K Reference
Equation or symbol description not availablea Equation or symbol description not available25nm 2
Equation or symbol description not availablecCu2+ of bulk solution Equation or symbol description not available0.1M 2
Equation or symbol description not availablecH+ of bulk solution Equation or symbol description not available1.1M 2
Equation or symbol description not availablecHSO4− of bulk solution Equation or symbol description not available1.3M 2
Equation or symbol description not availableDM(=DCu2+) Equation or symbol description not available0.714×10−9m2∕s 29
Equation or symbol description not availableDH+ Equation or symbol description not available9.311×10−9m2∕s 29
Equation or symbol description not availableDHSO4− Equation or symbol description not available1.545×10−9m2∕s 29
Equation or symbol description not availabledi Equation or symbol description not available0.012m 2
Equation or symbol description not availableDp Equation or symbol description not available8.73×10−12m2∕s 30
Equation or symbol description not availableSc 1400 
ν Equation or symbol description not available1.0×10−6m2∕s 2
Equation or symbol description not availableρAl2O3 Equation or symbol description not available3970kg∕m3 30
Equation or symbol description not availableρCu Equation or symbol description not available8960kg∕m3 30

The surface concentration for the reactive ion Equation or symbol description not availablecMs can be estimated from the fraction of the limiting current density23

Equation (4)cMs=cMb(1−iniL)

where Equation or symbol description not availablein is the current density at the cathode surface. Because the ratio of the hydrodynamic boundary layer thickness Equation or symbol description not availableδ0 to the diffusion boundary layer thickness δ is proportional to Equation or symbol description not availableSc0.356 by Eq. 1, we obtain24

Equation (5)δ0=δ×Sc0.356=12.64di0.3ν0.7V−0.7

In general, the value of Equation or symbol description not availableSc for an electrolytic solution is large, such that the thickness of the hydrodynamic boundary layer is much larger than that of the diffusion boundary layer; for Equation or symbol description not availableSc=1400 in this study Equation or symbol description not availableδo=13.2δ.25 Because the alumina particle size Equation or symbol description not available(50nm) is small enough for Brownian diffusion to be important, the boundary layer thickness for particles Equation or symbol description not availableδp was calculated from Eq. 3, with Equation or symbol description not availableDp being substituted for Equation or symbol description not availableDM, such that Equation or symbol description not availableδp=0.21δ. Using the values in Table I, the three boundary layer thicknesses are compared in Table II for the RCE rotational rates of the experiments of Stojak and Talbot.2

Table II. Comparison of boundary layer thicknesses Equation or symbol description not available(δ) for various RCE rotational rates.

Rotational rate (rpm)Particle Equation or symbol description not availableδp Equation or symbol description not available(μm) Ionic δ Equation or symbol description not available(μm) Hydrodynamic Equation or symbol description not availableδ0 Equation or symbol description not available(μm)
5007.536476
10004.622293
15003.517221

Lagrangian description of a particle within the diffusion boundary layer

In order to determine the mass flux of particles Equation or symbol description not availableJp in an RCE system, the motion of a single nonconducting particle within the diffusion boundary layer thickness is assumed to be governed by Newton’s second law

Equation (6)m∂v∂t=Fd+Ff+Fc+Fb

where in cylindrical coordinates

Equation (a)Fd=−∇μc=−kBTc∇c=−kBTc(∂c∂rer+1r∂c∂θeθ+∂c∂zez)

Equation (b)Ff=βv=β(vrer+vθeθ+vzez)

Equation (c)Fc=mω2rer

Equation (d)Fb=−mfω2rer

and Equation or symbol description not availablem(=Vpρp) is the mass of the single particle, Equation or symbol description not availableρp is the density of the particle, and Equation or symbol description not availablev is the particle velocity. The diffusion force is Equation or symbol description not availableFd, where Equation or symbol description not availableμc is chemical potential, Equation or symbol description not availablekB is Boltzmann’s constant, Equation or symbol description not availableT is temperature, and Equation or symbol description not availablec is concentration of the particles. The frictional force is Equation or symbol description not availableFf, where Equation or symbol description not availableβ(=6πμa) is the resistance of a spherical particle to the fluid in Stokes flow. The centrifugal and buoyant forces are Equation or symbol description not availableFc and Equation or symbol description not availableFb, respectively, where Equation or symbol description not availablemf(=Vpρf) is the mass of the displaced fluid, and Equation or symbol description not availableρf is the density of the fluid. The gravitation force is not considered as it is perpendicular to the forces taken into account and also was found to be small for the Equation or symbol description not available50nm diameter alumina particles considered (approximately 10 times smaller than the centrifugal force).26 In Searson’s studies,17, 18 the diffusional and gravitational forces were considered the main forces acting on a particle in a rotating disk electrode system. Also, as in those studies, stochastic (Brownian) motion of the particles is neglected.

If only the force balance in the radial direction Equation or symbol description not availabler at steady state is considered, the inner product of Eq. 6 with the normal Equation or symbol description not availableer gives the following force balance

Equation (7)−Ff∙er=−βυr=(Fd+Fc+Fb)∙er=−kBTc∂c∂r+Vp(ρp−ρf)ω2r

The particle flux Equation or symbol description not availableJp=−cvr can then be calculated by substituting the radial velocity Equation or symbol description not availablevr from Eq. 7 to obtain

Equation (8)Jp=−Dp∂c∂r+Ac

where Equation or symbol description not availableA[≡Vp(ρp−ρf)ω2r∕β] is a parameter that represents the effect of the centrifugal force and Equation or symbol description not availableDp=kBT∕β is the diffusion coefficient of the particle. Note that the parameter Equation or symbol description not availableA depends on the density difference between the particle and fluid, particle size, the electrode rotational speed, and fluid viscosity. When the density of a particle is larger than that of the fluid, the buoyancy force is weaker than the centrifugal force, so that the force acting on a particle is always from the rotating cylinder.

By using the boundary layer thickness Equation or symbol description not availableδp, the mass flux for a particle at the surface in Eq. 8 may be approximated by

Equation (9)Jp≅−Dpcb−csδp+Acs=−Bcb+(A+B)cs

where Equation or symbol description not availableB(≡Dp∕δp=0.0791di−0.3ν−0.344Dp0.644V0.7) is the average mass transfer coefficient for a particle, as compared to the mass transfer coefficient for the reactive ion Equation or symbol description not available(Cu2+), Equation or symbol description not availablek≡DM∕δ. By calculating the parameters Equation or symbol description not availableA and Equation or symbol description not availableB and the electrode surface concentration of reactive ions, the mass flux can be determined based on the flow characteristics. In the work of Shao et al. ,18 a similar relationship was developed for the particle flux, except that the two transport parameters represented the effects of the gravitational force and the diffusional flux for the RDE.

Instead of solving the convective-diffusion equation for a particle, a statistical approach is taken in which Equation or symbol description not availableP is defined as the probability for particles to be in the neighborhood of an electrode. Then, the surface concentration of particle Equation or symbol description not availablecs=Pcb, where Equation or symbol description not availableP is estimated from experimental results. If Equation or symbol description not availableP=0, the surface concentration is zero, and as Equation or symbol description not availableP approaches 1, the surface concentration increases to the bulk concentration of particles. Thus, Eq. 9 can be written in terms of the bulk particle concentration as

Equation (10)Jp=[(A+B)P−B]cb

Volume fraction of particles in the deposit

The volume fraction Equation or symbol description not availablexv of particles incorporated in a metal deposit can be expressed as a ratio of the incorporated volume of particles to the total volume of the composite17

Equation (11)xv=npVpNAnpVpNA+nMVm,M

where Equation or symbol description not availablenp is the number of moles of particles in the film, Equation or symbol description not availablenM is the number of moles of metal atoms in the film, Equation or symbol description not availableVp(=4πa3∕3) is the particle volume, Equation or symbol description not availablea is the particle radius, Equation or symbol description not availableVm,M is the molar volume of the metal film, and Equation or symbol description not availableNA is Avogadro’s number.

From the mathematical model for electrocodeposition of Celis et al. ,19 the flux of particles to the surface Equation or symbol description not availableJp is given by

Equation (12)Jp=JMnpnM(itrin)α

where Equation or symbol description not availableJM(=in∕zMF) is the flux of the metallic ions to the electrode, which is proportional to the current density from the reduction of metallic ions, where Equation or symbol description not availablezM is the number of electrons transferred in the reduction reaction.

In electrocodeposition a maximum in the weight (or volume) percentage of particles incorporated as a function of current density has been observed frequently.1, 2 The parameter Equation or symbol description not availableitr was proposed by Celis et al.19 as the transition current density at this maximum incorporation from kinetic to mass transfer limited deposition. The variable α is equal to zero if Equation or symbol description not availablein<itr for kinetic limited deposition and nonzero if Equation or symbol description not availableinitr for mass transfer control. Therefore, under kinetically controlled deposition, the ratio of the particle flux to metallic ion flux is directly equal to the respective molar concentration in the film Equation or symbol description not available(np∕nM=Jp∕JM). In this study, the values of Equation or symbol description not availableitr and α are determined from fitting experimental data.

The ratio Equation or symbol description not availablenp∕nM can be expressed in terms of the volume fraction from Eq. 11

Equation (13)npnM=Vm,MVpNAxv1−xv

By combining Eq. 12, 13, Equation or symbol description not availableJp can be written as a function of volume fraction and current density that can be measured from experiments

Equation (14)Jp=Vm,MVpNAzMFxv1−xvin(itrin)α

By rearranging, the ratio of the total volume of particles Equation or symbol description not availableVptot to that of the deposited metal Equation or symbol description not availableVmetal can be determined

Equation (15)VptotVmetal=xv1−xv=VpNAzMFVm,Min(initr)αJp

Finally, the volume ratio in Eq. 15 can be recast by substitution of Equation or symbol description not availableJp from Eq. 10 as

Equation (16)VptotVmetal=VpNAzMFVm,Min(initr)α[(A+B)P−B]cb

Vereeken et al.17 introduced a probability of particle incorporation to modify the surface concentration of the particle mass flux in Eq. 15, where the probability was assumed to be proportional to the deposition current. The surface concentration of particles was represented in terms of an extrapolated particle surface concentration and a characteristic current density at which the probability is 1. Based on the characteristic current density Equation or symbol description not availableic, the volume ratio was divided into two regions depending on the magnitude of the current density being either larger or smaller than Equation or symbol description not availableic. The characteristic current density, Equation or symbol description not availableic, corresponds to the transition current density, Equation or symbol description not availableitr, of the model of Celis et al.19 However, when Equation or symbol description not availableinic, the volume ratio was taken as inversely proportional to the current density. In this study, using the definition from Celis et al. ’s model, we show that the volume ratio is proportional to current density to the power of α. When Equation or symbol description not availableα=0, the volume ratio equation reduces to the model of Vereeken et al.17 If Equation or symbol description not availableα=1, the volume ratio is independent of current density. To distinguish between the two regions with respect to Equation or symbol description not availableitr, Equation or symbol description not availableα1 and Equation or symbol description not availableα2 are used. When Equation or symbol description not availableinitr, Equation or symbol description not availableα=α1 and the volume ratio will decrease as current density increases, whereas when Equation or symbol description not availablein<itr, Equation or symbol description not availableα=α2 and the volume ratio behaves according to Eq. 15. The magnitude of the volume ratio depends on the magnitude of the transport parameters, Equation or symbol description not availableA and Equation or symbol description not availableB, the probability Equation or symbol description not availableP, and the fitting parameters of Equation or symbol description not availableitr and α. The dependence of volume ratio on current density in both regions is discussed later.

Calculation of ionic concentrations and potential

In a supporting electrolyte where no concentration gradients exist, the current density in the bulk solution is governed by Ohm’s law, and the current is carried by nonreactive ions, such as Equation or symbol description not availableH+ and Equation or symbol description not availableHSO4− as in the acid copper plating system. As the electrical conductivity of the solution in a supporting electrolyte is very large, the potential gradient is very small. Within the diffusion boundary layer thickness, however, the electric field force competes with the concentration gradient in the movement of ions which carry the current.27 Although the concentration of the reactive ion is smaller than the nonreactive ions, the concentration gradient of the reactive ion is significant, as the current is due to the reduction of reactive species in a deposition process. To address the effects of an electric field and diffusion of a single reactive species in a supporting electrolyte solution, the transport equations for the ionic concentration and the electrostatic potential in a diffusion boundary layer δ in the absence of a particle were solved by the successive approximation as given in the Appendix1.

A perturbation method was used to obtain the asymptotic solutions for the ionic concentrations and the electrostatic potential gradient in the diffusion boundary layer28 for a Equation or symbol description not availableCu−H2SO4 plating solution. The variables Equation or symbol description not availablec1, Equation or symbol description not availablec2, and Equation or symbol description not availablec3 correspond to the concentration of Equation or symbol description not availableCu2+, Equation or symbol description not availableH+, and Equation or symbol description not availableHSO4−, respectively. The solutions (including up to the second-order terms) of the normalized concentrations Equation or symbol description not availableCi=ci∕[H2SO4]) and potential Equation or symbol description not availableΦ[=ϕ∕(RT∕F)] within the ionic diffusion boundary thickness δ are

Equation (17)C1(ξ)=ε[In(1−ξ)+1]+ε2[12z1λ1In(ξ2−1)−z1λ1(In+1)(ξ−1)]

Equation (18)C2(ξ)=1+ε[z2λ1(1−ξ)−1]+ε2[12λ3(ξ2−1)−λ6(ξ−1)]

Equation (19)C3(ξ)=1+ε[z3λ1(1−ξ)+1]+ε2[12λ4(ξ2−1)−λ7(ξ−1)]

Equation (20)dΦ(ξ)dξ=ελ1+ε2[λ2ξ+λ5]

where Equation or symbol description not availableε=[CuSO4]∕[H2SO4], and is used as the small parameter for the asymptotic solutions, which is reasonable for an acidic copper plating bath. Here, Equation or symbol description not availableIn is the normalized current density divided by Equation or symbol description not availablez1FD1b1∕δ and ξ is the radial direction coordinate normalized by δ. The coefficients Equation or symbol description not availableλj(j=1–7) are given in the Appendix1. The derivation of Eq. 17, 18, 19, 20 is shown in detail in the Appendix1. The normalized current density Equation or symbol description not availableIn at the electrode is

Equation (21)In∣ξ=0=Ψp+Ψd

where

Equation or symbol description not availableΨp=−z1{ε[In+1]+ε2[−12z1λ1In+z1λ1(In+1)]}{λ1+ελ5}

and

Equation or symbol description not availableΨd=In+εz1λ1(In+1)

In Eq. 21, Equation or symbol description not availableΨp and Equation or symbol description not availableΨd are the contributions of the electric field and diffusion to the current density, respectively.

Results and Discussion

To understand the transport processes for an RCE, the contributions of the electric field and diffusion to the current density using the solutions of Eq. 21 were calculated for small values of Equation or symbol description not availableε(=[CuSO4]∕[H2SO4]). As shown in Fig. 2, the errors from the asymptotic solution in Eq. 21 are 0.0525 for Equation or symbol description not availableε=0.083 (thick line) and 0.0008 for Equation or symbol description not availableε=0.01 (thin line). As ε is small (the solution is a well-supported electrolyte), the contribution of diffusion is dominant. In this study, the value Equation or symbol description not availableε=0.083 for a plating bath of Equation or symbol description not available0.1MCuSO4 and Equation or symbol description not available1.2MH2SO4 (Ref. 2) is used for all calculations. The diffusion of Equation or symbol description not availableCu2+ is the dominant process, which increases from Equation or symbol description not available∼80 to Equation or symbol description not available∼99% as the average current density increases. The current density due to electromigration of Equation or symbol description not availableCu2+ is negligible. Therefore, it is assumed that the current density is only due to diffusion; electromigration is ignored.

Figure 2. Refer to the following caption and surrounding text.

Figure 2. (Color online) The contributions of the electric field Equation or symbol description not availableΨp and diffusion Equation or symbol description not availableΨd to carry the current as a function of Equation or symbol description not availableIn, where the thin and thick lines correspond to Equation or symbol description not availableε=0.01 and Equation or symbol description not availableε=0.083, respectively.

The concentration profiles of the three ionic species were calculated by Eq. 17, 18, 19 for a plating bath of Equation or symbol description not available0.1MCuSO4 and Equation or symbol description not available1.2MH2SO4. The distribution of the concentration of each ion for Equation or symbol description not availableIn=−0.8 is shown in Fig. 3, where each concentration is normalized by its bulk concentration. The concentration gradients for Equation or symbol description not availableH+ and Equation or symbol description not availableHSO4− counteract the gradient of the electrostatic potential, so that their fluxes are zero at the electrode surface, which induces the small change of concentrations of Equation or symbol description not availableH+ and Equation or symbol description not availableHSO4− in the diffusion layer.27 The potential gradient ∂Φ∕∂ξ as calculated from Eq. 20 is almost constant in the diffusion layer and is small in the bulk solution because of the high conductivity bath.

Figure 3. Refer to the following caption and surrounding text.

Figure 3. (Color online) Normalized ionic concentration profiles in the diffusion boundary layer, where Equation or symbol description not availableIn=−0.8.

Figure 4 compares surface concentration for Equation or symbol description not availableCu2+ as a function of current density calculated from experimental data of Stojak and Talbot2 for electrocodeposition with a particle loading of Equation or symbol description not available158g∕L using Eq. 4 and predicted from the analytical solutions from Eq. 17. The surface concentration of Equation or symbol description not availableCu2+ is linear up to the limiting current density as determined from Eq. 4. The thin and thick lines for each set of data correspond to the surface concentration calculated from experimental limiting current density data2 and from Equation or symbol description not availableiLt=zMFDMcMb∕δ using the diffusion layer thickness in Table I. At a low rotational speed of Equation or symbol description not available500rpm, the theoretical predictions agree well with experimental results. At the higher rotational rate, the values from the asymptotic approximation are somewhat higher than the experimental results. This difference may be due to the absence of particles and convection in determining the surface concentration from the analytical solution of Eq. 17.

Figure 4. Refer to the following caption and surrounding text.

Figure 4. A comparison of normalized surface concentration of Equation or symbol description not availableCu2+ for a particle loading of Equation or symbol description not available158(g∕L) from experimental values using Eq. 4 (open symbols) and model Eq. 17 (filled symbols) for Equation or symbol description not available500, Equation or symbol description not available1000, and Equation or symbol description not available1500rpm. The thin and thick lines are calculated from Eq. 4 Equation or symbol description not availableiL=iLe from experiment results and Equation or symbol description not availableiL=iLt, where Equation or symbol description not availableiLt=zMFDMcMb∕δ, respectively.

Before discussing the results of the particle incorporation model for the RCE of Eq. 16, the two transport parameters Equation or symbol description not availableA and Equation or symbol description not availableB as defined in Eq. 9, which represent the effect of the centrifugal force and the mass transport of the particle, respectively, are discussed. Figure 5 illustrates the magnitude of Equation or symbol description not availableA and Equation or symbol description not availableB as a function of particle radius. For comparison, the lines at the top-right corner in Fig. 6 denote the mass transfer coefficient, Equation or symbol description not availablek, for Equation or symbol description not availableCu2+. The difference between Equation or symbol description not availableB and Equation or symbol description not availablek is due to the difference between the two diffusion boundary layer thicknesses and diffusivities for the reactive ion Equation or symbol description not available(Cu2+) and a particle Equation or symbol description not available(Al2O3), as shown in Table I and II. Thus, the effect of mass transfer for the reactive species is larger than that for the particle. The reason is that a reactive ion has a higher diffusivity than a particle, even if Equation or symbol description not availableδp is less than δ. The ratio Equation or symbol description not availablek∕B is then always larger than unity and approaches a constant value as the particle radius increases. Increasing the speed of the rotating cylinder has a larger effect on Equation or symbol description not availableA than on Equation or symbol description not availableB, because Equation or symbol description not availableA is proportional to Equation or symbol description not availableω2, while Equation or symbol description not availableB is proportional to Equation or symbol description not availableω0.7 by Eq. 5. Below a particle radius of Equation or symbol description not available100–200nm depending on the RCE rotational rate, the effect of mass transfer is more dominant. Above this particle radius, the centrifugal force is more pronounced as Equation or symbol description not availableA and Equation or symbol description not availableB are proportional to Equation or symbol description not availablea2 and Equation or symbol description not availablea−0.644, respectively. Therefore, it is expected that a thin boundary layer and a small particle size would result in an increased particle incorporation in the deposit.

Figure 5. Refer to the following caption and surrounding text.

Figure 5. Comparison of model parameters Equation or symbol description not availableA and Equation or symbol description not availableB as a function of particle radius for various rotational speeds with Equation or symbol description not availableρp=3.965(g∕cm3), Equation or symbol description not availabledi=1.2(cm), Equation or symbol description not availableμ=1(cp), Equation or symbol description not availableρ=1(g∕cm3), and Equation or symbol description not availableT=298K. The thick and thin lines correspond to Equation or symbol description not availableA and Equation or symbol description not availableB, respectively. The lines at the top-right corner show the mass transfer coefficient Equation or symbol description not availablek(=DM∕δ) for Equation or symbol description not availableCu2+.

Figure 6. Refer to the following caption and surrounding text.

Figure 6. (Color online) The ratio of particle volume Equation or symbol description not availableVptot to metal volume Equation or symbol description not availableVmetal in the composites at the rotating speed of Equation or symbol description not available500rpm. The dashed, dotted, and solid lines correspond to a particle loading of Equation or symbol description not available39(g∕L)(○), Equation or symbol description not available120(g∕L)(▿), and Equation or symbol description not available158(g∕L)(▵), respectively, whereas the symbols are experimental data.

Figure 7. Refer to the following caption and surrounding text.

Figure 7. (Color online) The ratio of particle volume Equation or symbol description not availableVptot to metal volume Equation or symbol description not availableVmetal in the composites at the rotating speed of Equation or symbol description not available1000rpm. The dashed, dotted, and solid lines correspond to a particle loading of Equation or symbol description not available39(g∕L)(○), Equation or symbol description not available120(g∕L)(▿), and Equation or symbol description not available158(g∕L)(▵), respectively, whereas the symbols are experimental data.

Figure 8. Refer to the following caption and surrounding text.

Figure 8. (Color online) The ratio of particle volume Equation or symbol description not availableVptot to metal volume Equation or symbol description not availableVmetal in the composites at the rotating speed of Equation or symbol description not available1500rpm. The dashed, dotted, and solid lines corresponds for a particle loading of Equation or symbol description not available39(g∕L)(○), Equation or symbol description not available120(g∕L)(▿), and Equation or symbol description not available158(g∕L)(▵), respectively, whereas the symbols are experimental data.

In order to approximate the surface concentration of particles, Eq. 9 can be rewritten as follows

Equation (22)cs=Bcb+JpA+B

where the order of magnitude of the particle mass flux Equation or symbol description not availableJp is Equation or symbol description not availableO(10−11)mol∕m2s as calculated from Eq. 14 using experimental results,2 which is much less than Equation or symbol description not availableBcb, which is Equation or symbol description not availableO(10−6)mol∕m2s. This is due to the very thin diffusion boundary layer thickness and the low particle mass flux. However, for our case of Equation or symbol description not availableBcb≫Jp, Equation or symbol description not availablecs≈Bcb∕(A+B) and the probability Equation or symbol description not availableP is equal to Equation or symbol description not availableB∕(A+B). The estimated probabilities Equation or symbol description not availableP at each rotational speed shown in Table III are independent of particle loading, and these results were used to predict the volume ratio in Eq. 14. The probability Equation or symbol description not availableP is thus dependent upon the transport parameters of the system, Equation or symbol description not availableA and Equation or symbol description not availableB. With Equation or symbol description not availableB≫A, i.e., the particle radius is very small as shown in Table IV and Fig. 5, the surface concentration is very close to the bulk concentration, i.e., Equation or symbol description not availableP≈1. Thus, nearly all of the particles at the surface become incorporated into the deposit.

Table III. The probability Equation or symbol description not availableP and the theoretical limiting current density for a rotational speed for a particle diameter of Equation or symbol description not available50nm.

Rotational rate (rpm) Equation or symbol description not availableP Equation or symbol description not availableA Equation or symbol description not available(×10−6cm∕s) Equation or symbol description not availableB Equation or symbol description not available(×10−4cm∕s) Equation or symbol description not availableiL Equation or symbol description not available(mA∕cm2)
5000.99420.671.1638
10000.98582.711.8862
15000.97626.102.5082

Table IV. The fitting parameters for the RCE electrocodeposition model.

Particle loading (g/L)Rotational speed (rpm) Equation or symbol description not availableα1 Equation or symbol description not availableα2 Equation or symbol description not availableitr∕iL
395000N/A0.39
10000N/A0.35
15000.21.50.22
1205000.61.50.29
10000.61.50.23
15000.41.60.47
1585000.51.60.34
10000.71.30.32
15000.61.10.33
Average value  Equation or symbol description not available0.5±0.2 Equation or symbol description not available1.4±0.2 Equation or symbol description not available0.33±0.08

Stojak and Talbot2 investigated extensively electrocodeposition using an RCE system, where the deposition conditions ranged from kinetically limited to mass transfer limited regions. Their experimental data were compared with the theoretical prediction from Eq. 16 of the ratio of particle volume to metal volume in the film, Equation or symbol description not availableVptot∕Vmetal. The best fit of the model to the experimental data was made by adjusting the parameters α and Equation or symbol description not availableitr. As discussed, the value for α was chosen to be less than unity when Equation or symbol description not availableinitr; the volume ratio decreases as the current density increases in the mass transfer limited region. Celis et al.19 plotted the weight percentage of incorporated particles versus average current density, where Equation or symbol description not availableitr was the maximum value which represents the transition from the kinetic limited region to the mass transfer limited region. In this study, Equation or symbol description not availableitr was chosen to adjust the magnitude of the volume ratio.

The experimental data of Stojak and Talbot2 were measured at three electrode rotational rates (500, 1000, and Equation or symbol description not available1500rpm) and three particle loadings (39, 120, and Equation or symbol description not available158g∕L). The volume ratio vs the current density normalized by the limiting current density, Equation or symbol description not availableiL (as listed in Table III) for these nine sets of data is compared with the model results in Fig. 6, 7 and 8 for each rotational rate. The fitting parameters are summarized in Table IV. As no data were available at small current densities at a particle loading of Equation or symbol description not available39g∕L at 500 and Equation or symbol description not available1000rpm, Equation or symbol description not availableα2 could not be determined.

As shown in Fig. 6, 7 and 8, the model agrees well with the experimental results for data both below and above the transition current density. For Equation or symbol description not availablecb=39g∕L and a low level of particle incorporation (a volume ratio less than 0.02), the Equation or symbol description not availableα1 equals 0 at low rotational rates and 0.2 at Equation or symbol description not available1500rpm. As the particle loading increased to Equation or symbol description not availablecb=120g∕L and Equation or symbol description not availablecb=158g∕L, the model prediction with Equation or symbol description not availableα=0 shows a large deviation from experimental results. Although the value of Equation or symbol description not availableα1 appears to slightly increase with particle loading (with an average value of 0.5), the values of Equation or symbol description not availableα2 and Equation or symbol description not availableitr∕iL are nearly constant for all the data sets with average values of 1.4 and 0.33, respectively.

The difference between our results and those of the Celis model19 is that the curve for Equation or symbol description not availablein<itr is not a straight line; Equation or symbol description not availableα2 is not equal to 2. The main concept of the Celis model is the competition of diffusion rate between particles and ions across the diffusion layer; the parameter α compensates for the relative diffusion speed of the particles. It was assumed that in the kinetically controlled region the diffusion speed for particles and ions was the same, so that the relationship was linear Equation or symbol description not available(α2=0). However, the experimental data from Stojak and Talbot show that the rate of particle incorporation is larger than that of deposited ions under kinetic control, i.e., Equation or symbol description not availableα2<2. In the mass transfer controlled region the reactive ion deposition rate is faster than that of particles, and the diffusion of particles becomes the controlling step. At a higher current density, this distinction is more obvious. This results in a positive value of Equation or symbol description not availableα1.

Conclusions

A model for electrocodeposition on an RCE was developed to predict the particle incorporation volume ratio which agrees well with experimental data. The asymptotic solution for the concentration profiles for ions within the diffusion boundary layer showed that diffusion of the reactive species in a supporting electrolyte solution is the dominant process to carry the current at the cathodic surface; the contribution of electromigration is negligible. The ratio of the total volume of particles to that of the deposited metal was expressed as a function of two transport parameters, current density, a probability Equation or symbol description not availableP, and two experimental fitting parameters (Equation or symbol description not availableitr and α). The model can be used easily to predict the volume ratio in the kinetically and mass transfer limited region of electrocodeposition with only few experimental data points.

University of California, San Diego, assisted in meeting the publication costs of this article.

List of Symbols

Equation or symbol description not availableA transport parameter for centrifugal force
Equation or symbol description not availablea particle radius, m
Equation or symbol description not availableb1 concentration of Equation or symbol description not availableCuSO4, Equation or symbol description not availablemol∕m3
Equation or symbol description not availableb2 concentration of Equation or symbol description not availableH2SO4, Equation or symbol description not availablemol∕m3
Equation or symbol description not availableB average mass transfer coefficient for particles
Equation or symbol description not availablec concentration of particles, Equation or symbol description not availablemol∕m3
Equation or symbol description not availableci concentration of Equation or symbol description not availablei species, Equation or symbol description not availablemol∕m3
Equation or symbol description not availableCi normalized concentration of Equation or symbol description not availablei species by Equation or symbol description not availableb2
Equation or symbol description not availabled particle diameter, m
Equation or symbol description not availabledi diameter of the inner cylinder, m
Equation or symbol description not availableh0 minimum distance between a particle and an electrode, m
Equation or symbol description not availableD diffusivity, Equation or symbol description not availablem2∕s
Equation or symbol description not availableF Faraday’s constant, Equation or symbol description not available96,485C∕equiv.
Equation or symbol description not availableg gravitation acceleration rate, Equation or symbol description not availablem2∕s
Equation or symbol description not availablein current density, Equation or symbol description not availableA∕m2
Equation or symbol description not availableIn normalized current density by Equation or symbol description not availablez1FD1b1∕δ
Equation or symbol description not availableiL limiting current density, Equation or symbol description not availableA∕m2
Equation or symbol description not availableitr transition current density, Equation or symbol description not availableA∕m2
Equation or symbol description not availableJp mass flux of particles on an electrode surface, Equation or symbol description not availablemol∕m2s
Equation or symbol description not availableJM mass flux of deposited atoms on an electrode surface, Equation or symbol description not availablemol∕m2s
Equation or symbol description not availablek mass transfer coefficient Equation or symbol description not availableDM∕δ, m/s
Equation or symbol description not availablekB Boltzmann’s constant
Equation or symbol description not availablem single particle mass, kg
Equation or symbol description not availablemf mass of the displaced fluid Equation or symbol description not available(=Vpρf), kg
Equation or symbol description not availablen unit normal vector pointing out an electrolyte solution
Equation or symbol description not availablenp number of moles of particles in a deposited film
Equation or symbol description not availablenM number of moles of metal atoms in a deposited film
Equation or symbol description not availableNA Avogadro’s number
Equation or symbol description not availableP probability of particle being at the surface
Equation or symbol description not availableR gas constant, Equation or symbol description not availableJ∕molK
Equation or symbol description not availabler coordinate perpendicular to an electrode surface
Equation or symbol description not availabler1 inner cylinder radius, m
Equation or symbol description not availableRe Reynolds number
Equation or symbol description not availableSc Schmidt number
Equation or symbol description not availableSh Sherwood number
Equation or symbol description not availableT temperature, K
Equation or symbol description not availableu velocity vector of a fluid, m/s
Equation or symbol description not availableV peripheral velocity at Equation or symbol description not availabler=r1, m/s
Equation or symbol description not availableVf fluid velocity at the center of a particle, m/s
Equation or symbol description not availableVm,M molar volume of a deposited film
Equation or symbol description not availableVp volume of a particle, Equation or symbol description not availablem3
Equation or symbol description not availablevi mobility of Equation or symbol description not availablei species defined by Equation or symbol description not availableDi∕RT
Equation or symbol description not availabley coordinate defined as Equation or symbol description not availabler−r1
Equation or symbol description not availablez charge number

Greek letters

αmodel fitting parameter
Equation or symbol description not availableα1 model fitting parameter when Equation or symbol description not availableinitr
Equation or symbol description not availableα2 model fitting parameter when Equation or symbol description not availablein<itr
βfriction coefficient Equation or symbol description not available(=6πμa)
δionic boundary layer thickness, m
Equation or symbol description not availableδ0 viscous sublayer boundary layer thickness, m
Equation or symbol description not availableδp boundary layer thickness for particles, m
ε Equation or symbol description not availableb1∕b2
Equation or symbol description not availableρM density of the cathodic deposit, Equation or symbol description not availablekg∕m3
Equation or symbol description not availableρp density of a particle, Equation or symbol description not availablekg∕m3
Equation or symbol description not availableρf density of the electrolyte solution, Equation or symbol description not availablekg∕m3
μviscosity of the electrolyte, Equation or symbol description not availablekgm∕s
νkinematic viscosity of the electrolyte, Equation or symbol description not availablem2∕s
ωelectrode rotational rate, rad/s
ϕelectrical potential, Equation or symbol description not availableV
Φnormalized electrical potential by Equation or symbol description not availableRT∕F
ξnormalized coordinate Equation or symbol description not available=y∕δ
Equation or symbol description not availablexv volume fraction of inert particles incorporated in a composite film

Subscripts

Equation or symbol description not availableb bulk solution
Equation or symbol description not availableM deposited species
Equation or symbol description not availablep particle
Equation or symbol description not availables electrode surface

: Appendix

The Nernst-Planck equation for the molar flux of the Equation or symbol description not availableith species is

Equation (A-1)Ji=−viziFci∇ϕ−Di∇ci+ciu

where Equation or symbol description not availablevi=Di∕RT is the mobility, Equation or symbol description not availablezi is the charge number, Equation or symbol description not availableF is Faraday’s constant, Equation or symbol description not availableci is concentration, Equation or symbol description not availableDi is diffusivity, Equation or symbol description not availableR is the gas constant, Equation or symbol description not availableT is the absolute temperature, Equation or symbol description not availableu is the velocity field of the fluid, and ϕ is the electrostatic potential function.29 The first, second, and third terms on the right-hand side of Eq. 1 are the contributions of the flux due to migration in an electric field, the diffusion due to a concentration gradient, and convection from the velocity of the fluid, respectively. For convenience, the subscripts Equation or symbol description not availablei=1,2,3 correspond to Equation or symbol description not availableCu2+, Equation or symbol description not availableH+, and Equation or symbol description not availableHSO4−, respectively. For an acidic copper sulfate plating bath, Equation or symbol description not availableb1=[CuSO4] and Equation or symbol description not availableb2=[H2SO4], such that Equation or symbol description not availablecCu2+=c1=b1, Equation or symbol description not availablecH+=c2=b2−b1, and Equation or symbol description not availablecHSO4−=c3=b2+b1 in the bulk region, where Equation or symbol description not availableb2 is much larger than Equation or symbol description not availableb1. An electrolytic solution is governed by the electroneutrality condition

Equation (A-2)∑zici=0

except in the thin regions near charged boundaries. The mass balance for each species with no reaction requires

Equation (A-3)∂ci∂t=−∇∙Jior∂ci∂t+u∙∇ci=viziF∇(ci∇ϕ)+Di∇2ci

where the fluid is incompressible. Because the boundary layer thickness δ is usually much smaller than the inner cylinder radius Equation or symbol description not availabler1, the distribution of ionic concentrations only in the direction of Equation or symbol description not availabley in rectangular coordinates is considered. When it is assumed that the fluid flows perpendicular to the Equation or symbol description not availabley−z plane, the convection term on the left-hand side of Eq. 3 vanishes for the two-dimensional analysis.

The following dimensionless variables are introduced to simplify the governing equation Eq. 3

Equation (A-4)Φ=ϕRT∕F,Ci=cib2,ξ=yδ,andIn=inz1FD1b1∕δ

The governing equations at steady state can be written as

Equation (A-5)ziddξ(CidΦdξ)+d2Cidξ2=0and∑ziCi=0

The boundary conditions on the surface of an electrode Equation or symbol description not available(ξ=0) are

Equation (A-6)In=−z1C1dΦdξ−dC1dξand0=ziCidΦdξ+dCidξfori=2,3

and at the diffusion boundary layer Equation or symbol description not available(ξ=1)

Equation (A-7)C1=ε,C2=1−ε,andC3=1+ε

where ε is a small parameter defined by Equation or symbol description not availableb1∕b2 and is assumed to be much less than 1 Equation or symbol description not available(ε<1). We approximate the solutions of Equation or symbol description not availableCi and Φ in a series expansion as

Equation (A-8)Ci=Ci0+εCi1+ε2Ci2+⋯,andΦ=Φ0+εΦ1+ε2Φ2+⋯

where the subscripts Equation or symbol description not availablek(=0,1,2,3,…) denote the Equation or symbol description not availablekth order of magnitude. Note that Equation or symbol description not availableC10=0 for the expansion of Equation or symbol description not availablei=1. By substituting these relations into Eq. A-5 through A-7, we obtain

Equation (A-9)ziddξ((Ci0+εCi1+⋯)d(Φ0+εΦ1+⋯)dξ)+d2(Ci0+εCi1+⋯)dξ2=0

Equation (A-10)∑zi(Ci0+εCi1+ε2Ci2+⋯)=0

The equations of order of magnitude of Equation or symbol description not availableO(1) from Eq. 5 are

Equation (A-11)ziddξ(Ci0dΦ0dξ)+d2Ci0dξ2=0,fori=2,3,and∑ziCi0=0

The boundary conditions at Equation or symbol description not availableξ=0 are

Equation (A-12)0=ziCi0dΦ0dξ+dCi0dξ,fori=2,3

and at the boundary layer Equation or symbol description not available(ξ=1)

Equation (A-13)C10=0,C20=1,C30=1

The solutions for Equation or symbol description not availableO(1) can be easily obtained by

Equation (A-14)C10=0,C20=1,C30=1,dΦ0∕dξ=0

The first-order equations to Equation or symbol description not availableO(ε) from Eq. 5 are

Equation (A-15)ziddξ(Ci0dΦ1dξ)+d2Ci1dξ2=0and∑ziCi1=0

The boundary conditions at Equation or symbol description not availableξ=0 are

Equation (A-16)In=−dC11dξand0=ziCi0dΦ1dξ+dCi1dξfori=2,3

where the flux of Equation or symbol description not availableC1 species is balanced by the applied current density and the fluxes of Equation or symbol description not availableC2 and Equation or symbol description not availableC3 are zero due to a no-penetration condition. Because Equation or symbol description not availableC10 is zero, Equation or symbol description not availableC11 is linear from Eq. 15. At the boundary layer Equation or symbol description not available(ξ=1), the concentrations are those of the bulk

Equation (A-17)C11=1,C21=−1,C31=1

The solution for Equation or symbol description not availableC11 is obtained using the boundary conditions of Eq. 16 and Eq. 17. The equation for the electrostatic potential Equation or symbol description not availableΦ1 is derived by summing Eq. 15 for all species and using the electroneutrality condition, i.e., Equation or symbol description not availabled2Φ1∕dξ2=0. One integration gives Equation or symbol description not availabledΦ1∕dξ=λ1, where Equation or symbol description not availableλ1 is an arbitrary constant and is determined from the electroneutrality condition. The solutions for Equation or symbol description not availableC21 and Equation or symbol description not availableC31 are obtained from Eq. 11 using the boundary conditions of Eq. 16, 17. The first-order solutions by simple algebra are

Equation (A-18)C11(ξ)=In(1−ξ)+1

Equation (A-19)C21(ξ)=z2λ1(1−ξ)−1

Equation (A-20)C31(ξ)=z3λ1(1−ξ)+1

Equation (A-21)dΦ1dξ=λ1

Because Equation or symbol description not available∑ziCi1=0 from Eq. 15, Equation or symbol description not availablez1[In(1−ξ)+1]+z2[z2λ1(1−ξ)−1]+z3[z3λ1(1−ξ)+1]=0.

At Equation or symbol description not availableξ=1, Equation or symbol description not availablez1−z2+z3=0. Substituting Equation or symbol description not availableξ=0 and arranging the resultant equation, the arbitrary constant Equation or symbol description not availableλ1 is then given by

Equation (A-22)λ1=−z1z22+z32In

In a similar way, the second-order equations to Equation or symbol description not availableO(ε2) are determined from Eq. 5

Equation (A-23)zi[ddξ(Ci0dΦ2dξ)+ddξ(Ci1dΦ1dξ)]+d2Ci2dξ2=0and∑ziCi2=0

The boundary conditions at Equation or symbol description not availableξ=0 are

Equation (A-24)0=zi(Ci0dΦ2dξ+Ci1dΦ1dξ)+dCi2dξfori=1,2,3

and at Equation or symbol description not availableξ=1

Equation (A-25)C11=0,C21=0,C31=0

The procedure to obtain the solutions is the same as that of the first-order case. The second-order solutions are

Equation (A-26)C12(ξ)=12z1λ1In(ξ2−1)−z1λ1(In+1)(ξ−1)

Equation (A-27)C22(ξ)=12λ3(ξ2−1)−λ6(ξ−1)

Equation (A-28)C32(ξ)=12λ4(ξ2−1)−λ7(ξ−1)

Equation (A-29)dΦ2dξ=λ2ξ+λ5

where

Equation (30a)λ2=z12λ1In+λ12(z23+z33)z22+z32,λ3=z22λ12−z2λ2,λ4=z32λ12−z3λ2

Equation (30b)λ5=−z1In[z1In(z23+z33)+(z22+z32)(z12+z22−z32)](z22+z32)2

Equation (30c)λ6=z1λ1(z1λ1−1)+z2λ5,andλ7=z3λ1(z3λ1+1)+z3λ5

Therefore, the solutions up to the second order of magnitude are

Equation (A-31)C1=ε[In(1−ξ)+1]+ε2[12z1λ1In(ξ2−1)−z1λ1(In+1)(ξ−1)]

Equation (A-32)C2=1+ε[z2λ1(1−ξ)−1]+ε2[12λ3(ξ2−1)−λ6(ξ−1)]

Equation (A-33)C3=1+ε[z3λ1(1−ξ)+1]+ε2[12λ4(ξ2−1)−λ7(ξ−1)]

Equation (A-34)dΦdξ=ελ1+ε2[λ2ξ+λ5]

Therefore, the solutions of Equation or symbol description not availableci and Equation or symbol description not availabledϕ∕dy can be rewritten by the dimensionless variables defined by Eq. 4.

Please wait… references are loading.
10.1149/1.2398819