Question
Question: The gap between the plates of a parallel-plate capacitor is filled with isotropic dielectric whose p...
The gap between the plates of a parallel-plate capacitor is filled with isotropic dielectric whose permittivity ϵ varies linearly from ϵ1 to ϵ2 (ϵ2>ϵ1) in the direction perpendicular to the plates. The area of each plate equals S, the separation between the plates is equal to d. Find
a) the capacitance of the capacitor;
b) the space density of the bound charges as a function of ϵ if the charge of the capacitor is q and the field E in it is directed toward

a) C=dln(ϵ1ϵ2)S(ϵ2−ϵ1), b) ρb(ϵ)=−Sdϵ2qϵ0(ϵ2−ϵ1)
Solution
The problem involves a parallel-plate capacitor with a non-uniform dielectric. The permittivity varies linearly between the plates.
a) Capacitance of the capacitor: Let the plates be parallel to the xy-plane, located at z=0 and z=d. Let the area of each plate be S. The permittivity ϵ varies linearly from ϵ1 at one plate (say z=0) to ϵ2 at the other plate (z=d). The linear variation of permittivity with position z is given by: ϵ(z)=ϵ1+dϵ2−ϵ1z. We can consider the capacitor as being composed of infinitesimally thin layers of thickness dz, stacked in series along the z-direction. Each layer at position z has an area S and permittivity ϵ(z). The capacitance of such a thin layer is dC=dzϵ(z)S. Since these layers are in series, the reciprocal of the total capacitance C is the sum of the reciprocals of the capacitances of the layers: C1=∫0ddC1=∫0dϵ(z)Sdz. Substitute the expression for ϵ(z): C1=∫0d(ϵ1+dϵ2−ϵ1z)Sdz=S1∫0dϵ1+dϵ2−ϵ1zdz. Let u=ϵ1+dϵ2−ϵ1z. Then du=dϵ2−ϵ1dz, so dz=ϵ2−ϵ1ddu. When z=0, u=ϵ1. When z=d, u=ϵ1+dϵ2−ϵ1d=ϵ1+ϵ2−ϵ1=ϵ2. The integral becomes: C1=S1∫ϵ1ϵ2u1ϵ2−ϵ1ddu=S(ϵ2−ϵ1)d∫ϵ1ϵ2u1du. C1=S(ϵ2−ϵ1)d[ln∣u∣]ϵ1ϵ2=S(ϵ2−ϵ1)d(lnϵ2−lnϵ1)=S(ϵ2−ϵ1)dln(ϵ1ϵ2). The capacitance is C=dln(ϵ1ϵ2)S(ϵ2−ϵ1). This formula assumes ϵ1 and ϵ2 are absolute permittivities. If they were relative permittivities, the absolute permittivity would be ϵ(z)=ϵ0ϵr(z), and the capacitance would be C=dln(ϵ1ϵ2)ϵ0S(ϵ2−ϵ1). Based on the phrasing "permittivity ϵ", we assume ϵ1,ϵ2 are absolute permittivities.
b) Space density of bound charges: The space density of bound charges ρb is related to the polarization P by ρb=−∇⋅P. In a linear, isotropic dielectric, the polarization is related to the electric field E by P=ϵ0χeE=ϵ0(ϵr−1)E=(ϵ−ϵ0)E, where χe is the electric susceptibility and ϵr=ϵ/ϵ0 is the relative permittivity. The electric field E inside the parallel-plate capacitor is directed perpendicular to the plates. Let the field be in the +z direction (from the plate with charge +q to the plate with charge -q). The problem states the field is directed "toward", which is ambiguous. Let's assume the standard convention where the field points from positive to negative charge, so if the charge is q, the field is directed from the positive plate to the negative plate. Let the positive plate be at z=0 and the negative plate at z=d. Then E=E(z)z^. The displacement field D is uniform within the dielectric and is related to the free charge density σf=q/S on the plates by D=σf=q/S. The relationship between D and E is D=ϵE. So, E(z)=D/ϵ(z)=ϵ(z)q/S. E(z)=Sϵ(z)qz^. The polarization is P(z)=(ϵ(z)−ϵ0)E(z)=(ϵ(z)−ϵ0)Sϵ(z)qz^=Sq(1−ϵ(z)ϵ0)z^. The space density of bound charges is ρb=−∇⋅P. Since P is in the z-direction and depends only on z, the divergence is ∂z∂Pz: ρb=−∂z∂[Sq(1−ϵ(z)ϵ0)]=−Sq∂z∂(1−ϵ0ϵ(z)−1). ρb=−Sq(−ϵ0)(−1)ϵ(z)−2dzdϵ=−S[ϵ(z)]2qϵ0dzdϵ. From the linear variation ϵ(z)=ϵ1+dϵ2−ϵ1z, we have dzdϵ=dϵ2−ϵ1. So, ρb(z)=−S[ϵ(z)]2qϵ0dϵ2−ϵ1. We are asked for the space density as a function of ϵ. Since ϵ varies linearly with z, z can be expressed in terms of ϵ: z=ϵ2−ϵ1d(ϵ−ϵ1). So specifying z is equivalent to specifying ϵ. The expression for ρb already has ϵ(z) in the denominator. We can just write ϵ instead of ϵ(z) to indicate the local value of permittivity. ρb(ϵ)=−Sdϵ2qϵ0(ϵ2−ϵ1). The direction of E is assumed to be from z=0 to z=d. If the field were directed toward the other plate, the sign of E would be reversed, but the sign of q would also be reversed (assuming q is the charge on the plate where the field originates), leaving the expression for ρb unchanged.
Final check: The total bound charge qb=∫ρbdV=∫0dρb(z)Sdz=∫0d−S[ϵ(z)]2qϵ0dϵ2−ϵ1Sdz. qb=−dqϵ0(ϵ2−ϵ1)∫0d[ϵ1+dϵ2−ϵ1z]2dz. Let u=ϵ1+dϵ2−ϵ1z, dz=ϵ2−ϵ1ddu. qb=−dqϵ0(ϵ2−ϵ1)∫ϵ1ϵ2u21ϵ2−ϵ1ddu=−qϵ0∫ϵ1ϵ2u−2du. qb=−qϵ0[−u−1]ϵ1ϵ2=−qϵ0(−ϵ21+ϵ11)=qϵ0(ϵ21−ϵ11)=qϵ0ϵ1ϵ2ϵ1−ϵ2. The bound surface charge density on the plates are σb1=P(0)⋅(−z^) and σb2=P(d)⋅(z^). P(z)=Sq(1−ϵ(z)ϵ0)z^. σb1=Sq(1−ϵ1ϵ0)(−1)=−Sq(1−ϵ1ϵ0). σb2=Sq(1−ϵ2ϵ0). The total bound charge is Qb=qb+σb1S+σb2S. Qb=qϵ0ϵ1ϵ2ϵ1−ϵ2−q(1−ϵ1ϵ0)+q(1−ϵ2ϵ0). Qb=qϵ0ϵ1ϵ2ϵ1−ϵ2−q+ϵ1qϵ0+q−ϵ2qϵ0=qϵ0ϵ1ϵ2ϵ1−ϵ2+qϵ0(ϵ11−ϵ21). Qb=qϵ0ϵ1ϵ2ϵ1−ϵ2+qϵ0ϵ1ϵ2ϵ2−ϵ1=0. The total bound charge in the volume and on the surfaces must sum to zero for a neutral dielectric. This confirms the expression for ρb.