betaFeedback

WorksPhysics and Astronomy

The Electric Force Between Two Straight Parallel Resistive Wires Carrying DC-Currents in the Asymptotic Limit of Infinitely Thin Wires

Abstract

During the years 1948-2019 the ampere was defined via the magnetic force between two long thin parallel wires carrying stationary current. However, if a stationary current flows through a resistive wire, static electric charges appear on the surface of the wire, and this will lead to an additional electric force between the wires. This article discusses the ratio of electric over magnetic forces in the asymptotic limit of infinitely thin wires, which is not accessible by numerical methods. The electric force between the two wires depends also on the choice of the common ground node. For extremely thin or extremely long resistive wires the electric force dominates over the magnetic one.

Introduction

In the International System of Units (SI) electric current is measured in the unit of ampere (A). Until recently (2019) the ampere was defined via the magnetic force between two current carrying wires[1, 2, 3, 4, 5, 6]:

The ampere is that constant current which, if maintained in two straight parallel conductors of infinite length, of negligible circular cross-section, and placed one metre apart in vacuum, would produce between these conductors a force equal to 2×10−72 \times10^{-7} newtons per metre of length.

{#eq:1}

Reading this definition one naturally assumes that we only have to make the wire long and thin enough to get a current measurement with arbitrarily high precision. However, for resistive wires this is utterly wrong! In this article I will show that the longer and the thinner the wire becomes the more this definition of the ampere is corrupted by an additional electric force between the wires. Yet, this paper does not intend to question the accuracy of state-of-the-art current measurements, which use different and more sophisticated geometries than in (1), see Refs. 7–9. Conversely, I intend to show that, without any additional counter-measures against electric forces (such as two electrical shields), the simple geometry of two straight parallel wires like in (1) gives infinite error in the asymptotic limit of infinitely thin resistive wires. This was already shown in Ref. 10, however, there the authors neglected the infinite charges at the ends of the wires, whereas this article gives a rigorous theory for wires of finite length in the limit of vanishing diameter. I do not know why electric forces were not mentioned in (1), although it is well known that a DC current through a conductor generates an electric field inside and outside the conductor[11, 12, 13, 14, 15, 16, 17]. One reason might be a lack in awareness of surface charges in electric circuits.

Current Flow Through A Prolate Ellipsoid

In this paper I model the wires as prolate ellipsoids with DC-currents flowing along their long axes. Although ellipsoidal coordinates have been used to study electrostatic charging of thin wires[18, 19, 20], no one seems to have used them to study stationary current flow in thin wires of finite length. The ellipsoidal model allows for a rigorous mathematical treatment, and it accurately models cylindrical wires in the asymptotic limit of infinite slimness.

Figure 1(a) shows a conductor in the shape of a prolate ellipsoid with the long axis of length 2c2c along the zz-axis. The conductor volume is parametrized by 0≤η≤η00 \le\eta\le\eta_0 with arbitrary θ\theta and ψ\psi in an ellipsoidal reference frame (η,θ,ψ)(\eta,\theta,\psi) defined by

x=asinh⁡(η)sin⁡(θ)cos⁡(ψ),y=asinh⁡(η)sin⁡(θ)sin⁡(ψ),z=acosh⁡(η)cos⁡(θ),(1)x=a\sinh(\eta)\sin(\theta)\cos(\psi),\qquad y=a\sinh(\eta)\sin(\theta)\sin(\psi),\qquad z=a\cosh(\eta)\cos(\theta), \tag*{(1)}

(x,y,z)(x,y,z) being the Cartesian coordinates. The ellipsoid has rotational symmetry—its cross-sections orthogonal to the zz-axis are circles. The largest circle is in z=0z=0, which is equivalent to θ=π/2\theta=\pi/2, with the diameter 2b:=2asinh⁡(η0)2b:=2a\sinh(\eta_0). The length of the ellipsoid is 2c:=2acosh⁡(η0)2c := 2a\cosh(\eta_0). Thus, the aspect ratio (slimness) of the ellipsoid is c/b=coth⁡(η0)c/b = \coth(\eta_0). It becomes infinite for η0→0\eta_0 \to0, c/b→∞c/b \to\infty. Then the ellipsoid degenerates to a straight line of length 2a2a with its ends being the foci of the ellipsoid in (x,y,z)=(0,0,±a)(x,y,z) = (0,0,\pm a). I call this the thin wire limit.

The Potential Inside the Ellipsoid and the Charge on the Contacts

Current flows between the hyperboloid coordinate surfaces θ=θin\theta= \theta_{\mathrm{in}} and θ=θex\theta= \theta_{\mathrm{ex}} for η<η0\eta< \eta_0 and arbitrary ψ\psi. There, the potential is constant (Dirichlet boundary condition). The lateral surface of the truncated ellipsoid η=η0\eta= \eta_0, θin≤θ≤θex\theta_{\mathrm{in}} \le\theta\le\theta_{\mathrm{ex}}, arbitrary ψ\psi, is insulating. There, the normal derivative of the potential vanishes (Neumann boundary condition). All boundary conditions do not depend on η\eta and ψ\psi, therefore the same applies to the potential, ϕinside(θ)\phi_{\mathrm{inside}}(\theta). For stationary current flow with uniform scalar conductivity κ\kappa the potential satisfies Laplace’s differential equation in ellipsoidal coordinates with the solution[21]

ϕinside(θ)=c1Q0(cos⁡(θ))+c2.(2)\phi_{\mathrm{inside}}(\theta) = c_1 Q_0(\cos(\theta)) + c_2. \tag*{(2)}

Q0(cos⁡(θ))=ln⁡(cot⁡(θ/2))Q_0(\cos(\theta)) = \ln(\cot(\theta/2)) is a Legendre function of the second kind and of order zero. We define ground potential in θ=π/2\theta= \pi/2, which means c2=0c_2 = 0. The constant c1c_1 is determined by the current I0I_0 flowing through the ellipsoid (Appendix A).

For the potential close to the zz-axis we write r=drr = \mathrm{d}r with ∣dr∣≪b|\mathrm{d}r| \ll b. From (1) we know r=asinh⁡(η)sin⁡(θ)r = a\sinh(\eta)\sin(\theta). Therefore small rr means either η→0\eta\to0 or θ→0\theta\to0 or θ→π\theta\to\pi. Let us consider the case of ’small θ\theta’. There we have the full ellipsoid of Figure 1(b), and the contacts are the lines a≤∣z∣≤ca \le|z| \le c in the shape of infinitely thin needles of infinite conductivity. They are on the zz-axis, pushed into the ellipsoid until the tips of the needles reach the focus points. We enter θ=dθ\theta= \mathrm{d}\theta with ∣dθ∣≪1|\mathrm{d}\theta| \ll1 into (1),

dr=asinh⁡(η)dθ,z=acosh⁡(η)⇒dr=z2−a2 dθwith a≤z≤c.(3)\mathrm{d}r = a\sinh(\eta)\mathrm{d}\theta,\qquad z = a\cosh(\eta)\quad\Rightarrow\quad\mathrm{d}r = \sqrt{z^2-a^2}\,\mathrm{d}\theta\quad\text{with }a \le z \le c. \tag*{(3)}

Inserting this into (2) gives for a≤z≤ca \le z \le c

ϕinside(dr,z)=I04πκ(c−a)ln⁡(4z2−a2(dr)2)+c2+O(dr),Ez(r=0,z)=−I0z2πκ(c−a)1z2−a2,Er(dr,z)=I02πκ(c−a)1dr+O(dr).(4)\begin{aligned} \phi_{\mathrm{inside}}(\mathrm{d}r,z) &= \frac{I_0}{4\pi\kappa(c-a)}\ln\left(4\frac{z^2-a^2}{(\mathrm{d}r)^2}\right)+c_2+O(\mathrm{d}r),\\ E_z(r=0,z) &= -\frac{I_0z}{2\pi\kappa(c-a)}\frac{1}{z^2-a^2},\\ E_r(\mathrm{d}r,z) &= \frac{I_0}{2\pi\kappa(c-a)}\frac{1}{\mathrm{d}r}+O(\mathrm{d}r). \tag*{(4)} \end{aligned}

Applying Gauss’s electric flux theorem[22] to ErE_r on the contact gives the charge on the contact

qcont=ϵ0I0/κ.(5)q_{\mathrm{cont}} = \epsilon_0 I_0/\kappa. \tag*{(5)}

With ϵ0=8.854×10−12\epsilon_0 = 8.854 \times10^{-12} Vs/Am, I0=1.06I_0 = 1.06 A, κ=5.88×107\kappa= 5.88 \times10^7 S/m (conductivity of copper), and the elementary charge e=1.6×10−19e = 1.6 \times10^{-19} As we find that there is only a single electron on the contact, regardless if the ellipsoid is very slim or not! Interestingly, the same tiny amount of surface charge qcontq_{\mathrm{cont}} sits in a 90° bend in an ordinary wire with conductivity κ\kappa to guide the current I0I_0 around the corner[17]. Hence, a very small charge has a huge effect on the local electric field.

Irrespective of the slimness of the ellipsoid the potential at the contact of current injection in (4) is infinite, because the contact has zero diameter (it is a line). Contacts with infinite contact resistance are not unusual in electrical engineering—they occur for example in Van der Pauw’s measurement of the sheet resistance of thin plates with point or line contacts[23] and they also occur in Hall plates with point or line contacts[24]. If current is injected in such a contact, its potential rises unboundedly with a logarithmic singularity. A more surprising finding is that in (4) the electric field EzE_z on the contact and parallel to the contact does not vanish. This is a consequence of the fact that the potential at the contact is infinite, which means that it takes infinite energy to shuffle a small test charge from infinity onto the contact. Conversely, the electric field along the contact is only finite, and therefore we need only finite energy to shift this test charge along the contact. The finite energy needed to shift a test charge along the contact is still negligible against the infinite energy to bring it onto the contact, and in this asymptotic sense the contact is still an iso-potential surface. The radial field ErE_r on the entire contact is infinite, see (4), and therefore the field lines of EE are perfectly perpendicular to the contact—in this respect the contact still behaves as we expect it from an ideal contact. Note that on the one hand the conductivity of the line contact is infinite, on the other hand its thickness is zero, and this may serve as an explanation for the line contact not being able to force zero tangential electric field. Note also that we cannot study this phenomenon by numerical simulation techniques like finite elements (FEM), because these programs cannot handle infinite quantities. They cannot represent infinite radial field ErE_r on the line contact and therefore erroneously they assume zero tangential field Ez=0E_z = 0 to arrive at correct E-field lines orthogonal to the contacts.

The same phenomenon occurs in electrostatics, if we charge up an infinitely thin straight wire of finite length[20]: The capacitance of such an ideal needle vanishes, see (21). Hence, it takes infinite energy to charge it. The charges distribute homogeneously on the ideal needle, yet they seem not to be in equilibrium, because there is a finite electric field acting on them in the direction of the needle. Again here the finite energy needed to shift a test charge along the ideal needle is negligible against the infinite energy needed to bring it onto the ideal needle, and the field lines are perfectly orthogonal to the ideal needle. Also here, standard FEM codes assume isopotential along the needle, thereby failing to predict the finite tangential electric field on the needle.

The Potential Outside the Ellipsoid and the Charge on the Ellipsoid

The general ansatz for the potential outside the full ellipsoid of Figure 1(b) is

Figure 1. (a) Cross-section of a prolate ellipsoid with its surface in η=η0\eta=\eta_0. The zz-axis is the axis of rotational symmetry. The contact of current input is in θ=θin\theta=\theta_{\mathrm{in}}, and the contact of current output is in θ=θex\theta=\theta_{\mathrm{ex}}. The current flows through the darker orange region. The focus points are on the zz-axis in z=±az=\pm a. The vertex points are on the zz-axis in z=±cz=\pm c. The long axis has length 2c2c and the maximum diameter is 2b2b, with c2=a2+b2c^2=a^2+b^2. (b) shows the ellipsoid with line contacts in θ=0\theta=0 (for current input) and in θ=π\theta=\pi (for current output). The contacts reach from each focus point to the closer vertex point.

ϕoutside(η,θ)=∑n=0∞(AnPn(cos⁡(θ))+BnQn(cos⁡(θ)))Qn(cosh⁡(η)).(6)\phi_{\mathrm{outside}}(\eta,\theta) = \sum_{n=0}^{\infty} \left(A_nP_n(\cos(\theta)) + B_nQ_n(\cos(\theta))\right)Q_n(\cosh(\eta)). \tag*{(6)}

In (6) we have to discard the terms Qn(cos⁡(θ))Q_n(\cos(\theta)), because they are singular in θ=0\theta= 0 and θ=π\theta= \pi. This means Bn=0B_n = 0. Indeed, the potential is singular in θ=0\theta= 0 and θ=π\theta= \pi, but only for η=η0\eta= \eta_0, whereas it is regular for η>η0\eta> \eta_0. Continuity of the potential at the surface η=η0\eta= \eta_0 means ϕinside(θ)=ϕoutside(η0,θ)\phi_{\mathrm{inside}}(\theta) = \phi_{\mathrm{outside}}(\eta_0,\theta). Splitting up this identity into even and odd functions of θ\theta gives

0=∑n=0∞A2nQ2n(cosh⁡(η0))P2n(cos⁡(θ)),0 = \sum_{n=0}^{\infty} A_{2n}Q_{2n}(\cosh(\eta_0))P_{2n}(\cos(\theta)),
c1Q0(cos⁡(θ))=∑n=0∞A2n+1Q2n+1(cosh⁡(η0))P2n+1(cos⁡(θ)).(7)c_1Q_0(\cos(\theta)) = \sum_{n=0}^{\infty} A_{2n+1}Q_{2n+1}(\cosh(\eta_0))P_{2n+1}(\cos(\theta)). \tag*{(7)}

Thus, A2n=0A_{2n} = 0. For A2n+1A_{2n+1} we multiply both sides with P2m+1(cos⁡(θ))sin⁡(θ)P_{2m+1}(\cos(\theta))\sin(\theta) and integrate over θ:0→π\theta: 0 \to\pi. Note that Pm(x),Pn(x)P_m(x), P_n(x) are orthogonal[25], ∫−11Pm(x)Pn(x) dx=2δm,n/(2n+1)\int_{-1}^{1} P_m(x)P_n(x)\,\mathrm{d}x = 2\delta_{m,n}/(2n + 1), but Pm(x),Qn(x)P_m(x), Q_n(x) are not orthogonal. In Appendix B we prove

∫0πQ0(cos⁡(θ))P2n+1(cos⁡(θ))sin⁡(θ) dθ=∫−11Q0(x)P2n+1(x) dx=1(n+1)(2n+1).(8)\int_{0}^{\pi} Q_0(\cos(\theta))P_{2n+1}(\cos(\theta))\sin(\theta)\,\mathrm{d}\theta= \int_{-1}^{1} Q_0(x)P_{2n+1}(x)\,\mathrm{d}x = \frac{1}{(n+1)(2n+1)}. \tag*{(8)}

Using this in (7) finally gives

A2n+1=c14n+32(n+1)(2n+1)Q2n+1(cosh⁡(η0))>0.(9)A_{2n+1} = c_1 \frac{4n+3}{2(n+1)(2n+1)Q_{2n+1}(\cosh(\eta_0))} > 0. \tag*{(9)}

The electric field component perpendicular to the surface of the conductor is[27]

E⊥,outside=Eoutside⋅nη=−1a(sinh⁡(η0))2+(sin⁡(θ))2dϕoutsidedη∣η0=−1a(sinh⁡(η0))2+(sin⁡(θ))2∑n=0∞AnPn(cos⁡(θ))ddη0Qn(cosh⁡(η0)).(10)\begin{aligned} E_{\perp,\mathrm{outside}} = \mathbf{E}_{\mathrm{outside}} \cdot\mathbf{n}_{\eta} = \left.\frac{-1}{a\sqrt{(\sinh(\eta_0))^2 + (\sin(\theta))^2}}\frac{\mathrm{d}\phi_{\mathrm{outside}}}{\mathrm{d}\eta}\right|_{\eta_0} \\ &= \frac{-1}{a\sqrt{(\sinh(\eta_0))^2 + (\sin(\theta))^2}}\sum_{n=0}^{\infty} A_nP_n(\cos(\theta))\frac{\mathrm{d}}{\mathrm{d}\eta_0}Q_n(\cosh(\eta_0)). \tag*{(10)} \end{aligned}

where nη\mathbf{n}_{\eta} is the unit vector in the direction of growing η\eta (with θ\theta and ψ\psi staying constant). In (10) we can further use

ddxQn(x)=n+1x2−1(−xQn(x)+Qn+1(x)).(11)\frac{\mathrm{d}}{\mathrm{d}x}Q_n(x) = \frac{n+1}{x^2-1}\left(-xQ_n(x) + Q_{n+1}(x)\right). \tag*{(11)}

Conversely, E⊥,inside=0E_{\perp,\mathrm{inside}} = 0, because the inside potential does not depend on η\eta, see (2). Therefore the charge density σq\sigma_q on the surface of the ellipsoid is equal to ϵ0E⊥,outside\epsilon_0 E_{\perp,\mathrm{outside}} (Ref. 22).

The total charge dqelldq_{\mathrm{ell}} on the surface of the ellipsoid in a ring of width dθd\theta at position θ\theta is

dqell=ϵ0E⊥,outside 2πasinh⁡(η0)sin⁡(θ) a(sinh⁡(η0))2+(sin⁡(θ))2∣dθ∣=ϵ0I0asin⁡(θ)∣dθ∣κ(c−a)∑n=0∞4n+32n+1P2n+1(cos⁡(θ))(cosh⁡(η0)−Q2n+2(cosh⁡(η0))Q2n+1(cosh⁡(η0))),(12)\begin{aligned} dq_{\mathrm{ell}} &= \epsilon_0 E_{\perp,\mathrm{outside}}\,2\pi a\sinh(\eta_0)\sin(\theta)\,a\sqrt{(\sinh(\eta_0))^2+(\sin(\theta))^2}\lvert d\theta\rvert\\ &= \frac{\epsilon_0 I_0a\sin(\theta)\lvert d\theta\rvert}{\kappa(c-a)}\sum_{n=0}^{\infty}\frac{4n+3}{2n+1}P_{2n+1}(\cos(\theta))\left(\cosh(\eta_0)-\frac{Q_{2n+2}(\cosh(\eta_0))}{Q_{2n+1}(\cosh(\eta_0))}\right), \tag*{(12)} \end{aligned}

where p(θ)p(\theta) is the perimeter of the ring at position θ\theta, and dwdw is the width of the ring in the direction tangentially to the surface of the ellipse. If the slimness of the ellipsoid gets infinite, it means η0→0\eta_0\to0 and dz→−asin⁡(θ)dθdz\to-a\sin(\theta)d\theta and cosh⁡(η0)=c/a→1\cosh(\eta_0)=c/a\to1 and cos⁡(θ)=z/c\cos(\theta)=z/c. In Appendix C we prove

lim⁡η0→0(cosh⁡(η0)−Q2n+2(cosh⁡(η0))Q2n+1(cosh⁡(η0)))=−1(n+1)ln⁡(cosh⁡(η0)−1).(13)\lim_{\eta_0\to0}\left(\cosh(\eta_0)-\frac{Q_{2n+2}(\cosh(\eta_0))}{Q_{2n+1}(\cosh(\eta_0))}\right)=\frac{-1}{(n+1)\ln(\cosh(\eta_0)-1)}. \tag*{(13)}

With λell=dqell/dz\lambda_{\mathrm{ell}}=dq_{\mathrm{ell}}/dz, we get for the line charge density of an infinitely slim ellipsoid

lim⁡b→0λell=−ϵ0I0κ(c−a)1ln⁡(c/a−1)∑n=0∞4n+3(n+1)(2n+1)P2n+1(zc).(14)\lim_{b\to0}\lambda_{\mathrm{ell}}=\frac{-\epsilon_0 I_0}{\kappa(c-a)}\frac{1}{\ln(c/a-1)}\sum_{n=0}^{\infty}\frac{4n+3}{(n+1)(2n+1)}P_{2n+1}\left(\frac{z}{c}\right). \tag*{(14)}

The infinite sum in (15) can be summed up, see Appendix D. With c−a→b2/(2c)c-a\to b^2/(2c) and ln⁡(c/a−1)→ln⁡(2)−2ln⁡(2c/b)\ln(c/a-1)\to\ln(2)-2\ln(2c/b), both valid for b/c→∞b/c\to\infty, we get

lim⁡b→0λell=qcontcb21ln⁡(2c/b)−ln⁡(2)/2ln⁡(c+zc−z).(15)\lim_{b\to0}\lambda_{\mathrm{ell}}=q_{\mathrm{cont}}\frac{c}{b^2}\frac{1}{\ln(2c/b)-\ln(2)/2}\ln\left(\frac{c+z}{c-z}\right). \tag*{(15)}

For ∣z∣≪c\lvert z\rvert\ll c the line charge density is linear in zz,

lim⁡b→0λell=qcont2b2zln⁡(2c/b)−ln⁡(2)/2+O(z3),(16)\lim_{b\to0}\lambda_{\mathrm{ell}}=q_{\mathrm{cont}}\frac{2}{b^2}\frac{z}{\ln(2c/b)-\ln(2)/2}+O(z^3), \tag*{(16)}

which is identical to Ref. 28 for b≪cb\ll c. If the wire becomes infinitely long, yet its diameter remains finite, the line charge density in (17) goes logarithmically to zero, which is also identical to Refs. 15 and 29. If the wire length is fixed and the diameter goes to zero, the line charge density grows unboundedly. The total charge in the upper half of the infinitely thin wire is

lim⁡b→0qell=∫0clim⁡b→0λell dz=(2cb)2ln⁡(2)2ln⁡(2c/b)−ln⁡(2)qcont>0.(17)\lim_{b\to0}q_{\mathrm{ell}}=\int_0^c\lim_{b\to0}\lambda_{\mathrm{ell}}\,dz=\left(\frac{2c}{b}\right)^2\frac{\ln(2)}{2\ln(2c/b)-\ln(2)}q_{\mathrm{cont}}>0. \tag*{(17)}

Therefore, in the case of an infinitely thin resistive wire carrying DCDC-current, the net charge becomes infinite on the surface of the upper half of the wire. The reason for this singular behavior is that the voltage drop along the wire rises faster than the capacitance of the wire diminishes while the ellipsoid gets thinner. Note that the logarithmic singularity of λell\lambda_{\mathrm{ell}} in z=±cz=\pm c is not responsible for the net charge becoming infinite: If we use (17) instead of (16) in the integration of (18), the net charge is only 2ln⁡2≈1.392\ln2\approx1.39 times smaller. In particular the charge on the surface of the ellipsoid is much larger than the charge on the contact, qell≫qcontq_{\mathrm{ell}}\gg q_{\mathrm{cont}}. This is important, because it justifies the neglection of qcontq_{\mathrm{cont}} when we compute the electric force between two wires in the next section.

The Electric Force between Two Infinitely Thin Resistive Wires with DC Currents

Let us consider two thin wires of finite lengths 2c2c, both being parallel to the zz-axis and extending from z=−cz=-c to z=cz=c. Their cross-sections are circular, and their center lines are spaced apart by a distance ss. A DC-current I0I_0 flows in opposite directions through both wires. The wires are thought to be prolate ellipsoids with their thickest diameters in z=0z=0 being 2b2b. Let the wire diameter shrink infinitely, η0→0\eta_0\to0, which means b→0b\to0 and c→ac\to a. The asymptotic limit of this process is identical to the thin wire limit of a cylindrical wire of constant diameter 2b2b, which also tends to zero, 2b→02b\to0. During this limit process the line contacts of the ellipsoids shrink to point contacts as c→ac\to a, whereas the circular disk contacts at the end faces of the cylindrical wires also become point contacts as b→0b\to0. For a finite diameter 2b2b the resistance RellR_{\mathrm{ell}} between the contacts of an elliptical wire is infinite (ϕinside(θ=0)=∞,(\phi_{\mathrm{inside}}(\theta=0)=\infty, see (3)), whereas the resistance of a cylindrical wire is finite, Rcyl=2c/(πκb2)R_{\mathrm{cyl}} = 2c/(\pi\kappa b^2). However, in the thin wire limit both resistances are identical, Rcyl→RellR_{\mathrm{cyl}} \to R_{\mathrm{ell}}, which means that the thinner the wires get the more similar their resistances become, although they both tend to infinity. Then the potentials along the wires, the electric field around them, the charges on them, and the forces between them converge to the very same limit. In the following we will see how the voltage between both wires and the choice of the common ground node affects the electric force.

Wires are grounded at their halves: Figure 2(a)

In this scenario a first current source is connected between the upper contacts of the wires, and a second identical current source is connected between their lower contacts (see Figure 2(a)). Both wires are grounded in z=0z = 0. No current flows into the ground node due to the symmetry. Thus, the left wire has a line charge density λell(z)\lambda_{\mathrm{ell}}(z) from (16) and the right wire has a line charge density (−1)×λell(z)(-1) \times\lambda_{\mathrm{ell}}(z). Then the xx-component of the electric force on the line charge of the right wire is given by Coulomb’s force law³⁰ between two differential charges on the first and second wires, summed up over both wires,

Two identical thin parallel wires in three grounding scenarios

Figure 2. Two identical thin parallel wires, spaced apart by a distance ss, carry anti-parallel DC-currents I0I_0. The wires have lengths 2c2c and maximum diameters 2b2b in the limit b→0b \to0. The batteries are ideal current sources. In scenario (a) the elliptical wires are grounded in the halves of the wires, i.e., in z=0z = 0. In scenario (b) the upper ends of the cylindrical wires are shorted and grounded in z=cz = c. In the thin wire limit scenario (c) defines the same potential at the upper ends of the elliptical wires as scenario (b) does for cylindrical wires.

lim⁡b→0Fx,el=∫z1=−cc∫z2=−cc−λell(z1)λell(z2)4πϵ01s2+(z1−z2)2ss2+(z1−z2)2 dz1 dz2=−ϵ0I02c2sπκb4(2ln⁡(2c/b)−ln⁡(2))2∫−cc∫−ccln⁡(c+z1c−z1)ln⁡(c+z2c−z2)(s2+(z1−z2)2)3/2 dz1 dz2.(18)\begin{aligned} \lim_{b \to0} F_{x,\mathrm{el}} &= \int_{z_1=-c}^{c}\int_{z_2=-c}^{c} \frac{-\lambda_{\mathrm{ell}}(z_1)\lambda_{\mathrm{ell}}(z_2)} {4\pi\epsilon_0} \frac{1}{s^2+(z_1-z_2)^2} \frac{s}{\sqrt{s^2+(z_1-z_2)^2}}\,\mathrm{d}z_1\,\mathrm{d}z_2 \\ &= -\frac{\epsilon_0 I_0^2 c^2s} {\pi\kappa b^4\left(2\ln(2c/b)-\ln(2)\right)^2} \int_{-c}^{c}\int_{-c}^{c} \frac{\ln\left(\frac{c+z_1}{c-z_1}\right) \ln\left(\frac{c+z_2}{c-z_2}\right)} {\left(s^2+(z_1-z_2)^2\right)^{3/2}}\,\mathrm{d}z_1\,\mathrm{d}z_2. \tag*{(18)} \end{aligned}

This equation can be massaged into the following form (see Appendix E),

lim⁡b→0Fx,el=−2πϵ032cs(I0cκb2(2ln⁡(2c/b)−ln⁡(2)))2Int⁡(s2c)<0,(19)\lim_{b \to0}F_{x,\mathrm{el}} = -\frac{2\pi\epsilon_0}{3}\frac{2c}{s} \left(\frac{I_0c}{\kappa b^2\left(2\ln(2c/b)-\ln(2)\right)}\right)^2 \operatorname{Int}\left(\frac{s}{2c}\right) < 0, \tag*{(19)}
Int⁡(s2c)=3π2(s2c)2∫α′=01∫β′=01−α′ln⁡(1+α′+β′1−α′−β′)ln⁡(1−α′+β′1+α′−β′)((α′)2+(s/2c)2)3/2 dα′ dβ′.\operatorname{Int}\left(\frac{s}{2c}\right) = \frac{3}{\pi^2}\left(\frac{s}{2c}\right)^2 \int_{\alpha'=0}^{1}\int_{\beta'=0}^{1-\alpha'} \frac{\ln\left(\frac{1+\alpha'+\beta'}{1-\alpha'-\beta'}\right) \ln\left(\frac{1-\alpha'+\beta'}{1+\alpha'-\beta'}\right)} {\left((\alpha')^2+(s/2c)^2\right)^{3/2}} \,\mathrm{d}\alpha'\,\mathrm{d}\beta'.

The negative sign of Fx,elF_{x,\mathrm{el}} in (20) means that the electric force due to anti-parallel currents in both wires is attractive. Figure 3 shows a plot of Int⁡(s/2c)\operatorname{Int}(s/2c). It is close to 1 if the wire spacing is less than 1% of the wire length. Then, the electrical force between both wires is dominated by the factors in front of Int⁡(s/2c)\operatorname{Int}(s/2c). In particular, for small spacing the electrical force is proportional to c/sc/s, whereas for very large spacing, s≫2cs \gg2c, the electric force is proportional to c2/s2c^2/s^2. For constant current the electric force grows unboundedly if the wire diameter diminishes.

Figure 3. Plot of the function Int⁡(s/2/c)\operatorname{Int}(s/2/c) defined in (19).

For anti-parallel currents I0I_0 in both wires, the magnetic force on the right wire is given by the Lorentz law, Fx,magn=μ0I02c/(πs)>0F_{x,\mathrm{magn}}=\mu_0 I_0^2c/(\pi s)>0, with μ0=4π×10−7\mu_0=4\pi\times10^{-7} Vs/Am. The positive sign means repulsion. Finally, the ratio of electric over magnetic force is

lim⁡b→0Fx,elFx,magn=−(cLκb21ln⁡(2c/b)−ln⁡(2)/2)2Int⁡(s2c)withLκ=π3κZ0.(20)\lim_{b \to0}\frac{F_{x,\mathrm{el}}}{F_{x,\mathrm{magn}}} = -\left(\frac{cL_\kappa}{b^2} \frac{1}{\ln(2c/b)-\ln(2)/2}\right)^2 \operatorname{Int}\left(\frac{s}{2c}\right) \quad\text{with}\quad L_\kappa=\frac{\pi}{\sqrt{3\kappa Z_0}}. \tag*{(20)}

wherein Z0=μ0/ϵ0≈376.7 ΩZ_0 = \sqrt{\mu_0/\epsilon_0} \approx376.7\ \Omega is the impedance of free space. The characteristic length LκL_\kappa is ≈1.8\approx1.8 times the ratio of the wire resistivity over Z0Z_0. For metal wires the resistivity in Ω×m\Omega\times\mathrm{m} is much smaller than Z0Z_0 in Ω\Omega, and therefore LκL_\kappa is very small (fractions of a nano-meter, for copper wires Lκ≈0.08 nmL_\kappa\approx0.08\ \mathrm{nm}). Consequently, the electric force is much smaller than the magnetic force, as long as the wire length is less than a few meters and the wire diameter is more than a tenth of a milli-meter (Table I).

#ccbbssFx,magnF_{x,\mathrm{magn}} for 100 A/mm2^2I0I_0Fx,el/Fx,magnF_{x,\mathrm{el}}/F_{x,\mathrm{magn}} with (21), Fig. 2(a)Fx,el/Fx,magnF_{x,\mathrm{el}}/F_{x,\mathrm{magn}} with (24), Fig. 2(b)
110 m0.1 mm1 cm3.95 mN3.14 A-47.2 ppm-99.2 ppm
21 m0.1 mm1 cm395 μ\muN3.14 A-0.692 ppm-1.48 ppm
31 m0.1 mm10 cm39.5 μ\muN3.14 A-0.546 ppm-1.34 ppm
41 m1 mm10 cm0.395 N314 A−9.5×10−11-9.5 \times10^{-11}−2.3×10−10-2.3 \times10^{-10}
50.5 m0.5 mm6 mm0.206 N78.5 A−4.8×10−10-4.8 \times10^{-10}−1.0×10−9-1.0 \times10^{-9}
61 m1 μ\mum1 mm4×10−114 \times10^{-11} N314 μ\muA-33.08-70.24
71 m1 nm1 mm4×10−234 \times10^{-23} N314 pA−1.5×1013-1.5 \times10^{13}−3.2×1013-3.2 \times10^{13}

Table 1. Electric and magnetic forces between the wires in Figures 2(a,b) in the thin wire limit. The wires are made of copper with κ=5.88×107\kappa= 5.88 \times10^7 S/m. The current density is always 100 A/mm2^2, which gives the current I0I_0. The electrostatic induction between both wires is neglected (see § 0.4).

Wires are grounded at their upper ends: Figure 2(b)

Here we compute the electric force between the two cylindrical wires of Figure 2(b) in the limit b→0b \to0. The striking difference to the preceding Section is that now the wires are shorted at their upper ends. In the thin wire limit the scenario in Figure 2(b) is equivalent to the scenario in Figure 2(c), which we use to compute the electric force. There the potential in the middle of the left elliptical wire is ϕ(x=−s/2,z=0)=−RcylI0/2\phi(x=-s/2,z=0)=-R_{\mathrm{cyl}}I_0/2 and in the thin wire limit the potential in the middle of the right elliptical wire is ϕ(x=s/2,z=0)=RcylI0/2\phi(x=s/2,z=0)=R_{\mathrm{cyl}}I_0/2 (because Rcyl→RellR_{\mathrm{cyl}} \to R_{\mathrm{ell}}). Thus, we may consider the potential in Scenario (c) as a linear superposition of the scenario in Figure 2(a) and an electrostatic scenario. In the electrostatic scenario no current flows through the ellipsoids, and they are charged up to ±RcylI0/2\pm R_{\mathrm{cyl}}I_0/2. For the electric potentials it holds ϕ(c)=ϕ(a)+ϕstatic\phi_{(c)}=\phi_{(a)}+\phi_{\mathrm{static}}, where the indices ’(a), ’(c)’ refer to the Figures 2(a),(c) and the index ’static’ denotes the electrostatic case. The charges also add up analogously. The amount of charge needed to hold a prolate ellipsoid at potential RcylI0/2R_{\mathrm{cyl}}I_0/2 is given by

qstatic=CellRcylI02withlim⁡b→0Cell=4πϵ0cln⁡(2cb)(21)q_{\mathrm{static}}=C_{\mathrm{ell}}R_{\mathrm{cyl}}\frac{I_0}{2} \quad\text{with}\quad \lim_{b\to0} C_{\mathrm{ell}}=\frac{4\pi\epsilon_0 c}{\ln\left(\frac{2c}{b}\right)} \tag*{(21)}

whereby the capacitance of a prolate ellipsoid, CellC_{\mathrm{ell}}, is derived in Appendix F. In the limit of an infinitely thin cylindrical wire it is known that the charge in electrostatic equilibrium (I0=0I_0=0) distributes uniformly on it18–20,33.

Hence, the line charge density in the electrostatic case is λstatic=qstatic/(2c)\lambda_{\mathrm{static}} = q_{\mathrm{static}}/(2c) with

lim⁡b→0λstatic=lim⁡b→0πϵ0ln⁡(2c/b)I0Rcyl(22)\lim_{b\to0}\lambda_{\mathrm{static}}=\lim_{b\to0}\frac{\pi\epsilon_{0}}{\ln(2c/b)}\frac{I_{0}R_{\mathrm{cyl}}}{} \tag*{(22)}

If we add λstatic\lambda_{\mathrm{static}} to λell\lambda_{\mathrm{ell}} in (15), we finally get the electric force between infinitely thin cylindrical wires in Figure 2(b)

lim⁡b→0Fx,el=−πϵ0cs(I0Rcylln⁡(2c/b))2(π212ln⁡(s2c)+1).(23)\lim_{b\to0}F_{x,\mathrm{el}}=-\pi\epsilon_{0}\frac{c}{s}\left(\frac{I_{0}R_{\mathrm{cyl}}}{\ln(2c/b)}\right)^{2}\left(\frac{\pi^{2}}{12}\ln\left(\frac{s}{2c}\right)+1\right). \tag*{(23)}

Comparison of (23) with (19) shows that the electric force between infinitely thin wires grounded at their upper ends in Figure 2(b) is 2.2 times stronger than if the wires are grounded at their centers in Figure 2(a).

Electrostatic Induction between Both Wires

If the current carrying wires of length 2c2c are brought at a distance s≪2cs \ll2c, the surface charges on the first wire generate an electric field that acts on the second wire, and redistributes the charges there. So far, we have neglected this electrostatic induction, but here we want to estimate its order of magnitude. To this end we compare the electric field generated by the charges of a wire with the electric field generated by the charges of the other wire. Thereby we only need to consider EzE_z, because in the thin wire limit the charges cannot move in lateral xx-, yy-directions.

EzE_z generated by the charges of the first wire onto themselves:

In the thin wire limit we use η=dη\eta= \mathrm{d}\eta with ∣dη∣≪1|\mathrm{d}\eta| \ll1 in (1). It gives dr=asin⁡(θ)dη\mathrm{d}r = a\sin(\theta)\mathrm{d}\eta and z=acos⁡(θ)z = a\cos(\theta) for ∣z∣<a|z| < a. Inserting this into (2) and differentiating against zz gives

Ez(r=0,∣z∣<a)=−I0a2πκ(c−a)1a2−z2(24)E_z(r=0,|z|<a)=-\frac{I_0a}{2\pi\kappa(c-a)}\frac{1}{a^2-z^2} \tag*{(24)}

In the limit of an infinitely thin wire we replace a→ca \to c and c−a→b2/(2c)c-a \to b^2/(2c) and get

Ez(r=0,∣z∣<c)→−I0c2πκb21c2−z2.(25)E_z(r=0,|z|<c)\to-\frac{I_0c^2}{\pi\kappa b^2}\frac{1}{c^2-z^2}. \tag*{(25)}

EzE_z generated by the charges of one wire onto the other wire for the scenario in Figure 2(a):

Ez(r=s,z)=∫z′=−ccλell(z′)4πϵ01s2+(z−z′)2z−z′s2+(z−z′)2 dz′=I0c2πκb2(2ln⁡(2c/b)−ln⁡(2))1(c2+s2−z2)2+(2sz)2×{s2+(c+z)2ln⁡(s2+z2−c2+(c2+s2−z2)2+(2sz)22(s2+(c−z)2))+s2+(c−z)2ln⁡(s2+z2−c2+(c2+s2−z2)2+(2sz)22(s2+(c+z)2))}.(26)\begin{aligned} E_z(r=s,z) &=\int_{z'=-c}^{c}\frac{\lambda_{\mathrm{ell}}(z')}{4\pi\epsilon_0}\frac{1}{s^2+(z-z')^2}\frac{z-z'}{\sqrt{s^2+(z-z')^2}}\,\mathrm{d}z'\\ &=\frac{I_0c}{2\pi\kappa b^2(2\ln(2c/b)-\ln(2))}\frac{1}{\sqrt{(c^2+s^2-z^2)^2+(2sz)^2}}\\ &\quad\times\left\{\sqrt{s^2+(c+z)^2}\ln\left(\frac{s^2+z^2-c^2+\sqrt{(c^2+s^2-z^2)^2+(2sz)^2}}{2(s^2+(c-z)^2)}\right)\right.\\ &\qquad\left.+\sqrt{s^2+(c-z)^2}\ln\left(\frac{s^2+z^2-c^2+\sqrt{(c^2+s^2-z^2)^2+(2sz)^2}}{2(s^2+(c+z)^2)}\right)\right\}. \tag*{(26)} \end{aligned}

This integral can be computed with Mathematica, but its exact formula is not even necessary if we pull out the diverging terms in λell(z′)\lambda_{\mathrm{ell}}(z') for b→0b \to0,

lim⁡b→0Ez(r=0,∣z∣<c)Ez(r=s,z)∝ln⁡(2cb)→∞.(27)\lim_{b\to0}\frac{E_z(r=0,|z|<c)}{E_z(r=s,z)}\propto\ln\left(\frac{2c}{b}\right)\to\infty. \tag*{(27)}

This proves that the electric field acting on the charges of one wire produced by the charges on the other wire is infinitely smaller than the field of the charges on a wire on themselves, if both wires are infinitely thin while their spacing is finite. Therefore we can neglect electrostatic induction between both wires.

Conclusion

In this paper I discussed the surface charges on an infinitely thin straight resistive wire which carries a DC-current. Thereby the wire is replaced by a prolate ellipsoid in the limit of infinite slimness. The distribution of surface charges varies linearly with the position near the center of the wire, whereas it has a logarithmic singularity at the ends of the wire. If two wires run parallel, electrostatic induction is negligible, as long as the wires are infinitely thin and spaced apart at a finite distance. The electric force between the surface charges on both wires was computed. It depends on the voltage drop along the wires, on the voltage between both wires, on the wire lengths, diameters, and their spacing. If the current, the wires lengths, and their spacing are fixed while the wires diameters shrink, the voltage drop grows unboundedly, and this will give infinite surface charges and infinite electric force. This limit leads to infinite current density and heating in the wire, which eventually distroys the wire. However, if we consider the ratio of electric over magnetic force between both wires, this ratio is independent of the current, and we may scale down the current synchronously with the wire diameter to achieve constant current density during the limit process. Of course, the magnetic force decreases with the current accordingly, but in practice this just calls for a sufficiently sensitive method of force measurement. In such a scenario the electric force can indeed become even stronger than the magnetic force.

The electric force can be eliminated if each wire is surrounded by an electric shield (like a coaxial cable) and both shields are tied to the same potential (e.g. ground). Each shield has to be clamped mechanically to its conductor, because there might be an electric force between the shield and its conductor (depending on symmetry).

This manuscript was submitted to the American Journal of Physics, but it was rejected (too long, too mathematically dense, lack of interest).

The author has no conflicts to disclose.

How to determine the constant c1c_1 of the potential ϕinside\phi_{\mathrm{inside}}

Let us look at the potential close to z=0z=0, which means z=0+dzz=0+\mathrm{d}z with ∣dz∣≪a\lvert\mathrm{d}z\rvert\ll a and θ=π/2+dθ\theta=\pi/2+\mathrm{d}\theta with ∣dθ∣≪1\lvert\mathrm{d}\theta\rvert\ll1. Inserting this into (1) gives

x≈asinh⁡(η)cos⁡(ψ),y≈asinh⁡(η)sin⁡(ψ),dz≈−acosh⁡(η) dθ.(28)x \approx a\sinh(\eta)\cos(\psi),\quad y \approx a\sinh(\eta)\sin(\psi),\quad\mathrm{d}z \approx-a\cosh(\eta)\,\mathrm{d}\theta. \tag*{(28)}

With the radial distance r=x2+y2r=\sqrt{x^2+y^2} it follows with (A1)

(cosh⁡(η))2=1+(sinh⁡(η))2⇒(dza dθ)2=1+(ra)2.(29)\left(\cosh(\eta)\right)^2=1+\left(\sinh(\eta)\right)^2 \quad\Rightarrow\quad\left(\frac{\mathrm{d}z}{a\,\mathrm{d}\theta}\right)^2=1+\left(\frac{r}{a}\right)^2. \tag*{(29)}

We solve (A2) for dθ\mathrm{d}\theta, whereby dθ\mathrm{d}\theta and dz\mathrm{d}z have opposite sign. Inserting this into (3) gives

dϕinside(r,dz)=c12ln⁡(1−dθ1+dθ)=c12ln⁡(1+dz/a2+r21−dz/a2+r2)=c1 dza2+r2.(30)\mathrm{d}\phi_{\mathrm{inside}}(r,\mathrm{d}z)=\frac{c_1}{2}\ln\left(\frac{1-\mathrm{d}\theta}{1+\mathrm{d}\theta}\right)=\frac{c_1}{2}\ln\left(\frac{1+\mathrm{d}z/\sqrt{a^2+r^2}}{1-\mathrm{d}z/\sqrt{a^2+r^2}}\right)=\frac{c_1\,\mathrm{d}z}{\sqrt{a^2+r^2}}. \tag*{(30)}

From (A3) we get the zz-component of the electric field in z=0z=0

Ez(r,z=0)=−∂ϕinside(r,z)∂z∣z=0=−c1a2+r2<0.(31)E_z(r,z=0)=-\left.\frac{\partial\phi_{\mathrm{inside}}(r,z)}{\partial z}\right|_{z=0}=\frac{-c_1}{\sqrt{a^2+r^2}}<0. \tag*{(31)}

With Ohm’s law the zz-component of the current density is Jz=κEzJ_z=\kappa E_z, and the total current I0I_0 downward through the ellipsoid is given by the integral

I0=∫r=0b(−Jz)2πr dr=2πκc1∫r=0bra2+r2 dr=2πκc1(c−a),⇒lim⁡b→0c1=lim⁡b→0I02πκ(c−a)=I0cπκb2.(32)\begin{aligned} I_0=\int_{r=0}^{b}(-J_z)2\pi r\,\mathrm{d}r=2\pi\kappa c_1\int_{r=0}^{b}\frac{r}{\sqrt{a^2+r^2}}\,\mathrm{d}r=2\pi\kappa c_1(c-a), \\ &\Rightarrow\lim_{b\to0}c_1=\lim_{b\to0}\frac{I_0}{2\pi\kappa(c-a)}=\frac{I_0c}{\pi\kappa b^2}. \tag*{(32)} \end{aligned}

Here we used a2+b2=c2a^2+b^2=c^2. Inserting c1c_1 from (A5) into (3) gives the potential everywhere inside the ellipsoid, if the current is known.

Proof of (9)

∫−11P2n+1(x)Q0(x) dx=∫−11P2n+1(x)12ln⁡(1+x1−x) dx=∫−11P2n+1(x)∑m=0∞x2m+12m+1 dx=∑m=0∞∫−11P2n+1(x)2m+1∑ℓ=0m22ℓ+1(4ℓ+3)(2m+1)!(m+ℓ+1)!(2m+2ℓ+3)!(m−ℓ)!P2ℓ+1(x) dx,(33)\begin{aligned} \int_{-1}^{1} P_{2n+1}(x)Q_0(x)\,\mathrm{d}x &= \int_{-1}^{1} P_{2n+1}(x)\frac{1}{2}\ln\left(\frac{1+x}{1-x}\right)\,\mathrm{d}x \\ &= \int_{-1}^{1} P_{2n+1}(x)\sum_{m=0}^{\infty}\frac{x^{2m+1}}{2m+1}\,\mathrm{d}x \\ &= \sum_{m=0}^{\infty}\int_{-1}^{1}\frac{P_{2n+1}(x)}{2m+1}\sum_{\ell=0}^{m}\frac{2^{2\ell+1}(4\ell+3)(2m+1)!(m+\ell+1)!}{(2m+2\ell+3)!(m-\ell)!}P_{2\ell+1}(x)\,\mathrm{d}x, \tag*{(33)} \end{aligned}

where we first developped the logarithm into a Taylor series, and then we used the Legendre series/sum of odd powers of xx from Ref. 31. Next we apply the orthogonality of the Legendre polynomials on (B1), ∫−11P2n+1(x)P2ℓ+1(x) dx=2δℓ,n/(4n+3)\int_{-1}^{1}P_{2n+1}(x)P_{2\ell+1}(x)\,\mathrm{d}x=2\delta_{\ell,n}/(4n+3), and we reverse the sequence of summations, ∑m=0∞∑ℓ=0m=∑ℓ=0∞∑m=ℓ∞\sum_{m=0}^{\infty}\sum_{\ell=0}^{m}=\sum_{\ell=0}^{\infty}\sum_{m=\ell}^{\infty}. Then (B1) becomes

=22n+2∑m=n∞(2m)!(m+n+1)!(2m+2n+3)!(m−n)!=1(n+1)(2n+1)=:Rn.(34)=2^{2n+2}\sum_{m=n}^{\infty}\frac{(2m)!(m+n+1)!}{(2m+2n+3)!(m-n)!}=\frac{1}{(n+1)(2n+1)}=:R_n. \tag*{(34)}

The summation in (B2) is handled by Mathematica. For its proof we start with the following identity[34]

Q2n+1(x)=22n+1∑ℓ=0∞(2ℓ+2n+1)!(ℓ+2n+1)!(2ℓ+4n+3)!ℓ!x−2ℓ−2n−2.(35)Q_{2n+1}(x)=2^{2n+1}\sum_{\ell=0}^{\infty}\frac{(2\ell+2n+1)!(\ell+2n+1)!}{(2\ell+4n+3)!\ell!}x^{-2\ell-2n-2}. \tag*{(35)}

Integration of (B3) gives

∫1∞Q2n+1(x) dx=22n+1∑ℓ=0∞(2ℓ+2n)!(ℓ+2n+1)!(2ℓ+4n+3)!ℓ!(2ℓ+2n+1)∫1∞x−2ℓ−2n−2 dx=12Rn.(36)\begin{aligned} \int_{1}^{\infty}Q_{2n+1}(x)\,\mathrm{d}x &=2^{2n+1}\sum_{\ell=0}^{\infty}\frac{(2\ell+2n)!(\ell+2n+1)!}{(2\ell+4n+3)!\ell!}(2\ell+2n+1)\int_{1}^{\infty}x^{-2\ell-2n-2}\,\mathrm{d}x \\ &=\frac{1}{2}R_n. \tag*{(36)} \end{aligned}

With the recurrence relation[35]

(2n+1)Qn(x)=Qn+1′(x)−Qn−1′(x),(37)(2n+1)Q_n(x)=Q'_{n+1}(x)-Q'_{n-1}(x), \tag*{(37)}

it follows for 0<ϵ≪10<\epsilon\ll1

∫1+ϵ∞Q2n+1(x) dx=∫1+ϵ∞Q2n+2′(x)−Q2n′(x)4n+3 dx=Q2n(1+ϵ)−Q2n+2(1+ϵ)4n+3.(38)\int_{1+\epsilon}^{\infty}Q_{2n+1}(x)\,\mathrm{d}x=\int_{1+\epsilon}^{\infty}\frac{Q'_{2n+2}(x)-Q'_{2n}(x)}{4n+3}\,\mathrm{d}x=\frac{Q_{2n}(1+\epsilon)-Q_{2n+2}(1+\epsilon)}{4n+3}. \tag*{(38)}

In (B5,B6) the primes denote differentiation with respect to xx. We insert (C4) into (B6), let ϵ→0\epsilon\to0, and insert this into (B4),

12Rn=ω2n+2−ω2n4n+3.(39)\frac{1}{2}R_n=\frac{\omega_{2n+2}-\omega_{2n}}{4n+3}. \tag*{(39)}

Inserting (C8) and (C10) into (B7) finally gives

Rn=2(ω2n+2−ω2n+1)+(ω2n+1−ω2n)4n+3=24n+3(12n+2+12n+1)=1(n+1)(2n+1).(40)\begin{aligned} R_n&=2\frac{(\omega_{2n+2}-\omega_{2n+1})+(\omega_{2n+1}-\omega_{2n})}{4n+3}\\ &=\frac{2}{4n+3}\left(\frac{1}{2n+2}+\frac{1}{2n+1}\right)=\frac{1}{(n+1)(2n+1)}. \tag*{(40)} \end{aligned}

Proof of (14)

With Mathematica we compute

lim⁡x→1+0(x−Q2n+2(x)Q2n+1(x))ln⁡(x−1)=−1n+1.(41)\lim_{x\to1+0}\left(x-\frac{Q_{2n+2}(x)}{Q_{2n+1}(x)}\right)\ln(x-1)=-\frac{1}{n+1}. \tag*{(41)}

For a proof we start with the recurrence relation[35, 36]

nQn(x)=(2n−1)xQn−1(x)−(n−1)Qn−2(x),⇒lim⁡x→1+0nQn(x)=lim⁡x→1+0(2n−1)Qn−1(x)+(n−1)Qn−2(x)(42)\begin{aligned} nQ_n(x)&=(2n-1)xQ_{n-1}(x)-(n-1)Q_{n-2}(x),\\ \Rightarrow\lim_{x\to1+0}nQ_n(x)&=\lim_{x\to1+0}(2n-1)Q_{n-1}(x)+(n-1)Q_{n-2}(x) \tag*{(42)} \end{aligned}

with the limits of the first three Q-functions

lim⁡x→1+0Q0(x)=lim⁡x→1+012ln⁡(2x−1),\lim_{x\to1+0}Q_0(x)=\lim_{x\to1+0}\frac{1}{2}\ln\left(\frac{2}{x-1}\right),
lim⁡x→1+0Q1(x)=lim⁡x→1+012ln⁡(2x−1)−1,\lim_{x\to1+0}Q_1(x)=\lim_{x\to1+0}\frac{1}{2}\ln\left(\frac{2}{x-1}\right)-1,
lim⁡x→1+0Q2(x)=lim⁡x→1+012ln⁡(2x−1)−32.(43)\lim_{x\to1+0}Q_2(x)=\lim_{x\to1+0}\frac{1}{2}\ln\left(\frac{2}{x-1}\right)-\frac{3}{2}. \tag*{(43)}

Inserting (C3) into (C2) shows that the logarithmic term is identical for all Qn(x→1+0)Q_n(x\to1+0). Thus, we can write

lim⁡x→1+0Qn(x)=lim⁡x→1+012ln⁡(2x−1)−ωn,(44)\lim_{x\to1+0}Q_n(x)=\lim_{x\to1+0}\frac{1}{2}\ln\left(\frac{2}{x-1}\right)-\omega_n, \tag*{(44)}

where ωn\omega_n is a rational number. We insert (C4) into the recurrence relation (C2) and get

nωn=(2n−1)ωn−1−(n−1)ωn−2,with ω0=0, ω1=1, ω2=3/2.(45)n\omega_n=(2n-1)\omega_{n-1}-(n-1)\omega_{n-2},\quad\text{with }\omega_0=0,\ \omega_1=1,\ \omega_2=3/2. \tag*{(45)}

Inserting (C5) into the left side of (C1) gives

lim⁡x→1+0Q2n+2(x)Q2n+1(x)=lim⁡x→1+01+2ω2n+2−ω2n+1ln⁡(x−1),⇒lim⁡x→1+0(x−Q2n+2(x)Q2n+1(x))ln⁡(x−1)=−2(ω2n+2−ω2n+1).(46)\begin{aligned} \lim_{x\to1+0}\frac{Q_{2n+2}(x)}{Q_{2n+1}(x)}&=\lim_{x\to1+0}1+\frac{2\omega_{2n+2}-\omega_{2n+1}}{\ln(x-1)},\\ \Rightarrow\lim_{x\to1+0}\left(x-\frac{Q_{2n+2}(x)}{Q_{2n+1}(x)}\right)\ln(x-1)&=-2(\omega_{2n+2}-\omega_{2n+1}). \tag*{(46)} \end{aligned}

If we express ω2n+2\omega_{2n+2} by ω2n+1,ω2n\omega_{2n+1},\omega_{2n} via the recurrence relations (C5) we get after a few re-arrangements

(2n+2)(ω2n+2−ω2n+1)=2n(ω2n−ω2n−1).(47)(2n+2)(\omega_{2n+2}-\omega_{2n+1})=2n(\omega_{2n}-\omega_{2n-1}). \tag*{(47)}

For the first few indices nn this gives

n=1: 4(ω4−ω3)=2(ω2−ω1)=2(32−1)=1,n=2: 6(ω6−ω5)=4(ω4−ω3)=1,n=3: 8(ω8−ω7)=6(ω6−ω5)=1,n: ω2n+2−ω2n+1=12n+2.(48)\begin{aligned} n=1:\ &4(\omega_4-\omega_3)=2(\omega_2-\omega_1)=2\left(\frac{3}{2}-1\right)=1,\\ n=2:\ &6(\omega_6-\omega_5)=4(\omega_4-\omega_3)=1,\\ n=3:\ &8(\omega_8-\omega_7)=6(\omega_6-\omega_5)=1,\\ n:\ &\omega_{2n+2}-\omega_{2n+1}=\frac{1}{2n+2}. \tag*{(48)} \end{aligned}

Inserting this into (C6) completes the proof of (C1). Moreover, setting n→2n+1n\to2n+1 in (C5) gives

(2n+1)(ω2n+1−ω2n)=2n(ω2n−ω2n−1).(49)(2n+1)(\omega_{2n+1}-\omega_{2n})=2n(\omega_{2n}-\omega_{2n-1}). \tag*{(49)}

For the first few indices nn this gives

n=1: 3(ω3−ω2)=2(ω2−ω1)=2(32−1)=1,n=2: 5(ω5−ω4)=4(ω4−ω3)=1,n=3: 7(ω7−ω6)=6(ω6−ω5)=1,n: ω2n+1−ω2n=12n+1.(50)\begin{aligned} n=1:\ &3(\omega_3-\omega_2)=2(\omega_2-\omega_1)=2\left(\frac{3}{2}-1\right)=1,\\ n=2:\ &5(\omega_5-\omega_4)=4(\omega_4-\omega_3)=1,\\ n=3:\ &7(\omega_7-\omega_6)=6(\omega_6-\omega_5)=1,\\ n:\ &\omega_{2n+1}-\omega_{2n}=\frac{1}{2n+1}. \tag*{(50)} \end{aligned}

How to compute the sum in (15)

We want to compute

S=∑n=0∞4n+32n+1P2n+1(x)n+1=2∑n=0∞P2n+1(x)n+1⏟=S1+∑n=0∞12n+1P2n+1(x)n+1⏟=S2.(51)S=\sum_{n=0}^{\infty}\frac{4n+3}{2n+1}\frac{P_{2n+1}(x)}{n+1}=2\underbrace{\sum_{n=0}^{\infty}\frac{P_{2n+1}(x)}{n+1}}_{=S_1}+\underbrace{\sum_{n=0}^{\infty}\frac{1}{2n+1}\frac{P_{2n+1}(x)}{n+1}}_{=S_2}. \tag*{(51)}

For the sum S1S_1 we start with the generating function of the Legendre polynomials[32]

11−2xt+t2=∑n=0∞Pn(x)tn∀t∈[−1,1].(52)\frac{1}{\sqrt{1-2xt+t^2}}=\sum_{n=0}^{\infty}P_n(x)t^n\quad\forall t\in[-1,1]. \tag*{(52)}

We integrate (D2) once over t:0→1t:0\to1 and once over t:0→−1t:0\to-1. Adding both results cancels out even indices nn,

∫0111−2xt+t2 dt+∫0−111−2xt+t2 dt=ln⁡(1+21−x)−ln⁡(1+21+x)=∑odd nPn(x)1+(−1)n+1n+1=∑n=0∞P2n+1(x)n+1=S1.(53)\begin{aligned} \int_0^1\frac{1}{\sqrt{1-2xt+t^2}}\,dt+\int_0^{-1}\frac{1}{\sqrt{1-2xt+t^2}}\,dt =\ln\left(1+\sqrt{\frac{2}{1-x}}\right)-\ln\left(1+\sqrt{\frac{2}{1+x}}\right) \\ &=\sum_{\mathrm{odd}\ n}P_n(x)\frac{1+(-1)^{n+1}}{n+1}=\sum_{n=0}^{\infty}\frac{P_{2n+1}(x)}{n+1}=S_1. \tag*{(53)} \end{aligned}

For the sum S2S_2 we set t=yzt=yz in (D2) and divide both sides by yy. This gives

1y1−2xyz+y2z2=∑n=0∞Pn(x)yn−1zn∀y,z∈[−1,1].(54)\frac{1}{y\sqrt{1-2xyz+y^2z^2}}=\sum_{n=0}^{\infty}P_n(x)y^{n-1}z^n\quad\forall y,z\in[-1,1]. \tag*{(54)}

Like above, we integrate (D4) once over z:0→1z:0\to1 and once over z:0→−1z:0\to-1, and we add both results to cancel out even indices nn,

∫011y1−2xyz+y2z2 dz+∫0−11y1−2xyz+y2z2 dz=∑odd nPn(x)yn−11+(−1)n+1n+1=∑n=0∞P2n+1(x)n+1y2n.(55)\begin{aligned} \int_0^1\frac{1}{y\sqrt{1-2xyz+y^2z^2}}\,dz+\int_0^{-1}\frac{1}{y\sqrt{1-2xyz+y^2z^2}}\,dz \\ &=\sum_{\mathrm{odd}\ n}P_n(x)y^{n-1}\frac{1+(-1)^{n+1}}{n+1}=\sum_{n=0}^{\infty}\frac{P_{2n+1}(x)}{n+1}y^{2n}. \tag*{(55)} \end{aligned}

Next, we integrate (D5) over y:−1→1y:-1\to1 whereby we take the Cauchy principal value around the singularity in y=0y=0,

\begin{aligned} \underset{-1}{\overset{1}{\fint}}\left(\int_0^1\frac{1}{y\sqrt{1-2xyz+y^2z^2}}\,dz+\int_0^{-1}\frac{1}{y\sqrt{1-2xyz+y^2z^2}}\,dz\right)\,dy =2\ln\left(\frac{3+x+2\sqrt{2}\sqrt{1+x}}{3-x+2\sqrt{2}\sqrt{1-x}}\right) \\ &=\sum_{n=0}^{\infty}\frac{P_{2n+1}(x)}{n+1}\frac{1-(-1)^{2n+1}}{2n+1}=2S_2. \tag*{(56)} \end{aligned}

The computation of the integrals in (D6) is easier, if one starts with the integration over yy before the integration over zz,

\underset{-1}{\overset{1}{\fint}}\frac{1}{y\sqrt{1-2xyz+y^2z^2}}\,dy =\ln\left(\frac{1+xz+\sqrt{1+2xz+z^2}}{1-xz+\sqrt{1-2xz+z^2}}\right),
∫01ln⁡(1+xz+1+2xz+z21−xz+1−2xz+z2) dz=ln⁡(3+x+221+x3−x+221−x).(57)\int_0^1\ln\left(\frac{1+xz+\sqrt{1+2xz+z^2}}{1-xz+\sqrt{1-2xz+z^2}}\right)\,dz =\ln\left(\frac{3+x+2\sqrt{2}\sqrt{1+x}}{3-x+2\sqrt{2}\sqrt{1-x}}\right). \tag*{(57)}

Combining (D3) and (D6) finally gives S=2S1+S2=ln⁡(1+x)−ln⁡(1−x)S=2S_1+S_2=\ln(1+x)-\ln(1-x).

Transformation of coordinates $(z_1,z_2)\mapsto(\alpha,\beta)$

Figure 4. Transformation of coordinates (z1,z2)↦(α,β)(z_1,z_2)\mapsto(\alpha,\beta) to compute the integral in (E4).

How to derive Int⁡(s2c)\operatorname{Int}\left(\frac{s}{2c}\right) in equation (20)

Let us call the integrand in the last line of (19) f(z1,z2)f(z_1,z_2). It is symmetrical, because it is identical to f(z2,z1)f(z_2,z_1). Therefore, we may halve the integration domain in (19),

∫z1=−cc∫z2=−ccf(z1,z2) dz1dz2=2∫z1=−cc∫z2=−cz1f(z1,z2) dz1dz2.(58)\int_{z_1=-c}^{c}\int_{z_2=-c}^{c} f(z_1,z_2)\,\mathrm{d}z_1\mathrm{d}z_2=2\int_{z_1=-c}^{c}\int_{z_2=-c}^{z_1} f(z_1,z_2)\,\mathrm{d}z_1\mathrm{d}z_2. \tag*{(58)}

In Figure 4 this reduces the integration domain from the square z1∈[−c,c]∧z2∈[−c,c]z_1\in[-c,c]\land z_2\in[-c,c] to the dark triangle. Next, we transform the integration variables (z1,z2)↦(α,β)(z_1,z_2)\mapsto(\alpha,\beta) according to

α=z1−z2,β=z1+z2,↔z1=α+β2,z2=−α+β2,(59)\alpha=z_1-z_2,\quad\beta=z_1+z_2,\quad\leftrightarrow\quad z_1=\frac{\alpha+\beta}{2},\quad z_2=\frac{-\alpha+\beta}{2}, \tag*{(59)}

which is also shown in Figure 4. Thereby the differential surface elements relate via the Jacobian determinant,

dz1dz2=∣∂(z1,z2)∂(α,β)∣dαdβ,with∂(z1,z2)∂(α,β)=∣1/2−1/21/21/2∣=12.(60)\mathrm{d}z_1\mathrm{d}z_2=\left|\frac{\partial(z_1,z_2)}{\partial(\alpha,\beta)}\right|\mathrm{d}\alpha\mathrm{d}\beta,\quad\text{with}\quad\frac{\partial(z_1,z_2)}{\partial(\alpha,\beta)}=\left|\begin{matrix}1/2&-1/2\\1/2&1/2\end{matrix}\right|=\frac{1}{2}. \tag*{(60)}

It holds

2∫z1=−cc∫z2=−cz1f(z1,z2) dz1dz2=2∫α=02c(∫β=α−2c0g(α,β)dαdβ2+∫β=02c−αg(α,β)dαdβ2),(61)\begin{aligned} 2\int_{z_1=-c}^{c}\int_{z_2=-c}^{z_1}f(z_1,z_2)\,\mathrm{d}z_1\mathrm{d}z_2 &=2\int_{\alpha=0}^{2c}\left(\int_{\beta=\alpha-2c}^{0}g(\alpha,\beta)\frac{\mathrm{d}\alpha\mathrm{d}\beta}{2}+\int_{\beta=0}^{2c-\alpha}g(\alpha,\beta)\frac{\mathrm{d}\alpha\mathrm{d}\beta}{2}\right), \tag*{(61)} \end{aligned}

with g(α,β)=ln⁡(2c+α+β2c−α−β)ln⁡(2c−α+β2c+α−β)(s2+α2)−3/2g(\alpha,\beta)=\ln\left(\frac{2c+\alpha+\beta}{2c-\alpha-\beta}\right)\ln\left(\frac{2c-\alpha+\beta}{2c+\alpha-\beta}\right)(s^2+\alpha^2)^{-3/2}.

g(α,β)g(\alpha,\beta) is an even function of β\beta. Therefore, in (E4) the two integrals over g(α,β)g(\alpha,\beta) are identical, and the final result is given in (20), whereby we used the transformations α=2cα′\alpha=2c\alpha' and β=2cβ′\beta=2c\beta'. In (20) the integration over β′\beta' can be done in closed form,

Int⁡(x)=x2π2∫α′=01f1(α′)(α′2+x2)3/2 dα′,\operatorname{Int}(x)=\frac{x^2}{\pi^2}\int_{\alpha'=0}^{1}\frac{f_1(\alpha')}{(\alpha'^2+x^2)^{3/2}}\,\mathrm{d}\alpha',
f1(α′)=2π2−3(1−α′)(ln⁡(1−α′))2−3α′(ln⁡(α′))2+6α′Li⁡2(α′)−6(1+α′)(Li⁡2(1+α′)+ıπln⁡(1+α′)),(62)\begin{aligned} f_1(\alpha')={}&2\pi^2-3(1-\alpha')(\ln(1-\alpha'))^2-3\alpha'(\ln(\alpha'))^2+6\alpha'\operatorname{Li}_2(\alpha')\\ &-6(1+\alpha')\left(\operatorname{Li}_2(1+\alpha')+\imath\pi\ln(1+\alpha')\right), \tag*{(62)} \end{aligned}

with the imaginary unit ı=−1\imath=\sqrt{-1}, and with the poly-logarithm Li⁡2(x)=∑k=1∞xk/k2\operatorname{Li}_2(x)=\sum_{k=1}^{\infty}x^k/k^2. f1(α′)f_1(\alpha') is real, all imaginary portions are in the last line of (E5) and cancel out.

We compute the limit of Int⁡(x)\operatorname{Int}(x) for x→0x \to0 with partial integration,

x2∫α′=ϵ1−ϵf1(α′)(α′2+x2)3/2 dα′=α′f1(α′)α′2+x2∣ϵ1−ϵ−∫α′=ϵ1−ϵα′f2(α′)α′2+x2 dα′,(63)x^2 \int_{\alpha'=\epsilon}^{1-\epsilon} \frac{f_1(\alpha')}{(\alpha'^2+x^2)^{3/2}}\,\mathrm{d}\alpha' = \left.\frac{\alpha' f_1(\alpha')}{\sqrt{\alpha'^2+x^2}}\right|_{\epsilon}^{1-\epsilon} - \int_{\alpha'=\epsilon}^{1-\epsilon} \frac{\alpha' f_2(\alpha')}{\sqrt{\alpha'^2+x^2}}\,\mathrm{d}\alpha', \tag*{(63)}
f2(α′)=df1(α′)dα′=3((ln⁡(1−α′))2−(ln⁡(α′))2−2iπln⁡(1+α′)+2Li⁡2(α′)−2Li⁡2(1+α′)).f_2(\alpha') = \frac{\mathrm{d}f_1(\alpha')}{\mathrm{d}\alpha'} = 3\left((\ln(1-\alpha'))^2-(\ln(\alpha'))^2-2\mathrm{i}\pi\ln(1+\alpha')+2\operatorname{Li}_2(\alpha')-2\operatorname{Li}_2(1+\alpha')\right).

For the integral on the right hand side of in (E6) we get again with partial integration

α′2+x2f2(α′)∣ϵ1−ϵ−∫α′=ϵ1−ϵ6α′α′2+x2(1+α′)ln⁡(1−α′)+(1−α′)ln⁡(α′)α′2−1 dα′.(64)\left.\sqrt{\alpha'^2+x^2}f_2(\alpha')\right|_{\epsilon}^{1-\epsilon}-\int_{\alpha'=\epsilon}^{1-\epsilon}\frac{6}{\alpha'}\sqrt{\alpha'^2+x^2}\frac{(1+\alpha')\ln(1-\alpha')+(1-\alpha')\ln(\alpha')}{\alpha'^2-1}\,\mathrm{d}\alpha'. \tag*{(64)}

For x→0x \to0, the integral in (E7) tends to π2/2+6ϵ(ln⁡(ϵ)−1)+3(ln⁡(ϵ))2\pi^2/2+6\epsilon(\ln(\epsilon)-1)+3(\ln(\epsilon))^2 (computed with Mathematica). We re-insert this into (E7) and (E6), and compute with Mathematica the limit for ϵ→0\epsilon\to0. The result is lim⁡x→0Int⁡(x)=1\lim_{x\to0}\operatorname{Int}(x)=1.

The Capacitance of a Prolate Ellipsoid

The potential outside a charged metallic ellipsoid has the same ansatz as in (6) with the only non-zero coefficient A0=ϕ0/Q0(cosh⁡(η0))A_0=\phi_0/Q_0(\cosh(\eta_0)), whereby the ellipsoid is at potential ϕ0\phi_0 and its surface has the ellipsoidal coordinate η=η0\eta=\eta_0. The electric field on the surface of the ellipsoid is given analogous to (10),

E⊥,outside=−1a(sinh⁡(η0))2+(sin⁡(θ))2A0P0(cos⁡(θ))⏟=1ddη0Q0(cosh⁡(η0))⏟=−1/sinh⁡(η0).(65)E_{\perp,\mathrm{outside}}=\frac{-1}{a\sqrt{(\sinh(\eta_0))^2+(\sin(\theta))^2}}A_0\underbrace{P_0(\cos(\theta))}_{=1}\frac{\mathrm{d}}{\mathrm{d}\eta_0}\underbrace{Q_0(\cosh(\eta_0))}_{=-1/\sinh(\eta_0)}. \tag*{(65)}

The charge on this ellipsoid is given analogous to (12)

qell=∫0πdqell=ϵ0ϕ0Q0(cosh⁡(η0))2πa∫0πsin⁡(θ) dθ.(66)q_{\mathrm{ell}}=\int_0^\pi\mathrm{d}q_{\mathrm{ell}}=\frac{\epsilon_0\phi_0}{Q_0(\cosh(\eta_0))}2\pi a\int_0^\pi\sin(\theta)\,\mathrm{d}\theta. \tag*{(66)}

We insert a=c2−b2a=\sqrt{c^2-b^2} into (F2) and compute qell/ϕ0q_{\mathrm{ell}}/\phi_0, which gives CellC_{\mathrm{ell}} from (21), which is consistent to Refs. 37–39.

References

References

  1. [1]Richard Becker. Theorie der Elektrizität. Volume I: Einführung in die Maxwellsche Theorie (Teubner Verlag, Stuttgart 1973), 21st edition, chapter 5.4 (in German) also available in English: Richard Becker. Electromagnetic Fields and Interactions. Volume I: Electromagnetic Theory and Relativity (Blaisdell, New York 1964)
  2. [2]John David Jackson. Classical Electrodynamics. (Walter de Gruyter, Berlin 1983), 2nd edition, Appendix. Units and Dimensions
  3. [3]Erik Hallén. Electromagnetic Theory. (John Wiley and Sons, New York 1962), chapter 11.1
  4. [4]Károly Simonyi. Theoretische Elektrotechnik. (VEB Deutscher Verlag der Wissenschaften, Berlin 1971), 4th edition, chapter 1.11 (in German) also available in English: K. Simonyi. Foundations of Electrical Engineering: Fields—Networks—Waves. (Elsevier 2016).
  5. [5]Hellmut Hofmann. Das elektromagnetische Feld. (Springer Verlag, Wien 1986), 3rd edition, §3.3.2.1.8 (in German)
  6. [6]Günther Lehner. Elektromagnetische Feldtheorie. (Springer, Heidelberg 2010), 7. edition, chapter 13 (in German)
  7. [11]A. Marcus. “The electric field associated with a steady current in long cylindrical conductor,” American Journal of Physics 9 (4), 225-226 (1941).DOI
  8. [12]Julius Adams Stratton. Electromagnetic Theory. (McGraw-Hill, New York, 1941), §4.21, problem 2.
  9. [13]E. Merzbacher. “A puzzle from professor Eugen Merzbacher,” Am. J. Phys 48, 178 (1980).
  10. [14]J. D. Jackson. “Surface charges on circuit wires and resistors play three roles,” American Journal of Physics, 64 (7), 855-870 (1996).DOI
  11. [15]A. K. T. Assis, W. A. Rodrigues Jr, and A. J. Mania. “The electric field outside a stationary resistive wire carrying a constant current,” Foundations of Physics 29 (5), 729-753 (1999).DOI
  12. [16]Andre Koch Torres Assis, and Julio Akashi Hernandes, The electric force of a current: Weber and the surface charges of resistive conductors carrying steady currents, 1st edition (Apeiron, Montreal, Canada, 2007).
  13. [17]Andrew Zangwill, Modern Electrodynamics, 1st edition (Cambridge Univ. Press, Cambridge, UK, 2013), chapter 9.7.4.DOI
  14. [18]J. C. Maxwell, “On the electrical capacity of a long narrow cylinder, and of a disk of sensible thickness,” Proc. London Math. Soc. IX, 94-101 (1878).
  15. [19]M. Andrews, “Equilibrium charge density on a conducting needle,” American Journal of Physics 65 (9), 846-850 (1997).DOI
  16. [20]J. D. Jackson. “Charge density on thin straight wire, revisited,” American Journal of Physics 68.9 (2000): 789-799.DOI
  17. [21]Parry Moon and Domina Eberle Spencer, Field Theory for Engineers, 1st edition (D. van Nostrand Company Inc., Princeton, New Jersey, 1961), §9.02, eqs. (9.12), (9.13).DOI
  18. [22]William Ralph Smythe. Static and Dynamic Electricity, 3rd ed., (Taylor & Francis, 1989) chapter I, 1.10 (1).
  19. [23]L. J. van der Pauw. “A method of measuring the resistivity and Hall coefficient on lamellae of arbitrary shape,” Philips technical review 20, 220-224 (1958).
  20. [24]M. G. Buehler, and G. L. Pearson. “Magnetoconductive correction factors for an isotropic Hall plate with point sources,” Solid-State Electronics 9 (5), 395-407 (1966).DOI
  21. [25]George Arfken. Mathematical Methods for Physicists, 3rd edition (Academic Press Inc., Boston 1985), chapter 12.3.
  22. [27]see Ref. 21, §9.06, after eq. (9.32).
  23. [32]see Ref. 25, eq.(12.20).
  24. [34]see Ref. 25, exercise (12.10.4).
  25. [35]see Ref. 25, exercise (12.10.5).
  26. [36]Ernest William Hobson. The Theory of Spherical and Ellipsoidal Harmonics, (Chelsea Pub., N.Y. 1965), chapter II, 42., (70).

Paper details

vlnc:W_mw34sc6z
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents