Physics H7C Fall 1999

Solutions to Problem Set 2

Derek Kimball

The purpose of physics is to understand the universe... the purpose of mathemat- (b) ics is, well, obscure to me... The Ampere-Maxwell equation, since there is no real (conduction) current J be- Prof. Seamus Davis, UC Berkeley tween the plates of the capacitor, reduces to: If you have any questions, suggestions or corrections to the solutions, dont hesitate to e-mail me at dfk@uclink4.berkeley.edu! B = 0
0

E t

(5)

If youre interested in the possibility of magnetic monopoles, you might want to Then using our result from part (a) and integrating (we choose an Amperian loop look up a paper by Blas Cabrera (Physical Review Letters, vol. 48, no. 20, centered on the z-axis of radius r), we nd: 1982 pp. 1378-81), where the possible detection of a single magnetic monopole 0 Ir2 is discussed. There have been no further monopoles detected since that time, so B d = 2rB = (6) this report remains unconrmed. There is also an excellent discussion of magnetic b2 monopoles in J.D. Jacksons Classical Electrodynamics. Thus we nd the magnetic eld in the direction to be: A discussion of the additional problem presented in discussion section this week can be found in a paper by Robert Romer (American Journal of Physics vol. 50, 0 Ir . (7) B = no. 12, 1982 pp. 1089-93). 2b2 Problem 1 (a) (c)

Far from the capacitor, there is no changing electric eld and therefore only conWe use Gausss law and choose a cylindrical surface of radius r centered on the axis duction current, so this is the familiar Amperes law: (well call it z ) of the parallel plate capacitor, far from the edges of the capacitor B d = 0 Iencl , (8) (r b). Then: E dA = Qencl
0

= r2 / 0 ,

(1)

from which we nd a magnetic eld in the direction: B = 0 I . 2r (9)

where is the surface charge density of the capacitor. We nd directly from Eq. (1) that: E = (/ 0 ). z (2)

Since there is a current I, the surface charge density changes with time by an amount: d I = 2, dt b (3)

which you will note is equivalent to Eq. (7) when r b. Also note that inside the capacitor, the magnetic eld grows with r while far from the capacitor the eld falls as 1/r. (d)

Lets consider the electric eld in two dierent regions. First, well consider E far from the capacitor in the vicinity of one of the long axial leads (as in part (c)). where we assume the current is owing in the z direction. So from Eqs. (2) and The changing current produces a changing magnetic eld, and from Maxwells (3), we nd that: equations we know this creates an electric eld: dE = dt I z. 2 0 b (4) September 2, 1999 Ed = B dA. t (10)

Physics H7C Fall 1999 From Eq. (9), we see that

Solutions to Problem Set 2

Derek Kimball

We choose an Amperian loop as indicated in Fig. 1. There is no electric eld perpendicular to the wire (along r). This can be deduced from symmetry consid A C= erations. Suppose there was an electric eld in the r direction. How does it know d whether to point in the + or direction? That has to be decided by either the r r direction of the current or the change in current. If we reverse these quantities, where A is the area of the plates and d is the plate separation. So for C1 : the electric eld in the r direction should reverse. But on the opposite sides of the 0A wire, these quantities have opposite signs! The only way this can be true is if the C1 = d electric eld in the r direction is zero. Furthermore, we know that the electric eld must go to zero as r . But since E d = 0, it must be the case that we have an electric eld in the z direction which varies with r. In other words, it is apparent that the electric eld is larger closer to the wire (z-axis). This can be done explicitly, of course, from Eqs. (10) and (11): E(r2 ) E(r1 ) =

We can simplify the problem by thinking of C2 and C3 as two capacitors in series or in parallel, respectively (Fig. 1). The capacitance C of a parallel plate capacitor is given by: (15)

(16)

C2a
Axial Lead r1 r r2 Amperian Loop Figure 1

C2
z

C2b

C3

C3a

C3b

0 I ln r1 /r2 . 2 t

(12)

Lets now consider the electric eld inside the capacitor, far from the fringe (as in part (b)). Once again we apply Eq. (10), but in this case: 0 r I B = , t 2b2 t (13)

Figure 2
For C2 we break up the problem into two parts, solving for C2a and C2b (shown in Fig. 1), then determining C2 using: C2 = 1 1 + C2a C2b
1

We see that there is also a component of the electric eld in the z direction which varies with r by utilizing similar arguments as those presented above: E(r) = 0 r2 I . 4b2 t (14)

(17)

From Eq. (15) we can nd C2a and C2b , where: C2a =

(18)

September 2, 1999

Physics H7C Fall 1999 and C2b = A =2 d/2

Solutions to Problem Set 2

Derek Kimball

(28)

So with a wee bit of algebra, we nd that: C2 = 2C1 . 0/ + 1

(29)

Similarly for C3 , we break up the capacitor into two parts C3a and C3b , and then The Q-factor is the ratio of these two quantities, U and P , multiplied by the solve for C3 using: frequency: C3 = C3a + C3b . We use Eq. (15) to solve for C3a and C3b , nding: C3a = and C3b = A/2 = C1 . d 2 0 (23)
0 A/2

(21)

Q=

(30)

1 C1 2

(22)

If we plug in the numbers for seawater, we nd that Q 1.1. This suggests that decimeter waves cannot propagate very far in seawater, since the energy in the wave falls to 1/e its initial value in about one decimeter! Problem 4 First, we can write down the the electric and magnetic elds of the incident, transmitted and reected waves:

So here the overall capacitance is given by: C3 = C1 ( / 2

+ 1).

(24)

z Ei sin (ky t) xBi sin (ky t) z E0 sin (k0 y t) xB0 sin (k0 y t)

Problem 3 The energy per unit volume U stored in an electromagnetic wave is given by: U= 1 2 E2 + 1 2 B = E2. (25)

If we then time average the energy, we nd that the average energy stored is:
2 U = E0

z Er sin (ky + t) xBr sin (ky + t)

(31)

cos2 (t)dt =

2 E0 . 2

(26)

We note that k0 = nk since the transmitted wave is in glass. Then we can impose the condition |B| = | E| (32)

The average power P dissipated per unit volume is given by the relation: P = J2 , (27)

1 on each of the waves, and demand that the Poynting vector, S = 0 E B, is along the direction of propagation of the waves. This xes the amplitudes and signs of

September 2, 1999

Physics H7C Fall 1999 the magnetic elds with respect to the electric elds: z Ei sin (ky t) x 0 0 Ei sin (ky t) z E0 sin (k0 y t) x 0 E0 sin (k0 y t) z Er sin (ky + t) 0 0 Er sin (ky + t) x

Solutions to Problem Set 2

Derek Kimball

If there were magnetic charges, a magnetic charge density m and a magnetic current density Jm would appear in Maxwells equations. To avoid confusion, lets denote the traditional electric charge density e and electric current density Je . We can place both of these, with some constants c1 and c2 which will be dened later, in Maxwells equations to make them nice and symmetric: E = e / (33)
0

B = c1 m B + c2 Jm t E B = 0 0 + 0 Je t E =

Now we consider the elds at y = 0, the interface between the block of glass and vacuum. We require that the electric and magnetic elds parallel to the surface of the glass satisfy: E =E B B = . After substitution, this leaves us with two equations: Ei + Er = E0 Ei + E r =
0

(38)

(34)

We can go further and work out a relationship between magnetic charge density and current density. We begin by demanding that magnetic charges and currents satisfy the continuity equation, namely: Jm + m = 0. t (39)

E0 .

(35) Then if we take the divergence of the new third Maxwells equation, we get: E = B + c2 Jm . t (40)

We can then eliminate E0 from these equations yielding the ratio of Er to Ei : Er = Ei / /

There is a vector derivative rule that states for any vector eld A, A = 0. The energy is proportional to E 2 (as can be readily seen by considering the Poynt- So the left-hand side of (40) is 0. The derivatives on the right hand side, and , can be swapped and we get: ing vector S), and in this case the index of refraction n = / 0 . Thus the ratio t of reected to incident energy Ur /Ui is given by: B + c2 Jm = 0. (41) 2 2 t Er n1 Ur = = . (37) Ui Ei n+1 From the second Maxwell equation we know that B = c1 m , so we nd: For n = 1.6, 5% of the energy is reected. m + c2 Jm = 0. (42) c1 t If we then apply the continuity equation, Eq. (39), we nd that c1 = c2 c. September 2, 1999

+1

(36)

Physics H7C Fall 1999 Thus the nal form of Maxwells equations is: E = e /
0

Solutions to Problem Set 2

Derek Kimball

B = cm E =

B cJm Anyhow, in this case its no problem, we nd that: t E (46) B = 0 M . B = 0 0 + 0 Je , (43) t where c is a constant of proportionality between the magnetic charge unit and the inside the iron plate and zero outside the plate. magnetic eld it produces (the equivalent of 1/ 0 for electric elds). This problem now reduces to the traditional problem of solving for the Problem 6 cyclotron orbit of a moving charged particle in a magnetic eld. An imIron Plate The rst part of this problem is to portant dierence, as pointed out by M calculate the magnetic eld B inside Paul Wright in section (thanks!), is Iron Plate the magnetized iron. We can use the that in this case we need to be careauxiliary eld H to make our job a ful about relativistic corrections to little easier. We know that: the radius of the cyclotron orbit. B To nd the radius of the cyclotron H= M. (44) 0 d orbit R, we balance the Lorentz force qvB with the relativistic centrifugal Also, we have the relation: force mv 2 /R. This tells us: H d = If ree , (45) mv . (47) R= qB Amperian Loop where in our problem If ree = 0 ev where R is the radius of the circular erywhere. We choose an Amperian orbit. If you take a look at Fig. 4, loop as pictured in Fig. 3 (M is in hopefully the simple geometric arguthe z direction), taking advantage of ments suggested convince you that the planar symmetry of the problem in fact: (we can assume the iron plate is inR z nite). Since the component of H perdqBc dqBc R sin = d/R = = , pendicular to the surface of the iron mvc pc plate must be zero based on symme(48) y try, and outside the iron plate H 0 where is the deection angle and as y , we conclude that in fact Figure 4 d is the thickness of the plate. The x H = 0 everywhere. rest of the problem is working out I would like to pause here and point the correct units... out that this conclusion is not enFigure 3 tirely trivial. If there is no free curFirst lets get B in SI units. B = 0 M = 4 107 N/A2 1.5 1029 electron rent If ree , that does not necessarily magnetic moments per m3 9 1024 J/T, or about 1.7 T. Then dqcB = 108 eV, September 2, 1999

mean that H = 0 everywhere. The fundamental reason for this is that in order to completely determine a vector eld you must know both its curl and divergence. Only in cases where we have planar, cylindrical, toroidal or solenoidal symmetry can we conclude that H = 0, and get H quickly. This is dierent from Amperes law with B where we always know B = 0. So, be careful when using H!

Physics H7C Fall 1999 so sin = a deection... Problem 7 Fowles 1.4 The 3D wave equation is: 2 f = 1 2f u2 t2
dqBc pc

Solutions to Problem Set 2

Derek Kimball

= 10

rad. Thats only about half a degree, so not too big of where k = 2/ is the wave vector. We can also express in terms of u and : 2u . If you take the derivative of with respect to : = ku = d 2u 2 du = 2 + . d d Now we calculate (49)
d dk :

(55)

(56)

2 d = . (57) dk 2 We employ spherical coordinates, and since our wavefunction is a function only of If we then substitute the expressions in Eqs. (56) and (57) into Eq. (54), we arrive at our result: r, 2 is also a function only of r: u (58) ug = u . f 1 2 f = 2 r2 . (50) r r r (b) Plugging in the spherical harmonic wavefunction f = 1 e(krt) , we get: r We use similar tricks to derive the result: 2 1 1 0 dn k = . (59) (51) 2 f = e(krt) = k 2 f. ug u c d0 r If we evaluate the right-hand side of Eq. (49), and use the fact that k = /u, we nd that: 1 2f = k 2 f. u2 t2 which veries that f is a solution to the 3D wave equation. Problem 8 Fowles 1.6 (a) Lets begin by deriving ug = u u . (53) and d0 2 = 0 . (63) d 2c Substituting these results back into Eq. (60) gives us the answer we were looking for: 1 1 0 dn = . (64) ug u c d0 Thats all folks! (52) We begin by noting that dk d0 dk 1 = = ug d d0 d Lets write the wave vector in terms of 0 and n: 2n k= 0 We can then take some derivatives, and nd that: 2n 2 dn dk = 2 + d0 0 0 d0 (62) (60)

(61)

We can begin by using Fowles (1.33), the denition of the group velocity: ug = d d d = , dk d dk (54)

September 2, 1999