UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede LECTURE NOTES 16 THE MAGNETIC VECTOR POTENTIAL A ( r ) We saw in electrostatics that ∇ × E = 0 {always} (due to intrinsic / microscopic nature of the electrostatic field) permitted us to introduce a scalar potential V ( r ) such that: E ( r ) ≡ −∇V ( r ) {n.b. V ( r ) is uniquely defined, up to an (arbitrary) constant.} Analogously, in magnetostatics, the ∇i B ( r ) = 0 (always) { ⇒ ∃ no magnetic charges / no magnetic monopoles} permits us to introduce a magnetic vector potential A ( r ) such that: B (r ) ≡ ∇ × A(r ) Teslas 1 ⇒ S.I. units of the magnetic vector potential A ( r ) = Tesla-meters TeslaMeters m ( ) Then: ∇i B ( r ) = ∇i ∇ × A ( r ) = 0 {always} The divergence of a curl of a vector field F ( r ) is always zero Ampere’s Law: In differential form: ( ) ( ) ∇ × B ( r ) = ∇ × ∇ × A ( r ) = ∇ ∇i A ( r ) − ∇ 2 A ( r ) = μo J free ( r ) Now, just as in the case of electrostatics, where V ( r ) was uniquely defined up to an arbitrary constant (Vo ) , then let: V ′ ( r ) ≡ V ( r ) + Vo =0 then: E ( r ) = −∇V ′ ( r ) = −∇ (V ( r ) + Vo ) = −∇V ( r ) − ∇Vo = −∇V ( r ) i.e. E ( r ) = −∇V ′ ( r ) = −∇V ( r ) An analogous thing occurs in magnetostatics - we can add / we have the freedom to add to the magnetic vector potential A ( r ) the gradient of any scalar function A ( r ) ≡ ∇Φ m ( r ) where Φ m ( r ) ≡ magnetic scalar potential SI Units of magnetic scalar potential Φ m ( r ) = Tesla-m2 Then: A′ ( r ) ≡ A ( r ) + A ( r ) = A ( r ) + ∇Φ m ( r ) ⇐ Formally known as a Gauge Transformation The curl of the gradient of a scalar field ( Φ m ( r ) here) automatically/always vanishes, i.e.: ≡ 0 Always !!! ( ) B ( r ) = ∇ × A′ ( r ) = ∇ × A ( r ) + A ( r ) = ∇ × A ( r ) + ∇ × A ( r ) = ∇ × A ( r ) + ∇ × ∇Φ m ( r ) = ∇ × A(r ) © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 1 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Note that the magnetic scalar potential Φ m ( r ) has same physical units as magnetic flux Φ m : Tesla-m2 = Weber (Magnetic flux, Φ m = ∫ B ( r )idA !!) S eeek!!!! ⇒ Please do NOT confuse the magnetic scalar potential Φ m ( r ) (= a scalar point function, whose value can change at each/every point in space, r ) with the magnetic flux Φ m (which is a constant scalar quantity (i.e. a pure number), independent of position) Φ m ( r ) ≠ Φ m !!! Thus, like the scalar potential V ( r ) , the magnetic vector potential A ( r ) is (also) uniquely defined, but only up to an (arbitrary) vector function A ( r ) = ∇Φ m ( r ) . A′ ( r ) ≡ A ( r ) + A ( r ) = A ( r ) + ∇Φ m ( r ) The definition B ( r ) ≡ ∇ × A ( r ) specifies the curl of A ( r ) , but in order to fully specify the vector field A ( r ) , we additionally need to specify the divergence of A ( r ) , ∇i A ( r ) . We can exploit the freedom of the definition of A ( r ) to eliminate the divergence of A ( r ) - i.e. a specific choice of A ( r ) will make A ( r ) divergenceless: ∇i A ( r ) = 0 ⇐ Coulomb Gauge If: A′ ( r ) ≡ A ( r ) + A ( r ) = A ( r ) + ∇Φ m ( r ) Then: ∇i A′ ( r ) = ∇i A ( r ) + ∇iA ( r ) = ∇i A ( r ) + ∇i∇Φ m ( r ) = ∇i A ( r ) + ∇ 2 Φ m ( r ) While the original magnetic vector potential, A ( r ) is not/may not be divergenceless, we can make A′ ( r ) = A ( r ) + A ( r ) = A ( r ) + ∇Φ m ( r ) divergenceless, i.e. ∇i A′ ( r ) = 0 if we chose A ( r ) = ∇Φ m ( r ) such that ∇iA ( r ) = ∇ 2 Φ m ( r ) = −∇i A ( r ) ⇐ Coulomb Gauge A Simple Illustrative Example: Suppose ∃ a region of space that has a uniform/constant magnetic field, e.g. B ( r ) = Bo zˆ . ⎛ ∂A ( r ) ∂Ax ( r ) ⎞ Then: B ( r ) = Bo zˆ = ∇ × A ( r ) = ⎜ y − ⎟ zˆ . ∂y ⎠ ⎝ ∂x Thus (here): A ( r ) = Ax ( r ) xˆ + Ay ( r ) yˆ + Az ( r ) zˆ = Ax ( r ) xˆ + Ay ( r ) yˆ =0 If Ax ( r ) = − Bo y and Ay ( r ) = Bo x , then A ( r ) = − 12 Bo yxˆ + 12 Bo xyˆ , and thus we see that this 1 2 1 2 choice of magnetic vector potential indeed gives us the correct B -field: ⎛ ∂A ( r ) ∂Ax ( r ) ⎞ 1 1 − B (r ) = ∇ × A(r ) = ⎜ y ⎟ zˆ = 2 Bo + 2 Bo = Bo zˆ ∂ ∂ x y ⎝ ⎠ 2 © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede ⎛ ∂A ∂A ∂A ⎞ ⎛ ∂ ( − 12 Bo y ) ∂ ( 12 Bo x ) ∂ ( 0 ) ⎞ Is ∇i A = 0 satisfied? ∇i A = ⎜ x + y + z ⎟ = ⎜ + + ⎟ = 0 Yes!!! ∂z ⎟⎠ ∂y ∂y ∂z ⎠ ⎜⎝ ∂x ⎝ ∂x Note that we could also have instead chosen/used a different magnetic vector potential: A′ ( r ) = A ( r ) + A ( r ) = A ( r ) + ∇Φ m ( r ) where e.g. A ( r ) = ∇Φ m ( r ) = Ao , i.e. where Ao is any (arbitrary) constant vector, Ao = Aox xˆ + Aoy yˆ + Aoz zˆ . Since (here) A ( r ) = ∇Φ m ( r ) = Ao , then A ( r ) = ∇Φ m ( r ) = Ao = Aox xˆ + Aoy yˆ + Aoz zˆ means that the gradient of the magnetic scalar ∂ ( Aoy y ) ∂ ( Aox x ) ∂ ( Aoz z ) xˆ + yˆ + zˆ = Aox xˆ + Aoy yˆ + Aoz zˆ = Ao = A ( r ) ∂x ∂y ∂z and thus the magnetic scalar potential itself (here) is: Φ m ( r ) = Aox xxˆ + Aoy yyˆ + Aoz zzˆ . potential (here) is: ∇Φ m ( r ) = Thus, here for the case of a constant/uniform magnetic field B ( r ) = Bo zˆ we see that there is in fact a continuum of allowed magnetic vector potentials A′ ( r ) = A ( r ) + Ao = A ( r ) + ∇Φ m ( r ) that simultaneously satisfy B ( r ) = Bo zˆ = ∇ × A′ ( r ) and ∇i A′ ( r ) = 0 with the addition of an (arbitrary) constant magnetic vector potential Ao = Aox xˆ + Aoy yˆ + Aoz zˆ contribution with corresponding magnetic scalar potential Φ m ( r ) = Aox xxˆ + Aoy yyˆ + Aoz zzˆ . Note that this is exactly analogous to the situation in electrostatics where the scalar electric potential V ( r ) is unique, up to an arbitrary constant, Vo because there exists no absolute voltage reference in our universe – i.e. absolute measurements of the scalar electric potential are meaningless - only potential differences have physical significance!!! We used this simple example of the constant/uniform magnetic field B ( r ) = Bo zˆ to elucidate this particular aspect of the magnetic vector potential A ( r ) . Here in this particular example, we found that the addition of an arbitrary constant vector A ( r ) = Ao = Aox xˆ + Aoy yˆ + Aoz zˆ = ∇Φ m ( r ) to the magnetic vector potential A ( r ) was allowed, i.e. A′ ( r ) = A ( r ) + A ( r ) = A ( r ) + Ao , which leaves the magnetic field B ( r ) unchanged. In general there are many instances involving more complicated physics situations, where B ( r ) ≠ constant vector field, where indeed B ( r ) = ∇ × A′ ( r ) and ∇i A′ ( r ) = 0 are simultaneously satisfied for A′ ( r ) = A ( r ) + A ( r ) , because it is possible to determine/find a corresponding magnetic scalar potential Φ m ( r ) for the problem satisfying ∇ 2 Φ m ( r ) = −∇i A ( r ) , but it is (very) important to understand that, in general, the allowed A ( r ) = ∇Φ m ( r ) (very likely) may not be simply a constant vector field, but indeed one which varies in space (i.e. with position vector, r )! Here again, however, the new A′ ( r ) = A ( r ) + A ( r ) will also be such that B ( r ) = ∇ × A′ ( r ) will be unchanged, exactly analogous to V ′ ( r ) = V ( r ) + Vo leaving E ( r ) = −∇V ′ ( r ) unchanged. © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 3 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede So we see that if ∇ 2 Φ m ( r ) = −∇i A ( r ) then yes, ∇i A′ ( r ) = 0 . It is always possible to find an A ( r ) = ∇Φ m ( r ) in order to make ∇i A′ ( r ) = 0 . Note however that this situation is then formally mathematically identical to Poisson’s Equation, for the magnetic scalar potential Φ m ( r ) because: ∇ 2 Φ m ( r ) = − ρ m ( r ) ⇐ Analogous to ∇ 2V ( r ) = − Equivalent magnetic volume charge density ρTot ( r ) in electrostatics!!! ε0 Physically, ρ m ( r ) could e.g. be due to bound effective ⇐ magnetic charges associated with a magnetic material… If we assume that the equivalent magnetic volume charge density, ρ m ( r ) ≠ 0 and we want ∇i A′ ( r ) = 0 Then: ∇i A ( r ) + ∇iA ( r ) = ∇i A ( r ) + ∇ 2 Φ m ( r ) = 0 Or: ∇i A ( r ) − ρ m ( r ) = 0 ⇒ ∇i A ( r ) = ρ m ( r ) Then, the solution to Poisson’s equation for the magnetic scalar potential Φ m ( r ) is of the form: Φm ( r ) = 1 4π ∫ v′ ρm ( r′) r dτ ′ ⇐ Analogous to V ( r ) = 1 4πε 0 ∫ v′ ρTot ( r ′ ) r dτ ′ in electrostatics with r = r − r ′ (n.b. these two relations are both valid assuming that ρ m ( r ′ ) and ρToT ( r ′ ) vanish when r ′ → ∞ !) So then if ρ m ( r ′ ) = ∇i A ( r ′ ) , and ρ m ( r ′ ) = ∇i A ( r ′ ) vanishes as r ′ → ∞ , then the magnetic scalar potential Φ m ( r ) is given by: Φm ( r ) = 1 4π ∫ v′ ρm ( r′) r dτ ′ = 1 4π ∇i A ( r ′ ) dτ ′ v′ r ∫ {Note that if ∇i A ( r ′ ) = ρ m ( r ′ ) does not go to zero at infinity, then we’ll have to use some other means in order to obtain an appropriate Φ m ( r ) , e.g. in an analogous manner to that which we’ve had to do for the (electric) scalar potential V ( r ) associated with problems that have electric charge distributions extending out to infinity.} Thus, this choice of Φ m ( r ) ensures that indeed we can always make the magnetic vector potential A′ ( r ) divergenceless, i.e. the condition ∇i A′ ( r ) = 0 (Coulomb Gauge) can always be met, for the case of magnetostatics. Note that if ρ m ( r ′ ) = 0 then ρ m ( r ′ ) = ∇i A ( r ′ ) = 0 . 4 © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede With the choice of the magnetic scalar potential: 1 4π Φm ( r ) = ∫ ρm ( r′) v′ r dτ ′ and A′ ( r ) = A ( r ) + Ao = A ( r ) + ∇Φ m ( r ) , and ∇i A′ ( r ) = 0 Then Ampere’s Law (in differential form) becomes: ( ) ( ) ∇ × B ( r ) = ∇ × ∇ × A ′ ( r ) = ∇ ∇ i A ′ ( r ) − ∇ 2 A′ ( r ) = μ 0 J free ( r ) =0 Which gives: ∇ A′ ( r ) = − μ0 J free ( r ) ⇐ Vector form of Poisson’s equation for magnetostatics. 2 i.e.: ⎧∇ 2 Ax′ ( r ) = − μo J x free ( r ) ⎫ The three separate/independent scalar forms ⎪⎪ 2 ⎪⎪ ⎨∇ A′y ( r ) = − μo J y free ( r ) ⎬ ⇐ of Poisson's equation are connected by: ⎪ 2 ⎪ J free ( r ) = J x free ( r ) xˆ + J y free ( r ) yˆ + J z free ( r ) zˆ ⎩⎪∇ Az′ ( r ) = − μo J z free ( r ) ⎭⎪ n.b. in Cartesian coordinates: ∇ 2 A′ ( r ) = ( ∇ 2 Ax′ ( r ) ) xˆ + ( ∇ 2 A′y ( r ) ) yˆ + ( ∇ 2 Az′ ( r ) ) zˆ However, in curvilinear coordinates (i.e. spherical-polar or cylindrical coordinates) ⎧rˆ = sin θ cos ϕ xˆ + sin θ sin ϕ yˆ + cos θ zˆ ⎫ ⎪ ⎪ e.g. spherical-polar coordinates: ⎨θˆ = cos θ cos ϕ xˆ + cos θ sin ϕ yˆ − sin θ zˆ ⎬ ⎪ϕˆ = − sin ϕ xˆ + cos ϕ yˆ ⎪ ⎩ ⎭ Note that the unit vectors rˆ,θˆ + ϕˆ for spherical-polar coordinates are in fact explicit functions of the vector position, r i.e. rˆ = rˆ ( r ) , θˆ = θˆ ( r ) and ϕˆ = ϕˆ ( r ) and therefore rˆ, θˆ + ϕˆ must also be explicitly differentiated in calculating the Laplacian ∇ 2 of a vector function (here, A′ ( r ) ) in curvilinear (i.e. either spherical-polar and/or cylindrical) coordinates!!! This is extremely important to keep in mind, for the future… ⇒ The safest way to calculate the Laplacian of a vector function ∇ 2 A ( r ) in terms of curvilinear coordinates, is to NOT use curvilinear coordinates!!! Failing that, then one should use: ( ) ( ) ( ∇ 2 A ( r ) = ∇ ∇i A ( r ) − ∇ × ∇ × A ( r ) = −∇ × ∇ × A ( r ) ) = 0 in the Coulomb Gauge If ρ m ( r ′ ) = 0 then (automatically) ρ m ( r ′ ) = ∇i A ( r ′ ) = 0 and we can use A ( r ) directly. Hence, if ∇ 2 A ( r ) = − μ0 J free ( r ) (vector Poisson equation for magnetostatics), then if J free ( r ) → 0 as r → ∞ , then A ( r ) = μ0 4π ∫ v′ J free ( r ) r dτ ′ where r = r = r − r ′ © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 5 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Generalizing this for a moving point charge as well as for line, surface and volume current densities (with B ( r ) = ∇ × A ( r ) ), we summarize these results in the following table: v ( r ′) μo q free 4π r I free ( r ′ ) μ A(r ) = o ∫ d ′ 4π C ′ r μ d ′ = o I free ∫ C′ r 4π A(r ) = A(r ) = μ0 4π ∫ A(r ) = μo 4π ∫ K free ( r ′ ) S′ r J free ( r ′ ) v′ r B (r ) = v ( r ′ ) × rˆ μo q free 4π r2 B (r ) = μo 4π = ∫ (I free ( r′) d ( μo I free ∫ ′ C 4π B (r ) = μo 4π ∫ dτ ′ B (r ) = μo 4π ∫ ) 2 C′ da′ ′ × rˆ ( r d ′ ( r ′ ) × rˆ r K free ( r ′ ) × rˆ 2 S′ ( v′ ) 2 r J free ( r ′ ) × rˆ r 2 ) da′ ) dτ ′ Note that: A v , I , d , K , J i.e. A is always parallel to the direction of motion of current, with ( ) relative velocity v , whereas B = ∇ × A ⊥ v , I , d , K , J . Note also that B and A both vanish when v → 0 (e.g. in the rest frame of a current (e.g. a proton or an electron beam)). A Tale of Two Reference Frames: For a pure point electric charge/point electric monopole moment, q we know that if it is moving in the lab frame with speed v << c {c = speed of light in vacuum} that the magnetic field Bq ( r ) observed in the lab frame is: Bq ( r ) = ∇ × Aq ( r ) = q ⎛ rˆ ⎞ μ ⎛ rˆ ⎞ 1 v × Eq ( r ) = v × 2 ⎟ = o ⎜ qv × 2 ⎟ 2 2 ⎜ c r ⎠ 4π ⎝ r ⎠ 4πε 0 c ⎝ ( ) Thus in the lab frame where this charged particle is moving, the magnetic vector potential Aq ( r ) associated with this moving charged particle (as observed in the lab frame) has a non-zero curl. Contrast this with the situation in the rest frame of this pure point electric charge particle, where the magnetic field vanishes, i.e. the magnetic vector potential Aq ( r ) associated with this charged particle has no curl!!! 1 ∂V ( r , t ) in electrodynamics. c2 ∂t ⇒ ∃ connection between the A ( r , t ) − field and electric scalar potential V ( r , t ) - they are in fact We will find out (next semester, in P436) that: ∇i A′ ( r , t ) = − the 3 spatial & 1 temporal components of the relativistic 4-potential in electrodynamics !!! 6 © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Uses of the Magnetic Scalar Potential Φ m ( r ) : In certain (limited) circumstances for magnetostatics, it is actually possible to have the magnetic field B ( r ) directly related to the (negative) gradient of a magnetic scalar potential Φ m ( r ) , i.e. B ( r ) = − μo∇Φ m ( r ) , in direct analogy to that for electrostatics E ( r ) = −∇V ( r ) . However, while ∇i B ( r ) = − μo∇i∇Φ m ( r ) = − μo ∇ 2 Φ m ( r ) = 0 is satisfied, i.e. ∇ 2 Φ m ( r ) = 0 is Laplace’s equation for the magnetic scalar potential, Φ m ( r ) (n.b. implying that ρ m ( r ) ≡ 0 ), Ampere’s law ∇ × B ( r ) = − μo ∇ × ∇Φ m ( r ) = μo J ( r ) is not satisfied/is violated (!!!) unless ≡ 0 Always !!! J ( r ) ≡ 0 everywhere in the region(s) of interest. These current-free regions must also be simplyconnected. {A region D (e.g. in a plane) is connected if any two points in the region can be connected by a piecewise smooth curve lying entirely within D. A region D is a simply connected region if every closed curve in D encloses only points that are in D.} The use of B ( r ) = −∇Φ m ( r ) is in fact helpful for determining the magnetic fields associated with e.g. current-carrying filamentary wires, current loops/magnetic dipoles, and e.g. the magnetic fields associated with magnetized materials/magnetized objects. The SI Units of the magnetic vector potential A are Tesla-meters (= magnetic field strength per unit length), which is also equal to Newtons/Ampere (force per unit current) = kg-meter/Ampere-sec2 = kg-meter/Coulomb-sec = (kg-meter/sec)/Coulomb = momentum per unit charge, since (kg-meter/sec) are the physical units associated with momentum p = " mv " . Thus, for the A -field: 1 Tesla-meter = 1 unit of force momentum = 1 unit of Ampere of current Coulomb of charge and for the B -field, from B ( r ) = ∇ × A ( r ) : 1 Tesla = 1 unit of force N momentum meter = = meter Ampere of current A − m Coulomb of charge Physically, the A -field has units of force per Ampere of current (or momentum per Coulomb of electric charge), and thus physically, the magnetic field B ( r ) = ∇ × A ( r ) is the curl of the force per unit current (or momentum per unit charge) field. Note also that force, F = dP dt and current, I = dQ dt such that the magnetic vector potential A physically also has units of ⎛ Force ⎞ F dp / dt dp / dt Δp = = and thus B is the curl of this physical quantity. ⎜ ⎟= = ⎝ Current ⎠ I dq / dt dq / dt Δq © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 7 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede The Magnetic Vector Potential of a Long Straight Wire Carrying a Steady Current For a filamentary wire carrying steady current I, the magnetic vector potential and magnetic field d ′ × rˆ d ′ ⎛μ ⎞ ⎛μ ⎞ B (r ) = ∇ × A(r ) = ⎜ o ⎟ I ∫ and are: A ( r ) = ⎜ o ⎟ I ∫ ′ r2 ⎝ 4π ⎠ C r ⎝ 4π ⎠ C ′ Let the length of wire = 2L, and I = Izˆ and thus d ′ = d ′zˆ = dzzˆ ( r = r = r − r′ = R2 + ) 2 The infinitesimal magnetic vector potential, dA ( r = Ryˆ ) due to the current-carrying line segment d ′ carrying current I is: ⎛μ dA ( r = Ryˆ ) = ⎜ o ⎝ 4π ⎞ d ′ ( r ′ ) ⎛ μo =⎜ ⎟I r ⎠ ⎝ 4π ⎞ d ′ ( r′) ⎟I ⎠ R2 + 2 The corresponding infinitesimal magnetic field increment is: d ′ ( r ′ ) ⎛ μo ⎞ ⎛ dzzˆ ⎞ ⎛μ ⎞ =⎜ dB ( r = Ryˆ ) = ∇ × dA ( r = Ryˆ ) = ⎜ o ⎟ I ∇ × ⎟ ⎟ I ⎜∇× r ⎝ 4π ⎠ ⎝ 4π ⎠ ⎝ R2 + 2 ⎠ d ⎛ μ ⎞ z =+ L dzzˆ ⎛μ ⎞ L = 2⎜ o ⎟ ∫ A ( r = Ryˆ ) = ∫ dA ( r = Ryˆ ) = ⎜ o ⎟ I ∫ 2 2 =− 0 C′ z L ⎝ 4π ⎠ ⎝ 4π ⎠ R + R2 + 2 Now: L ⎛μ ⎞ ⎛μ ⎞ = ⎜ o ⎟ I ln ⎡ + R 2 + 2 ⎤ zˆ = ⎜ o ⎟ I ln ⎡ L + R 2 + L2 ⎤ − ln R ⎣ ⎦0 ⎣ ⎦ ⎝ 4π ⎠ ⎝ 4π ⎠ { } More generally, for a ⊥ distance R away from a long straight wire of length 2L carrying a steady current I: 2 ⎫⎤ ⎡⎛ L ⎞ ⎧ ⎛μ ⎞ A ( r = Ryˆ ) = ⎜ o ⎟ I ln ⎢⎜ ⎟ ⎨1 + 1 + R ⎬⎥ zˆ L R ⎝ ⎠ ⎝ 4π ⎠ ⎩ ⎭⎦ ⎣ ( ) ⎛μ If L >> R, then: A ( r = Ryˆ ) ≈ ⎜ o ⎝ 4π ⎞ ⎛ 2L ⎞ ⎟ zˆ {Since 1 + ε ≈ 1 for ε ⎟ I ln ⎜ ⎝ R ⎠ ⎠ 1 .} Note that if L → ∞ (or R → 0), then A ( r = Ryˆ ) diverges (logarithmically)! This is OK (unphysical anyway!) because even if A → ∞, B = ∇ × A ≠ ∞ !!! (necessarily!!!) 8 © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede So, for a ⊥ distance R away from a long straight wire of length 2L carrying a steady current I: ⎛μ A ( r = Ryˆ ) = ⎜ o ⎝ 4π 2 ⎫⎤ ⎡⎛ L ⎞ ⎧ ⎞ R ln 1 1 I zˆ = Az ( r = Ryˆ ) zˆ + + ⎨ ⎢⎜ ⎟ ⎟ L ⎬⎥ ⎠ ⎭⎦ ⎣⎝ R ⎠ ⎩ ( ) Then B ( r ) = ∇ × A ( r ) Let’s do this in cylindrical coordinates: (note: ρ = x 2 + y 2 = R here): =0 =0 ⎛ ⎜ 1 ∂Az ∂Aϕ − B ( ρ = R) = ⎜ ∂z ρ ϕ ∂ ⎜ ⎝ ρˆ B ( ρ = R) = Or: =0 ⎞ ⎛ =0 ⎞ ⎛ =0 ∂Aρ 1⎜ ∂ ⎟ˆ ⎜ ∂Aρ ∂Az ⎟ ˆ +⎜ − + ⎜ ( ρ Aϕ ) − ∂ϕ ⎟ρ ⎟ϕ ∂ ∂ ∂ z ρ ρ ρ ⎟ ⎜ ⎟ ⎜ ⎠ ρ =R ⎝ ⎠ ρ =R ⎝ ϕˆ 1 ∂ R ∂ρ 0 zˆ ∂ ∂z Az 0 0 In Cylindrical Coordinates: =− ∂Az ϕˆ ∂ρ ρ = R ẑ ρ =R ϑ Bρ = 0 Bz = 0 x̂ Then: B ( ρ = R ) = − ∂Az ϕˆ ∂ρ ρ =R ( Now if: U ( R ) ≡ L + L2 + R 2 dU ( R ) = dR ϕ̂ ρ̂ ) ( ⎛μ ⎞ ∂ ⎡ = −⎜ o ⎟ I ln L + L2 + R 2 − ln R ⎤ ϕˆ ⎢ ⎥⎦ ⎝ 4π ⎠ ∂R ⎣ ) R 1 ⎡ dU ( R ) ⎤ ∂ ln (U ( R ) ) = ⎢ ⎥= ∂R U ( R ) ⎣ dR ⎦ L + L2 + R 2 Since: ŷ ϕ 1 ∂Az ∂Az =− ϕˆ ϕˆ Thus: Bϕ = i R ∂ρ ρ = R ∂ρ ρ = R R Then: ⎞ ∂Az ⎟ˆ =− ϕˆ ⎟z ∂ ρ ρ =R ⎟ ⎠ ρ =R ( R L2 + R 2 and: )( L2 + R 2 ) ∂ 1 ln ( R ) ) = ( ∂R R ⎧ ⎫ B-field associated ⎪ ⎪ with filamentary μ R 1 ⎪ ⎛ ⎞ ⎪ − ⎬ ϕˆ ⇐ wire of length 2L Then finally: B ( ρ = R ) = − ⎜ o ⎟ I ⎨ 2 ⎛ 2 ⎞ R⎪ ⎝ 2π ⎠ ⎪ 2 carrying steady L 1+ R 1+ 1+ R ⎜ ⎟ L L current I. ⎪⎩ ⎪ ⎝ ⎠ ⎭ ⎛μ ⎞I Note that as L → ∞, then B ( ρ = R ) = − ⎜ o ⎟ ϕˆ ⇐ i.e. exactly the same as we obtained for ⎝ 2π ⎠ R ∞-long straight filamentary wire carrying steady current I (see previous P435 Lecture Notes)!!! ( ) ( ) © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 9 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede The Magnetic Vector Potential A ( r ) and Magnetic Field B ( r ) = ∇ × A ( r ) Associated with a Pair of Long, Parallel Wires Carrying Steady Currents I1 = + Izˆ and I 2 = − Izˆ , Separated by a Perpendicular Separation Distance, d Plane ⊥ to wires and containing ⊥ separation distance d d 1 = + d zˆ = + dzzˆ r 1 = r1 − r1′ = r − r1′ d 2 = − d zˆ = − dzzˆ r 2 = r2 − r2′ = r − r2′ ( r1 = r2 = r ) For simplicity’s sake, assume L >> R1, R2. ⎛μ Then: A1 ( R1 ) ≈ + ⎜ o ⎝ 2π ⎛ 2L ⎞ ⎞ ⎟ zˆ and ⎟ I ln ⎜ ⎠ ⎝ R1 ⎠ ⎛μ A2 ( R2 ) ≈ − ⎜ o ⎝ 2π ⎛ 2L ⎞ ⎞ ⎟ zˆ ⎟ I ln ⎜ ⎠ ⎝ R1 ⎠ Then, using the principle of linear superposition: ⎛ 2 L ⎞ ⎛ μo ⎛μ ⎞ ATOT ( r ) = A1 ( r ) + A2 ( r ) ≈ + ⎜ o ⎟ I ln ⎜ ⎟ zˆ − ⎜ ⎝ 2π ⎠ ⎝ R1 ⎠ ⎝ 2π ⎛ 2L ⎞ ⎞ ⎟ zˆ ⎟ I ln ⎜ ⎠ ⎝ R2 ⎠ 2 Or: 10 ⎛R ⎞ ⎛R ⎞ ⎛μ ⎞ ⎛μ ⎞ ATOT ( r ) ≈ ⎜ o ⎟ I ln ⎜ 2 ⎟ zˆ = ⎜ o ⎟ I ln ⎜ 2 ⎟ zˆ for L >> R1, R2. ⎝ 2π ⎠ ⎝ 4π ⎠ ⎝ R1 ⎠ ⎝ R1 ⎠ © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Now let us re-locate the local origin to be at the LHS wire, where it intersects the ⊥ -plane: Top View I1 (out of page) d I 2 (into page) R12 = x 2 + y 2 d-y y R1 x Then: ATOT ( r ) R22 = x 2 + ( d 2 − y 2 ) R2 ⎛ μo ⎜ 4π ⎝ 2 ⎛ R2 ⎞ ⎞ ⎛ μo ⎟ I ln ⎜ R ⎟ zˆ = ⎜ 4π ⎠ ⎝ ⎝ 1⎠ ⎡ x 2 + ( d − y )2 ⎤ ⎞ ⎟ I ln ⎢ x 2 + y 2 ⎥ zˆ for L >> R1, R2. ⎠ ⎢⎣ ⎥⎦ ATOT ( y = 0 ) diverges (x = 0) ( ATOT ( r ) (for x = 0 i.e. observation point on ŷ -axis) y = 0 ATOT y = d 2 ) = 0 for I y = d/2 1 = − I 2 = Izˆ y=d y ATOT ( y = d ) diverges (x = 0) Note that ATOT = Az zˆ i.e. AxTOT = ATOT = 0 (since currents only in ± ẑ -direction) !!! y Note also that ATOT = Az zˆ changes sign – its direction is parallel to the closest current!!! Then: BTOT ( r ) = ∇ × ATOT ( r ) , in Cartesian ( xˆ − yˆ − zˆ ) coordinates: TOT x B ∂AzTOT ⎛ μ ⎞ ⎡(d − y) y ⎤ =+ = −⎜ o ⎟ I ⎢ + 2⎥ 2 R1 ⎦ ∂y ⎝ 2π ⎠ ⎣ R2 BTOT =+ y ∂AzTOT ⎛μ = −⎜ o ∂x ⎝ 2π x ⎤ ⎞ ⎡ x ⎟I ⎢ 2 − 2 ⎥ ⎠ ⎣ R2 R1 ⎦ R12 = x 2 + y 2 R22 = x 2 + ( d 2 − y 2 ) BzTOT = 0 © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 11 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede At the point x = 0 and y = d/2 (i.e. at the center point, midway between the two conductors): {where R1 = R2 = R = d/2}: TOT x B ⎛μ = −⎜ o ⎝ 2π ⎡ d d ⎤ ⎞ ⎢ 2 2 ⎥ = − ⎛ μo + ⎟I ⎢ ⎜ 2 2⎥ ⎠ ⎝ 2π d d ⎢⎣ 2 2 ⎥⎦ ( ) ( ) 2 ⎡ ⎛ μo ⎞ 4 1 ⎤ ⎛ 2 μo ⎞ I ⎞ ⎢ 1 ⎟ ⎟ I = −⎜ ⎟ I d + d ⎥ = −⎜ 2π ⎠ d ⎥ ⎝ π ⎠d ⎠ ⎢ ⎝ ⎣ 2 2⎦ BTOT =0 y BzTOT = 0 Top View: 2μ ⎛ I ⎞ BTOT x = 0, y = d , z = 0 = − o ⎜ ⎟ xˆ 2 π ⎝d ⎠ ( ) The Magnetic Vector Potential A ( r ) and Magnetic Field B ( r ) = ∇ × A ( r ) Associated with a Magnetic Dipole Loop (For Large Source-Observer Separation Distances) For simplicity’s sake, let us choose the observation / field point P ( r ) to lie in the x-z plane: Observation / field Point P ( r ) = P ( x, 0, z ) Magnetic Dipole Loop has radius R R = r′ I = Iϕˆ 12 © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 d ′ = Rdϕϕˆ (d Lecture Notes 16 Prof. Steven Errede cos Ψ = rˆirˆ′ = opening angle between r̂ and r̂ ′ r i r ′ r ir ′ r ir ′ = = cos Ψ = where: r ′ = r ′ = R r i r ′ r ir ′ rR ′ = Rdϕ ) {from the arc length formula “ S = Rθ ”} Id ′ ( r ′ ) ⎛ μo ⎛μ ⎞ =⎜ Now: A ( r ) = ⎜ o ⎟ ∫ ′ r ⎝ 4π ⎠ C ⎝ 4π d ′ ( r ′) ⎞ ⎟ I ∫C′ r ⎠ ⇒ A ( r ) (here) is a function of xˆ and yˆ only (actually only ϕ̂ ) and not ẑ since Id ′ = Id ′ lies in the x-y plane and I = Iϕˆ (here). Since we evaluate A in the x-z plane, the only component of d ′ that will contribute to A (there) will be in the ŷ direction (n.b. A is parallel to the closest current from the observation point P(r)). ⇒ We only want the component of d ′ ( r ′ ) along the ŷ -axis, d ( r ′) cos ϕ (Note: If we wanted to evaluate A e.g. in the y-z plane, then we would want only the component of d ′ ( r ′ ) along the x̂ -axis, d ( r ′) sin ϕ ) ⎛ μ ⎞ 2π ( Rdϕ ) cos ϕ ϕˆ ← ϕˆ = yˆ in the x-z plane for r = ( x, 0, z ) . Then: A ( r ) = A ( x, 0, z ) = ⎜ o ⎟ I ∫ r ⎝ 4π ⎠ 0 Now: r 2 = r 2 + R 2 − 2rR cos Ψ (from the Law of Cosines) and r = r − r ′ 1 And: And: 2 ⎛ R 2 2rR ⎞ 2 1⎛ R⎞ ⎛R⎞ = ⎜ 1 − 2 + 2 cos Ψ ⎟ ≈ 1 − ⎜ ⎟ + ⎜ ⎟ cos Ψ if R r ⎝ r r 2⎝ r ⎠ ⎝ r ⎠ ⎠ r ir ′ = rr ′ cos Ψ = ( xxˆ + zzˆ )i( R cos ϕ xˆ + R sin ϕ yˆ ) = xR cos ϕ r r, r and r ≈ r . (if the observation / field point P ( r ) = P ( x, 0, z ) lies in the x-z plane) 2 1 ⎛ R ⎞ xR cos ϕ 1− ⎜ ⎟ + Thus: for r >> R and r r = r − r ′ 2⎝ r ⎠ r2 r ⎛ 1 ⎛ R ⎞ 2 xR ⎞ ⎛ μ ⎞ I ϕ = 2π R cos ϕ ⎜1 − ⎜ ⎟ + 2 cos ϕ ⎟ dϕ yˆ Then: A ( r ) = A ( x, 0, z ) ⎜ o ⎟ ∫ ⎜ 2⎝ r ⎠ ⎟ r ⎝ 4π ⎠ r ϕ =0 ⎝ ⎠ 2 ⎛ μ ⎞ Iπ R A ( r ) = A ( x, 0, z ) ⎜ o ⎟ 3 xyˆ for r >> R and r r = r − r ′ Or: ⎝ 4π ⎠ r r © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 13 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 But notice that in the x-z plane: P ( r ) = P ( x, 0, z ) Prof. Steven Errede ẑ ⎛x⎞ sin θ = ⎜ ⎟ ⎝r⎠ θ r z ŷ out of page x x̂ ⎛ μ ⎞ Iπ R A ( r ) = A ( x, 0, z ) ⎜ o ⎟ 2 sin θ yˆ for r >> R and r r = r − r ′ ⎝ 4π ⎠ r But in the x-z plane, ϕˆ = ŷ , and since the direction of A ( r ) is always parallel to the current: 2 ∴ ∴ A(r ) 2 ⎛ μo ⎞ I π R ⎜ ⎟ 2 sin θϕˆ for r >> R and r ⎝ 4π ⎠ r r = r − r ′ {n.b. A ( r ) I ϕˆ } The magnetic dipole moment associated with this current-carrying loop is m = mzˆ m = Ia = Iazˆ (for this planar loop) where (here): a = π R 2 zˆ (by the right-hand rule) m = I π R 2 zˆ and: m = m = Ia = I π R 2 = π R 2 I ( ) ( ) Note that: zˆ × rˆ = cos θ rˆ − sin θθˆ × rˆ = − sin θ θˆ × rˆ = − sin θ ( −ϕˆ ) = + sin θϕˆ Thus, the quantity: m × rˆ = Iazˆ × rˆ = I π R 2 ( zˆ × rˆ ) = I π R 2 sin θϕˆ ∴ ⎛ μ ⎞ m × rˆ ⎛ μ ⎞ m × r A ( r ) ⎜ o ⎟ 2 = ⎜ o ⎟ 3 for r >> R and r ⎝ 4π ⎠ r ⎝ 4π ⎠ r r = r − r′ Now B ( r ) = ∇ × A ( r ) in spherical coordinates for the magnetic dipole (with magnetic dipole moment m = Ia = I π R 2 zˆ ) is: ⎛ μ ⎞ 2m Br ( r ) = ⎜ o ⎟ 3 cos θ ⎝ 4π ⎠ r ⎛μ ⎞m Bθ ( r ) = ⎜ o ⎟ 3 sin θ ⎝ 4π ⎠ r valid for r >> R and r r = r − r′ Bϕ ( r ) = 0 B ( r ) = Br ( r ) rˆ + Bθ ( r ) θ + Bϕ ( r ) ϕˆ ⎛μ Thus: B ( r ) = ⎜ o ⎝ 4π ( ⎞m ˆ ⎟ 3 2 cos θ rˆ + sin θθ ⎠r ) for r >> R and r r = r − r ′ and m = Ia = I π R 2 zˆ . cf w/ the E -field associated with a physical electric dipole with dipole moment p = qd : ⎛ 1 E (r ) = ⎜ ⎝ 4πε o 14 ⎞ p ⎟ 3 2 cos θ rˆ + sin θθˆ for r >> d and r ⎠r ( ) r = r − r ′ and p = qd . © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede We already know what this B ( r ) looks like it – it is “solenoidal” around the current loop: Cross-Sectional View of a Magnetic Dipole Loop: ⎛μ Thus if m = Ia = I π R 2 zˆ = a constant vector, A ( r ) ⎜ o ⎝ 4π m× r ⎛μ ⎞ B ( r ) = ∇ × A ( r ) = ⎜ o ⎟ ∇ × 3 for r >> R and r r r ⎝ 4π ⎠ ⎡ ⎢ ⎛ r ⎛ μo ⎞ ⎢ ⎜ ∇i r 3 − ∇ + i B (r ) = ⎜ m m ⎟⎢ 3 r ⎜ ⎝ 4π ⎠ r ⎝ ⎢ =0 ⎣ ( e.g. mx and: ∇i r r ∴ 3 = ) ⎞ m× r ⎟ 3 for r >> R and r ⎠ r r = r − r′ = r − r′ ⎤ ⎞⎥ ⎛ ⎞ ⎟ ⎥ ⇐ ⎜ ∇i r 3 = 0 ⎟ ⎜ r ⎟ ⎟⎥ ⎝ ⎠ ⎠⎥ ⎦ 3 ( mi r ) r 3 ( mirˆ ) rˆ − m r m ∂ ⎛ r ⎞ mx xˆ r =− ⎜ 3 ⎟ = 3 − 3mx x 5 ∴ mi∇ 3 = 3 − 5 3 ∂x ⎜⎝ r ⎟⎠ r r r r r r ( 3 r 3 − ri 3r r 5 = ⎛μ B (r ) = ∇ × A(r ) = ⎜ o ⎝ 4π 3 r 3 − 3r ir r 5 = 3 r 3 m × r ⎛ μo ⎞ ⎟∇× 3 = ⎜ ⎠ ⎝ 4π r for r >> R (far away) and r − ) 3r r 2 5 = 3 r 3 − 3 r 3 ⎞ ⎡ 3 ( mirˆ ) rˆ − m ⎤ ⎥ ⇐ ⎟⎢ 3 ⎠ ⎢⎣ r ⎥⎦ =0 Magnetic Field of a Magnetic Dipole Loop r = r − r′ ⇒ The magnetic field of a distant circuit (r >> R) does not depend on its detailed geometry, but only its magnetic dipole moment, m !!! ⇐ Important (conceptual) result!!! © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 15 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Compare this result for B ( r ) for the magnetic dipole loop, with magnetic dipole moment m = Ia , with result for the electric dipole field E ( r ) associated with a physical electric dipole moment p = qd : ⎛μ B (r ) = ⎜ o ⎝ 4π ⎞ ⎡ 3 ( mirˆ ) rˆ − m ⎤ ⎥ ⎟⎢ 3 ⎠ ⎢⎣ r ⎥⎦ m = Ia , for r R (far away) and r r = r − r′ ⎛ 1 ⎞ ⎡ 3 ( p irˆ ) rˆ − p ⎤ E (r ) = ⎜ ⎥ p = qd , for r ⎟⎢ 3 r ⎝ 4πε o ⎠ ⎢⎣ ⎦⎥ d (far away) and r r = r − r′ +q When r ≈ R (or less) for magnetic dipole loop Or r ≈ d (or less) for electric dipole, then will be able to “see” / observe / detect higherorder moments - e.g. quadrupole, octupole, sextupole, etc. . .moments of the B E fields. p = qd d ( ) −q The statement that the magnetic field of a distant circuit (r >> R) does not depend on its detailed geometry, but only its magnetic dipole moment, m = Ia (n.b. this is also true for the electrostatic case, with p = qd ) are very useful!!! If one can compute m = Ia then one can obtain A ( r ) and hence B ( r ) = ∇ × A ( r ) (or if have p = qd then can obtain E ( r ) ) for r >> R (or d) and r r = r − r ′ . EASY!!! Magnetic Flux Conservation ( ) If B ( r ) = ∇ × A ( r ) then ∇i B ( r ) = ∇i ∇ × A ( r ) = 0 is automatically satisfied everywhere ( ∀ r ) If ∇i B ( r ) ≡ 0 for each/every point, r in a volume v bounded by its surface S ˆ (by the Divergence Theorem) ∫ ∇i B ( r ) dτ = ∫ B ( r )indA ˆ ?? ∫ B ( r )indA ˆ ≡0 What is ∫ B ( r )indA Then: v S S S Recall Gauss’ Law for E (and/or D ) were: ˆ = Q enclosed Φ D ≡ ∫ DindA = net electric displacement flux through closed surface S . free S ΦE ≡ ˆ =Q ∫ E ( r )indA Thus: Φ m ≡ ˆ =0 ∫ B ( r )indA S S enclosed Tot ε o = net electric flux through closed surface S . = net magnetic flux through closed surface S ≡ 0 !!! ⇒ Magnetic flux is conserved → magnetic field lines have no beginning / no end points (because ∃ no magnetic charge(s)!) The SI units of magnetic flux Φ m are Tesla-m2 = Webers 16 © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Again: Do not confuse magnetic flux, Φ m with the magnetic scalar potential Φ m ( r ) (they even have the same units!!! Webers / Tesla-m2) WAA-HEE !!! They are not the same thing!!! A′ ( r ) = A ( r ) + ∇Φ m ( r ) B ( r ) = ∇ × A′ ( r ) = ∇ × A ( r ) + ∇ × ∇Φ m ( r ) ˆ Φ m = ∫ B ( r )indA S Magnetic Flux Magnetic Vector Potential Area element Magnetic Scalar Potential Don’t confuse these either!!! Φm ( r ) = ∇ × A ( r ) = ∇ 2Φ m ( r ) = − ρm ( r ) 1 4π ∫ ∇i A ( r ) r v dτ gives ∇i A′ ( r ) = 0 The magnetic flux through a surface S (not necessarily closed!!!): ˆ = ∫ ∇ × A ( r ) indS ˆ = ∫ A ( r )id by Stoke’s Theorem Φ m = ∫ B ( r )indS S S ( ) C n.b. not a closed surface! Φm = Magnetic flux enclosed by contour C : n.b. This d S ( ) C is NOT a line segment associated with a line current I !!! ˆ = ∫ ∇ × A ( r ) indS ˆ = Φ m = ∫ B ( r )indS S ∫ A ( r )i d ∫ A ( r )i d C = ∫ C′ ∇Φ m ( r )id = Φ m © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 17 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Griffiths Example 5.11: A spherical shell of radius R carries a uniform surface charge density σ and rotates with constant angular velocity ω , Determine the magnetic vector potential it produces at point r . A rotating surface charge density σ produces a surface/sheet current density K ( r ′ ) = σ v ( r ′ ) The magnetic vector potential is thus: A ( r ) = μo 4π ∫ S′ K ( r′) r da′ For ease of integration, choose the observation/field point P ( r ) = P ( zzˆ ) (i.e. r = zzˆ ) along the + ẑ -axis and ω to lie in the x-z plane. Choose the origin ϑ to be at the center of the sphere, as shown in the figure below: r = r − r ′ = R 2 + r 2 − 2 Rr cos θ ′ from the law of cosines K ( r ′) = σ v ( r′) v ( r ′) = ω × r ′ ω = ω sinψ xˆ + ω cosψ zˆ (here) r ′ = R sin θ ′ cos ϕ ′xˆ + R sin θ ′ sin ϕ ′ yˆ + R cos θ ′zˆ and: v ( r ′) = ω × r ′ xˆ v ( r ′) = yˆ zˆ ω sinψ ω cosψ 0 R sin θ ′ cos ϕ ′ R sin θ ′ sin ϕ ′ R cos θ ′ v ( r ′ ) = Rω ⎡⎣ − ( cosψ sin θ ′ sin ϕ ′ ) xˆ + ( cosψ sin θ ′ cos ϕ ′ − sinψ cos θ ′ ) yˆ + ( sinψ sin θ ′ sin ϕ ′ ) zˆ ⎤⎦ Now since ∫ 2π 0 2π sin ϕ ′dϕ ′ = ∫ cos ϕ ′dϕ ′ = 0 0 Then terms involving only sin ϕ ′ or cos ϕ ′ in the integral for A ( r ) contribute nothing. 18 © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Lecture Notes 16 Prof. Steven Errede σ (ω × r ′ ) da′ r r with K ( r ′ ) = σ (ω × r ′ ) and da′ = R 2 sin θ ′dθ ′dϕ ′ and r = R 2 + r 2 − 2 Rr cos ϕ ′ A(r ) = μo 4π ∫ S′ Then: A ( r ) = Let: K ( r ′) Fall Semester, 2007 da′ = μo 4π μo R 3σω sinψ ⎛ u ≡ cos θ ′ du = sin θ ′dθ ′ π ⎞ dθ ′ ⎟ yˆ R + r − 2 Rr cos θ ′ ⎠ cos θ ′ sin θ ′ 2 cos θ ′ sin θ ′ R 2 + r 2 − 2 Rr cos θ ′ (R =− 2 θ ′ = 0 ⇒ u = +1 θ ′ = π ⇒ u = −1 π 0 S′ ⎜ ∫0 ⎝ 2 ∫ ∫ 2 + r 2 + Rru ) 3R 2 r 2 dθ ′ = ∫ +1 −1 udu R 2 + r 2 − 2 Rru u =+1 R 2 + r 2 − 2 Rru u =−1 1 ⎡ R 2 + r 2 + Rr ) R − r − ( R 2 + r 2 − Rr ) ( R + r ) ⎤ 2 2 ⎣( ⎦ 3R r ⎛ 2r ⎞ If r < R (i.e. inside sphere) then this integral = ⎜ 2 ⎟ ⎝ 3R ⎠ ⎛ 2R ⎞ If r > R (i.e. outside sphere) then this integral = ⎜ 2 ⎟ ⎝ 3r ⎠ =− Now: ω × r = −ω r sinψ yˆ Then: A ( r ) = μo Rσ (ω × r ) for r < R (inside sphere) 3 μo R 4σ A(r ) = (ω × r ) for r > R (outside sphere) 3r 3 If we now rotate the problem so that ω = ω ẑ and r = ( r , ϑ , ϕ ) then ω × r = −ω r sinψ yˆ ⇒ ω r sin θϕˆ , thus with ω rotated to ω = ω ẑ and field point now located at r = ( r , ϑ , ϕ ) , the magnetic vector potential A ( r ) inside/outside the rotating sphere becomes: A ( r ,ϑ , ϕ ) = μo Rωσ r sin θϕˆ ( r < R , inside sphere) 3 μo R 4ωσ sin θ ϕˆ ( r > R , outside sphere) A ( r ,ϑ , ϕ ) = 3 r2 A(r ) Amax ( r = R ) ~r = μo R 2ωσ sin θ 1 3 ~ 1 r2 r=R Then: B ( r ) = ∇ × A ( r ) = = zˆ r 2μo Rωσ 2 2 cos θ rˆ − sin θθ = μo Rω zˆ = μo Rω !!! ( ω = ω ẑ ) 3 3 3 ( ) © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 19 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Griffiths Example 5.12: Determine the magnetic vector potential A ( r ) of an infinitely long solenoid with n turns / unit length, radius R and steady current I ⇒ n.b. The current extends to infinity, so we cannot use A ( r ) = μo 4π ∫ C′ But we do know that: ( ) I r d ′ because it diverges! Magnetic flux Φ m = ∫ A ( r )id = ∫ ∇ × A ( r ) ida = ∫ Bida C S′ S′ Flux-enclosing loop / contour Φ m = ∫ B ( r )ida S But we know from Ampere’s Circuital Law that: Binside ( r ≤ R ) = μo nIzˆ = uniform & constant = ( μo nI ) ∗ (π R 2 ) = μo nI π R 2 ∴ Φ inside m = ∫ A ( r = R )i d But: Φ inside m C where d = Rdϕϕˆ = A ( r = R ) 2π R ∴ Φ inside m Now Asolenoid must be parallel to I = Iϕˆ for the “ideal solenoid” (i.e. no pitch angle) ⇒ A ( r ) = A ( r ) ϕˆ Then: A ( r = R ) = Φ inside μ nI π R 2 1 m ϕˆ = μo nIRϕˆ = o 2 2π R 2π R If r > R, then more generally, we have: For r < R, then: Aoutside ( r > R ) = Ainside ⎛ R2 ⎞ 1 μo nI ⎜ ⎟ ϕˆ 2 ⎝ r ⎠ 1 ( r < R ) = μo nIrϕˆ 2 A(r ) Amax ( r = R ) = 12 μo nIR ~r ~1 r r=R Note that: A ( r ) = Aϕ ( r ) ϕˆ (only) for the infinitely long ideal solenoid. 20 © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. r UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Does B ( r ) = ∇ × A ( r ) ? B (r ) = ∇ × A(r ) = 1 ∂ ( rAϕ ( r ) ) zˆ in cylindrical coordinates r ∂r =0 2 1 1 1 ∂ ⎛1⎞ ∂ ⎛ R ⎞ Boutside ( r > R ) = μo nI ⎜ ⎟ ⎜ r R 2 ) zˆ ≡ 0 ( ⎟ zˆ = μo nI 2 r ∂r 2 ⎝ r ⎠ ∂r ⎝ r ⎠ 1 1 ⎛1⎞ ∂ ⎛1⎞ Binside ( r < R ) = μo nI ⎜ ⎟ ( r 2 ) zˆ = μo nI ⎜ ⎟ ( 2r ) zˆ = μo nIzˆ 2 2 ⎝ r ⎠ ∂r ⎝r⎠ A ( r ) = Aϕ ( r ) ϕˆ Does ∇i A ( r ) = 0 ?? (Coulomb Gauge) In Cylindrical Coordinates: 1 ∂Aϕ ( r ) ∇i A ( r ) = = 0 because A ( r ) has NO explicit ϕ -dependence! r ∂ϕ Aoutside ( r > R ) = Ainside ( r < R ) = ⎛ R2 ⎞ 1 μo nI ⎜ ⎟ ϕˆ 2 ⎝ r ⎠ 1 μo nIrϕˆ 2 Magnetostatic Boundary Conditions In the case of electrostatics, we learned (via use of Gauss’ Law enclosed ˆ = QTot Φ E = ∫ E ( r )inda ε o ) that the normal component of E ( r ) suffers a discontinuity S whenever there is a surface charge density (free or bound) present on a surface / interface: E2⊥ above − E1⊥ below = (Ε above 2 ) σ Tot (σ free + σ bound ) ∂V2above = = εo εo ∂n − surface ∂V1below ∂n surface = Ε1above ( Ε is continuous across interface) n.b. ⊥ = perpendicular component relative to surface, = parallel component relative to surface: © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 21 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Consider a thin conducting sheet of material carrying a surface current density of K = − Kxˆ = K ( − xˆ ) Amperes/meter Now imagine that this current sheet K = − Kxˆ = K ( − xˆ ) is “placed” in an external magnetic field, e.g. created / emanating from some other current-carrying circuit below this current sheet. Call this external magnetic field that is below the original current sheet B1below . ext What we discover is that the magnetic field above the current sheet B2above is not parallel to B1below ext ext - it has been refracted by the current sheet (in the tangential direction - with respect to the surface)! The physical origin for this is simple to understand. Below the current sheet, the current sheet 1 below below itself adds to the tangential component of Bext a component Bsheet = − μo Kyˆ (for K = − Kxˆ ), 2 above however, above the current sheet, the current sheet adds to the tangential component of Bext a 1 above component Bsheet = + μo Kyˆ (for K = − Kxˆ ). 2 ⊥ zˆ So if: Bext = Bextx xˆ + Bext y yˆ + Bext z below below below ⊥below = Bextx xˆ + Bext y yˆ + Bext zˆ Then: Bext z And: 22 above above ⊥ above above Bext = Bextx xˆ + Bext y yˆ + Bext zˆ z ║ = parallel to surface ⊥ = perpendicular to surface © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Then by the principle of linear superposition, BTot = Bext + Bsheet . Hence, below the current sheet ( K = − Kxˆ ): = B sheet ⎞ ⎛ ⎜ ⎟ 1 below ⊥below below below below ˆ ⎜ ⎟ yˆ + Bext = Bextbelow + − x B μ K zˆ = BTOT xˆ + BTOT yˆ + BTOT zˆ ext y o x z x y z 2 ⎜ ⎟ ⎜ ⎟ ⎝ ⎠ below below BTOT And above the current sheet ( K = − Kxˆ ): = B sheet ⎞ ⎛ ⎜ above 1 ⎟ ⊥ above above above above ˆ ⎜ ⎟ yˆ + Bext = Bextabove + + x B μ K zˆ = BTOT xˆ + BTOT yˆ + BTOT zˆ ext o x y z x y z 2 ⎜ ⎟ ⎜ ⎟ ⎝ ⎠ above above BTOT above below Thus, (comparing BTOT vs. BTOT component-by-component), we see that: 1) 2) below above BTOT = BTOT x x Tangential (to sheet / surface) component of BTOT parallel Bextbelow = Bextabove x x to sheet current K = − Kxˆ is continuous. BTOTy ≠ BTOTy Tangential (to sheet/surface) component of BTOT perpendicular below above above below BTOTy − BTOTy = μo K to sheet current K = − Kxˆ is discontinuous by an amount μo K across sheet / surface. 3) ⊥below ⊥ above BTOT = BTOT z z Normal (to sheet/surface) component of BTOT is continuous ⊥below ⊥ above Bext = Bext z z across sheet / surface. Mathematically, these 3 statements can be compactly combined into a single expression: above below BTOT − BTOT = μo K × nˆ where the unit normal to the surface, nˆ = zˆ (here, as drawn above). © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 23 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede As we found in electrostatics, that the scalar electric potential V ( r ) was continuous across any boundary Vabove ( r ) = Vbelow ( r ) , likewise, the magnetic vector potential A ( r ) is also continuous across any boundary, i.e. Aabove ( r ) = Abelow ( r ) provided that: ∇i A ( r ) = 0 , which ( ) guarantees that A⊥above ( r ) = A⊥below ( r ) and also provided that: ∇ × A ( r ) = B ( r ) , which, in integral form, i.e. ∫ A ( r )i d C = ∫ B ( r )ida = Φ S guarantees that Aabove ( r ) = Abelow ( r ) . m However, note that the normal derivative of A ( r ) , since A ( r ) K ( r ) then A ( r ) also “inherits” the discontinuity associated with B ( r ) : BTOTy − BTOTy = μo K (see #2 on previous above below page), and since B ( r ) = ∇ × A ( r ) , thus we have a discontinuity in the (normal) slope(s) of A ( r ) on either side of the boundary/current sheet. We can understand the origin of this condition on the normal derivative(s) of A ( r ) taken just above/below an “interface” e.g. for the specific case of the current sheet K = − Kxˆ . From above below BTOTy − BTOTy = μo K we know that the discontinuity in the B -field is in the ŷ -direction, whereas since the magnetic vector potential associated with the current sheet A ( r ) is always parallel to the current, and since K = − Kxˆ we know that the component of ATOT ( r ) that we are ( concerned with here is in the ŷ -direction. But from: BTOT = ∇ × ATOT , then: BTOTy = ∇ × ATOT ) y thus we need to worry only about the ŷ -component of the curl of ATOT ( r ) , which is: ( BTOTy = ∇ × ATOT ) y ⎛ ∂ATOTx ∂ATOTz =⎜ − ⎜ ∂z ∂x ⎝ ⎞ ⎟⎟ ⎠ Then, noting that the ẑ -direction is perpendicular (i.e. normal) to the plane of the current sheet: above below ⊥ above ⊥below ⎛ ∂ATOT ⎞ ⎛ ∂ATOT ⎞ ∂ATOT ∂ATOT above below x x z z ⎟ ⎟ − −⎜ − BTOTy − BTOTy = μo K = ⎜ ⎜ ∂z ⎟ ⎜ ∂z ⎟ ∂ ∂ x x ⎝ ⎠ surface ⎝ ⎠ surface above below ⊥ above ⊥below ⎛ ∂ATOT ⎛ ∂ATOT ⎞ ∂ATOTx ⎞ ∂ATOT x z z ⎟ =⎜ − −⎜ − ⎟ ⎜ ∂x ⎜ ∂z ∂z ⎟⎠ ∂x ⎟⎠ ⎝ ⎝ surface above below ⎛ ∂ATOT ∂ATOTx ⎞ x ⎟ =⎜ − ⎜ ∂n ⎟ ∂ n ⎝ ⎠ surface surface =0 ⊥ ATOT suffers no discontinuity z ⊥ nor ATOTy suffer discontinuities in their slopes at the current sheet – only Neither ATOT z above ATOT does - in the normal (i.e. ẑ ) direction. Therefore, we can most generally write this x condition on the discontinuity in the normal derivative on A ( r ) as: ∂Aabove ( r ) ∂Abelow ( r ) − = − μo K ∂n ∂ n surface surface 24 © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede The Magnetic Vector Potential A ( r ) Associated with a Finite Circular Disk Sheet Current We wish to delve a bit deeper into the nature of the magnetic vector potential, A ( r ) and also B ( r ) = ∇ × A ( r ) associated with current sheets. Consider a sheet current K = K o xˆ flowing on the surface of a finite circular disk of radius R, lying in the x-y plane as shown in the figure below: ẑ +z r ρ xˆ , K = K o xˆ ŷ −z To keep it simple, we’ll just calculate A ( r ) at an arbitrary point along the ẑ -axis above and below the x-y plane. The magnetic vector potential A ( r ) associated with a sheet current is: A(r ) = μo 4π K ( r ′) μ K xˆ da′ da′ = o o ∫ S′ 4π S ′ r r ∫ We deliberately chose a sheet current flowing on a finite circular disk of radius R so that we could easily carry out the integration. The area element da′ on the circular disk (in cylindrical coordinates) is da′ = d ρ ( ρ dϕ ) = ρ d ρ dϕ , and from the figure above, we see that: r = ρ 2 + z 2 . A( z ) = Thus: = ρ =R μo K o xˆ ρ = R ϕ = 2π ρ d ρ dϕ 2π μo K o xˆ ρ = R ρ d ρ 1 = μo K o xˆ ⎡ ρ 2 + z 2 ⎤ = ∫ ∫ ∫ 2 2 2 2 = 0 = 0 = 0 ρ ρ ϕ ⎣ ⎦ ρ =0 2 4π 4π ρ +z ρ +z 1 1 μo K o xˆ ⎡ R 2 + z 2 − z 2 ⎤ = μo K o ⎡ R 2 + z 2 − z 2 ⎤ xˆ ⎣ ⎦ 2 ⎣ ⎦ 2 Now there is a subtlety here that we need to notice before proceeding further – since we are interested in knowing A ( z ) at an arbitrary point along the ẑ -axis - above and/or below the x-y plane, thus z can be either positive or negative. Note that both the R 2 + z 2 and z 2 terms are always ≥ 0 for both positive and/or negative z (in particular: z 2 = z ≠ z !). Thus, in order to preserve this fact, we explicitly keep expression for the magnetic vector potential A ( z ) as: A( z ) = 1 μo K o 2 ( ) R 2 + z 2 − z 2 xˆ © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 25 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede A plot of the magnetic vector potential A ( z ) vs. z is shown in the figure below for a circular disk of radius R = 10 m and sheet current K = K o xˆ = 1.0 xˆ Amperes/meter. Note that A ( z ) is a maximum when z = 0, right on the sheet current. Note also the discontinuity in the slope(s) of A ( z ) on either side of z = 0, which arises due to the presence of the sheet current in the x-y plane, since: ∂Aabove ( r ) − ∂n surface ∂Abelow ( r ) = − μo K ∂n surface ∂Aabove ( z ≥ 0 ) ∂Abelow ( z ≤ 0 ) − = − μo K ∂z ∂z z =0 z =0 or: Care/thought must also be taken when carrying out the normal derivatives (slopes) above and below the x-y plane – look carefully at the slopes for z > 0 and z < 0 in the above figure, and compare this information to what we calculate: ∂A above ( z ≥ 0) = 1 μ o Ko ∂ ∂z ( ⎛ z z R 2 + z 2 − z 2 xˆ = 12 μo K o ⎜ − 2 2 z2 ⎝ R +z ( z ≤ 0) = 1 μ o Ko ∂ ∂z ( ⎛ ⎛ ⎞ z z ⎞ z 1 R 2 + z 2 − z 2 xˆ = 12 μo K o ⎜ − + 1⎟ xˆ ⎟ xˆ = 2 μo K o ⎜ 2 2 2 2 2 z ⎠ ⎝ R +z ⎝ R +z ⎠ ∂z ∂A below ∂z 26 2 2 ) ⎞ ⎛ ⎞ z 1 − 1⎟ xˆ ⎟ xˆ = 2 μo K o ⎜ 2 2 ⎠ ⎝ R +z ⎠ ) © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede Thus we see that indeed: ∂Aabove ( z ≥ 0 ) ∂Abelow ( z ≤ 0 ) − = − 12 μo K o xˆ − 12 μo K o xˆ = − μo K o xˆ = − μo K ∂z ∂ z z =0 z =0 The magnetic field B ( z ) at an arbitrary point along the along the ẑ -axis – either above and/or below the x-y plane is calculated using B ( z ) = ∇ × A ( z ) in Cartesian coordinates. Since A ( z ) = Ax ( z ) xˆ (only), then: B ( z ) = ∇ × A ( z ) = ∇ × Ax ( z ) xˆ = ∂Ax ( z ) yˆ ∂z Thus: ∂Axabove ( z ≥ 0 ) ⎛ ⎞ z − 1⎟ yˆ B yˆ = 12 μo K o ⎜ ( z ≥ 0 ) = ∇ × A ( z ≥ 0 ) = ∇ × A ( z ≥ 0 ) xˆ = 2 2 ∂z ⎝ R +z ⎠ below ∂A ( z ≤ 0 ) yˆ = 1 μ K ⎛ z + 1⎞ yˆ B below ( z ≤ 0 ) = ∇ × Abelow ( z ≤ 0 ) = ∇ × Axbelow ( z ≤ 0 ) xˆ = x ⎟ 2 o o⎜ 2 2 ∂z ⎝ R +z ⎠ above above above x The figure below shows the magnetic field B ( z ) vs. z along the ẑ -axis with a sheet current K = K o xˆ flowing on the surface of the finite disk of radius R, lying in the x-y plane: © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved. 27 UIUC Physics 435 EM Fields & Sources I Fall Semester, 2007 Lecture Notes 16 Prof. Steven Errede We now investigate what happens in the limit that the radius of the sheet current-carrying circular disc, R → ∞ , i.e. it becomes an infinite planar sheet current. We discover that the magnetic vector potential A ( r ) associated with the sheet current K = K o xˆ becomes infinite (i.e. A ( r ) diverges): ( ) lim A ( z ) = R→∞ 1 μo K o 2 ( ) R 2 + z 2 − z 2 xˆ → 1 μo K o 2 ( ) ∞ 2 + z 2 − z 2 xˆ whereas the boundary condition on the discontinuity in the normal derivative of A ( r ) across the sheet current lying in the x-y plane at z = 0 still exists, and is well-behaved (i.e. finite): ∂Aabove ( z ≥ 0 ) ∂Abelow ( z ≤ 0 ) − = − 12 μo K o xˆ − 12 μo K o xˆ = − μo K o xˆ = − μo K ∂z ∂ z z =0 z =0 We also discover that the magnetic field B ( r ) is also well-behaved (i.e. finite) – and constant – independent of the height/depth z above/below the x-y plane (!!): ⎛ ⎞ z − 1⎟ yˆ = − 12 μo K o yˆ lim B above ( z ≥ 0 ) = 12 μo K o ⎜ 2 2 R→∞ ⎝ ∞ +z ⎠ ⎛ ⎞ z + 1⎟ yˆ = + 12 μo K o yˆ lim B below ( z ≤ 0 ) = 12 μo K o ⎜ 2 2 R→∞ ⎝ ∞ +z ⎠ 28 ( ) ( ) © Professor Steven Errede, Department of Physics, University of Illinois at Urbana-Champaign, Illinois 2005-2008. All Rights Reserved.