27-1 (SJP, Phys 1120)

Electric flux, and Gauss' law

Finding the Electric field due to a bunch of charges is KEY! Once you know
E, you know the force on any charge you put down - you can predict (or
control) motion of electric charges! We're talking manipulation of anything
from DNA to electrons in circuits... But as you've seen, it's a pain to start from
Coulomb's law and add all those darn vectors.
Fortunately, there is a remarkable law, called Gauss' law, which is a universal
law of nature that describes electricity. It is more general than Coulomb's law,
but includes Coulomb's law as a special case. It is always true... and
sometimes VERY useful to figure out E fields! But to make sense of it, we
really need a new concept, Electric Flux (Called ). So first a "flux interlude":

Imagine an E field whose field lines "cut through" or "pierce" a loop.

Define as the angle between E and the "normal" E
or "perpendicular" direction to the loop.
We will now define a new quantity, the electric
!
flux through the loop, as
A, the
Flux, or = E A = E A cos "normal"
to the loop
E is the component of E perpendicular to the loop:
E = E cos.
For convenience, people will often characterize the area of a small patch (like
the loop above) as a vector instead of just a number.
The magnitude of the area vector is just... the area! (What else?)
But the direction of the area rvector
r is the normal to the loop.
That way, we can write " = E # A .
(Can you see that this just gives the formula we had above?)

!
Flux is a useful concept, used for other quantities besides E, too. E.g. if you have solar
panels, you want the flux of sunlight through the panel to be large. House #2 has poorly
designed panels. Although the AREA of the
panels is the exact same, and the sunshine
(Less flux) brightness is the exact same, panel 2 is less
useful: fewer light rays "pierce" the panel, there
(Lots of flux) is less FLUX through that panel.
1 2 Solar panel 2,
same area,
Solar panel diffferent tilt.
House
27-2 (SJP, Phys 1120)

Examples of calculating flux:

E Here (picture to the left) = E A,
A
because E is parallel to the area vector. (=0)
(If E pointed the opposite direction, you'd have a "negative" flux!)
Area,
Note: I've drawn an area "A" vector for you: remember, that's defined as
A
perpendicular to the surface in question.
Here (picture to the right), =0, because E is perpendicular to the A
area. (=90. ) No flux: the E field lines don't "pierce" this loop, they Area,
A
"skim" past it... E

E E" Here, (picture to the left), = E A cos. The flux is

! reduced a bit because it's not perfectly perpendicular.
Again, the A vector is (by definition) perpendicular to our
A
surface.
r r
(So is also the angle between E and A, and
" = E # A =E A cos)
_______________________________________________________________
What if E is not constant? Or, if the area curves? You can still find the flux,
you just break
! your surface up into little "patches", and add up (integrate!) all
the small fluxes through each little patch.
For Gauss' law (that's what we're after, remember!) we're going to have to
evaluate the electric flux through a CLOSED SURFACE: like the surface of a
balloon, or a soda can, or a cube... or just some imaginary surface in space.
(Closed means there's an inside and an outside. So the patches shown above
are not closed, but a cube is, or the funny ball shown below...)
Of course, any surface can be thought of as the sum of a
bunch of little patches! Just imagine taking any surface E E
dA
and drawing a little "grid" on it, completely covering it. dA
Each tiny patch will have tiny area (dA).
So, the flux through a closed surface is the SUM (really, E
INTEGRAL) of the flux through all the patches dA E
covering the surface. (The convention is always that dA
points everywhere out of a closed surface)
r r
We write this flux formally as " = $ E # dA
(The weird circle around the integral is to remind you that we're talking about
an integral, or sum, all the way around some closed surface!)
!
In a crude sense, flux counts how many "field lines" pierce out a surface!
27-3 (SJP, Phys 1120)

A3
Easy example #1: E is uniform (constant) from left to
3
right, and our surface is a cube with face area A. E
A1
What's the total flux through the (closed) cube shown? 1 2 A2

4
E

We need to add up the flux through all 6 faces. Now, the top and bottom
faces, (3 and 4), have an area vector which is perpendicular to E. So the flux
through them is zero. (E "skims" them, doesn't "poke through" them).
Convince yourself! The exact same argument is true for the front and back
faces (not numbered in the picture).rAgain,
r each has zero flux.
r r
What about face 2? Here, flux is # E " dA (through wall 2) = E " A = EA
What about face 1? This time, the flux is -EA.
The minus sign comes because area vectors always point OUT. That means A(face 1)
points left, but E points right, and the dot product is thus EA!cos(180) = -EA.
!
The total flux through entire surface is thus
Flux(1)+Flux(2)+ ... + Flux(6) = -EA + EA + 0 + 0 + 0 + 0 = 0.
There is NO net flux through this surface!
That's correct: the number of lines "poking in" on the left is equal to the
number "poking out" on the right, the total is thus zero.
Notice: If the E field and/or the surface are simple enough, you don't really
have to DO any integral - just think of the SUM of fluxes through all parts of
the surface, and add 'em up.
 27-4 (SJP, Phys 1120)

Easy example #2: You have a point charge q at the center E E

of a perfect sphere of radius R. What is the flux through the

q
sphere? dA

Here's the key thing you have to realize. Since E is E E

everywhere "radially outward", and the little area vectors

dA at any patch on a sphere are also "radially outward", then on ANY little
patch, E and dA are parallel! That means E.dA = E dA (the cosine in the dot
product ALWAYS gives one, at every point everywhere around the sphere!)
From Coulomb's law, we know that at the surface of this sphere, E = kQ/R^2
(It's the same in magnitude everywhere around the sphere!)
r r kq
So we want to compute " = $ E # dA , which is thus just "R 2
dA .
But now notice that those are all constants, and constant come out of integrals.
kq
So " = # dA. !
R2 !
Now think what that funny integral means. It means "sum up dA for all
patches everywhere around the sphere".
! But that's just the total area of a sphere, 4 R^2! So the R^2 cancels, and
Flux = 4 k q. It's a constant (independent of radius. )
It's positive (if q is +): there is a net flux pointing OUT of the sphere!
Remember, flux is basically "counting flux lines that poke out".
If we moved q a little to the side, the flux would be bigger on that side, but
smaller on the other. But still the total number of flux lines poking out would
be the same! The total flux would be the same no matter where q was, as long
as it is somewhere inside the sphere.
If you added a second charge, q2, it would add more flux = 4 k q2.
So the total flux would just be 4 k (q + q2). (No matter where precisely
those charges were inside the sphere.) This leads us to the big idea of this
chapter. If you have a surface with charges inside it, the total flux through that
surface will just be 4 k * (q total, inside)
1
Since we defined k = back in Chapter 22, we can write this as
4 "#0
r r qenclosed
"= % E # dA = $0
This equation!is called "Gauss' law". We just showed it for spherical surfaces.

!
 27-5 (SJP, Phys 1120)

Gauss' law his actually comes from experiment, and is true no matter WHAT
the shape of the closed surface is! Ut involves measurable quantities, and as
far as we know, there are NO EXCEPTIONS to it.
Again: if you have a bunch of charges enclosed by any surface (a closed
surface, which has an inside an an outside) then:
r r qenclosed
$ E " dA = Gauss' law, simple but cryptic! Let's review what it means:
#0
The left side is the electric flux through our surface.
q(enclosed) = the total amount of charge inside.
! "0 =8.85E-12 C^2/(N m^2) is a constant of nature (defined in Ch. 22).

You can pick a surface, any surface you want. It can be a real physical surface
(like a balloon) or an IMAGINARY surface in space. Doesn't matter.
!
Inside that surface will be some charges. They can be distributed arbitrarily.
(Could be a mix of + and -) Together, they add up to a total charge = qenclosed .
Those charges will produce an electric field (all throughout the universe...!)
and those E field lines will "pierce" your surface... We've just seen that you
can always (in principle) compute the FLUX of E field through ! any surface...:
that's the LEFT SIDE of Gauss' law: total electric flux through our surface.
It tells you "how many flux lines poke OUT the surface" (really, # out - # in)
The right side is simple as can be: proportional to total charge.
Gauss' law is amazing! It's going to take a while to sink in what it really says.
We'll start by looking at our simple examples (where we find both left and
right sides). We'll see how Gauss' law gives us Coulomb's law. We'll see some
useful consequences when we consider metals in the real world. We'll look at
some situations where Gauss' law can be used to figure out E fields from
useful (but rather complicated) charge distributions where "integrating"
Coulomb's law to find E, like in the last chapter, would be truly nightmarish.
Gauss' law is one of four "master equations" we'll learn this semester. These
equations are together called "Maxwell's Equations" (Maxwell tied them
together). With them, all classical electromagnetic phenomena can be
described, predicted, and explained!
 27-6 (SJP, Phys 1120)

Remember the cube example? The total flux was zero. (Number of lines
poking out = number of lines poking in). Is that consistent with Gauss' law?
Yes it is - the cube was empty! E field lines came in and went back out. (If
there were any charges in there, E field lines would start or end on the
charges: if they pass on through, there's no charge in there!)
Suppose we only knew Gauss's law. I claim we can DEDUCE Coulomb's law
from it! Here's how: Place a charge q at the origin. I'd like to figure out the E
field at some arbitrary point, a distance R away. (Coulomb tells me the
answer, but let's work it out from Gauss' law).
I do need to make an additional argument, based on symmetry. A point
charge is totally symmetric: no DIRECTION in space can be any different
than any other. North, South, East, West... the E field must be the same in
different directions. So by SYMMETRY, I can argue that whatever E is, it
must i) point radially away from the charge, and ii) must depend ONLY on
the distance away from the charge. We call this a
spherically symmetric field. I could say E = E(r) r E E

(That means the the E field might depend on radius, but at a q dA

given radius, it has a definite value and points outwards)
!
So let's draw an IMAGINARY sphere, centered on the E E

charge, with radius R. Let's see what Gauss' law says.

r r qenclosed
$ E " dA = . The right side is obviously just q/ "0 . What about the left?
#0
If we don't KNOW E yet, how can we do the integral? Well, the point is, I
don't know what E is, but I DID argue that E = E(r) r , i.e. it's the SAME at
!
! all points on the sphere, and always points out (i.e. parallel to dA). That means
everywhere on our surface, E.dA = E(R) dA. Since E(R) is just a constant,
pull it out of the integral: E(R) " dA = q /#0 . !Once again, that funny integral is
just the area of the sphere, 4 R^2, which means E(R) 4 " R 2 = q /#0 , or,
q
solving, E = . That's Coulomb's law!
4 "#0 R!2
!
Let me summarize: I could "do" the integral in Gauss' formula even without
knowing
! what exactly E(R) was, by cleverly CHOOSING a nice Gaussian
surface and invoking symmetry. And then I came back with Coulomb's law!
So Coulomb's law can be thought of as a consequence of Gauss' law, which is
the deeper and more fundamental of the two...
27-7 (SJP, Phys 1120)

Let's try another example. Suppose you have a spherical ball of charge (NOT
a point). What's the E field outside this ball? Coulomb's law doesn't tell us the
answer (it's only valid for point charges). But Gauss' law does: follow
EXACTLY the same steps as the previous example, and conclude that the E
field is the same as if all the charge was located at a point at its center!
What if we had a spherical shell of charge Q (not a sphere: a thin shell, like
the skin of a balloon). What's E outside that? The exact same argument says
it's the same as if all the charge was located at a point at the center of the shell.
Let's do another one: suppose I have this spherical shell of
E?
charge Q, spread evenly.
R
What's E inside the shell, a distance R from the origin?
Since the charge is spherically symmetric, my old
symmetry argument still holds: E can only depend on R
(not on angle), and it must point radially.
(Might be radially out, or in, that I can't tell, but it can't point at some funny
other angle: that would be non-symmetric.)

It's the same game as before!

Draw an imaginary spherical surface with radius R. Note: R

that's a new surface, it's NOT the same as the shell that has
the charge on it. It's inside that one!
r r qenclosed
Now apply Gauss' law: $ E " dA = .
#0

We're going to be integrating over our new imaginary sphere!

We've seen what to do with the left side: E.dA = E(R) dA, the E(R) comes out
of the integral, !
and we're left with E(R) 4 " R 2 = qenclosed /#0
But what is q(enclosed)? Think about it : the charge Q of the shell is outside
my imaginary surface. q(enclosed) is zero, there's no charge INSIDE the
dashed line. That means E(R)! = 0.
There is NO E field ANYWHERE inside that shell. Not even right up close to
all the charges just inside the shell!
This has useful physical consequences: we'll come back to it later. (It's why
you're safe in your car if lightning hits it and dumps charge all around you. If
you're INSIDE, E=0!) But we need to come back and see how this is still true
even with a "non-spherical" car, and "non-uniform" charge distribution!
27-8 (SJP, Phys 1120)

What if we lose symmetry? E.g, suppose you have a dipole near the origin.
What's E a distance R away from the origin?
E?
r r qenclosed
It is still true that $ E " dA = =0, but now we can't R
#0
make the symmetry argument, and I no longer can
+q -q
figure out E in any easy way!
I certainly!can NOT conclude that E=0, just because the
integral is zero. There's no reason to assume E is the
same everyone on the surface I drew, so if E is not a constant, you can't pull it
out of the integral.
We're stuck - I can't simplify the integral. I know that the total flux through
that sphere is zero - the #of lines poking out = # poking in. But that does NOT
tell us that those #'s are each separately zero!
We'd have to go back to more direct (and painful) ways: summing up dE from
each charge.
Bottom line: Gauss' law is great, especially when there's symmetry. But, it
doesn't ALWAYS tell you what E is at any given point in space. (So it's
always true, but not always helpful for problem solving! )
 27-9 (SJP, Phys 1120)

Another example: Suppose you have a large (effectively +

infinite) flat sheet of uniformly smeared out charge, with +
+ E?
uniform charge density = charge/area. +
What is the E field a distance L away from the sheet? + L
+
^z
We have symmetry again, but it's not spherical. ^
y
What I'll argue is that if the sheet is infinite, then E cannot ^
depend on z or y. (It might depend on x, we'll have to see..) x
Furthermore, E can't point "up" or "down" because of symmetry.
In fact, the only way it could point is away from the sheet (in the x direction).

Symmetry arguments are subtle, and maybe unfamiliar. But they're a really
important idea in lots of applications. Think about the argument I just made.
Ask yourself why E can't point any way except "directly away"!
So we've argued that E = E(x) i . It's time to apply Gauss' law.
But remember, you first have to pick a surface over which you will integrate.
What surface should we pick? We want on that takes advantage of the
symmetry. Picking a big! sphere is a bad idea, because this problem doesn't
have spherical symmetry.
(The E field would not point parallel to dA on a sphere, and wouldn't
necessarily be the same everywhere. )
Here we need a little creativity!
I'm going to propose the infamous "beer can"
surface, also known as a "Gaussian pillbox".
It's an imaginary cylinder, endcap area A,
A
centered around the charged sheet:
Now consider Gauss' law applied to that pillbox.
r r q L L
$ E " dA = enclosed
#0
That integral is a surface integral: we need to figure
out E.dA all around the pillbox.
!
I propose thinking of the pillbox as having 3 distinct parts: the left endcap, the
right endcap, and the rest of the cylinder. To find the total flux, we'll add the
flux through all three of those parts of the surface.
 27-10 (SJP, Phys 1120)

Consider first the "cylinder part". It has little dA vectors spiking dA

E
out, all of which are perpendicular to the x direction! (Do you
see that?) So if E points purely in the x direction, it's
"skimming" the can, not poking it.
The flux through that entire cylindrical surface is zero!
What about the right end cap? There, E points right, and so does dA.
E is totally uniform over that whole end, so integrating E.dA just E
gives you E A, where A is the area of the endcap, and E=E(L) is the dA
value of the E field out there a distance L away from the sheet.
What about the left end cap? There E points left, and so does dA!
E You need to think about both of those.
E points "away" from the sheet: on the left side, away = left.
dA We also agreed that dA always points OUT of a closed surface,
(which means left on the left endcap)
That means E.dA = +EA again!
The sign is important: if E and A are parallel, the dot product is positive, even
if both point left. There IS a net flux out of the can, E fields poke "out" of
BOTH endcaps!
Note also that I'm using symmetry to claim that E(-L) = E(L) I'm calling them
both just "E".
Bottom
r r line: The left side of Gauss' law is the flux:
# E " dA = 0 + EA + EA = 2EA.
How about the right side, q(enclosed)?
Our pillbox has surrounded a circular patch of the sheet, with area A, so the
! charge enclosed must just be A.
Let's put it all together:
2 E(L) A = A/ "0 . The area cancels. (That's good, because A was an
artifically introduced area of our imaginary surface!)
And E(L) = /2 "0 . Hey, check it out: there's no dependence on "L"!
E is the !same at ANY distance from the sheet!

E = /2!"0 i if x>0
E = - /2 "0 i if x<0

!
!
 27-11 (SJP, Phys 1120)

One more example: Suppose you have a very long (infinite) LINE of charge,
with linear charge density (Coulombs/meter).
What is the electric field at some distance r away from this long line?
By now perhaps you see the method. First use symmetry to convince yourself
that E can only depend on "r", the distance away from the line.
Also, E must point straight away from the line at all points.
Then choose a sensible imaginary Gaussian surface: E
another "pillbox" (beer can!) shape seems promising
r
++++++
Then, write down Gauss' law for the pillbox surface:
r r qenclosed + L
$ E " dA = #0
E
The left side requires integrating over the two endcaps
plus the "can" itself. This time, it is the endcaps that both give zero
! contribution (because E doesn't "pierce" them, it's parallel to them

For the rest: at ANY little patch of area, E . dA is just E dA (because E points
straight out from the line at the middle, and so does dA.) Convince yourself!
Once again, E is the same value, E(r) everywhere on the surface (by
symmetry) so we can pull it out of the integral, and we have
E(r) " dA . But that funny integral just means "add up the little area squares on
the cylindrical surface", and we know that: the area of a cylinder is 2 r L.

! The other side of Gauss' law requires q(enclosed), which is just L (do you
see that?)
Putting it together, E(r)2 r L = L/ "0 .
The "L"'s cancel (Good thing! Because L was the length of the imaginary
surface, not anything physically real!) and we've got
E = 2 r L = /(2 r "0 ) !

This E field DOES depend on r, the distance from the line. But it doesn't drop
off like r^2 (like
! Coulomb's law), instead it drops off less rapidly.
27-12 (SJP, Phys 1120)

E fields and metals: Electrostatic equilibrium

I said E is defined everywhere in space. What is the electric field E inside a
chunk of metal? I claim, in steady state, "electrostatic equilibrium", the
answer must be zero! Why? Because metals are filled with free electrons that
can move around (they conduct) - so if E was NOT zero at some point inside,
then the force on an electron there would be nonzero - the electron would start
to move. They would continue to move, building up a "counter" E field, a
canceling field, until finally everything settles down, nothing moves, E=0
throughout the metal (so F=0) No more motion!
(Steady state MEANS everything has settled down, no charges are moving,
so the feel no net force on them any more)
It's a bit of a subtle argument - you have to think about it. But it's a very
important idea: E WILL ALWAYS BE EXACTLY 0 everywhere inside a
metal (conductor), in steady state.
If you try to add an external E field, the charges (conduction electrons) inside
the chunk of metal will quickly move around,
Eext until E=0 everywhere inside.
- + Eext
-- E=0 ++
+
- inside +
-

Gauss' law applied to an imaginary surface anywhere INSIDE the conductor

r r qenclosed
will give $ E " dA = #0
Since E=0 everywhere inside, the left side always vanishes.
Thus, q(enclosed) vanishes for any surface inside the chunk of conductor,
which
! tells you there are no net charges anywhere in there.
The separated charges must live right on the surface of the conductor!
(External field lines end on one surface, and start on the other: starting or
stopping on surface charges, but never going into the metal.)
The same argument tells you that, in steady state, if there's a nonzero E field
outside of a chunk of metal, it will always be perpendicular Eext
to the surface right at the edge. +
E=0 +
If it wasn't perpendicular, there would be a component of E +
+
inside +
parallel to the surface, and electrons right at the surface
would then flow (surface currents), until that piece of the E
field got cancelled out.
27-13 (SJP, Phys 1120)

There are tons of practical consequences of the above simple statements. Just one example:
if you make a metal box and put it in a region of large E field, the electrons in the metal
quickly (essentially instantly) rearrange on the surface to make E=0 everywhere INSIDE
the box. If you then hollow out the box, it makes no difference, E=0 inside, still. If there's a
lightning storm (large, potentially fatal E fields all around you), a relatively safe place to be
is inside a metal car, because the E field INSIDE the car (box) is zero. (It would be better if
the car was entirely metal) If you're in a fancy fiberglass-body car, too bad - fiberglass
doesn't conduct, it's not a metal, so the above arguments fail to hold. At least you'll stay
dry..

If you look at the surface of a conductor, we've argued above that there might
be a "surface charge density" . But inside, E=0 and there
won't be any other net charges in there. + E
+
So let's look right close to the edge of a conductor, any +
conductor, and apply Gauss' law to a small imaginary +
cubical surface, half of which is inside and half outside.

Since we argued that E is perpendicular to the surface, there will only be flux
through ONE side of the cube, (the right side in the picture) and that will
simply be EA. So Gauss' law says
EA = q(enc)/ "0 = A / "0
Thus E = / "0 just outside of a conductor in steady state.
That's always true! The charge density will "adjust itself" to make it true.
It's a!very general!statement, no approximations involved.
!
27-14 (SJP, Phys 1120)

Recall that we found a couple of pages ago that E = /2 "0 just outside of a
sheet of charges. There's a factor of two difference from the formula we just
found for E outside a conductor.
!
We need to think about this a bit, it's a puzzle! Because it looks
from the picture on the right like what we have is ...,well, it + E
looks like a sheet of charge with density !? +
+
So why do we get a different answer (by 2) than we got before? +
The answer is subtle and worth thinking about: The infinite
sheet problem was a very special case. We had to assume we had that sheet,
and nothing else, no other charges anywhere in the universe (!) Otherwise, the
E field from the other charges would superpose, add in, and we'd get a
different answer.
So the formula /2 "0 is JUST what you get from a single sheet and
NOTHING ELSE.
But for the case of the conductor, we're not making any such assumption. Not
only CAN ! there be other charges around... there MUST be, in order to ensure
that E=0 inside the conductor!! (If the ONLY charges were the ones shown
above, then E would NOT be zero inside, it would be /2 "0 to the left)
So there must be other charges around, and there field will add to the /2 "0
from the sheet alone. The magic of Gauss' law is that we know what this sum
is going to turn out to be, even without hunting around
! and looking for those
other charges. They have to be there, they will be there, and they'll! make the
total field turn out to be / "0 right outside the conductor.
E.g., perhaps the "larger" picture looks more like this: + +
Perhaps we have on the right side, and on the left + + + Etot
! +
side of the metal. The resulting E field over on the + +
right is really the superposition of the E fields caused + + +
+
by the two sheets,
Etot = E1+E2. But both E1 and E2 are /2 "0 , arising from a sheet of charge.
The total is / "0 , just as we said it has to be: right outside a conductor.
Inside, the fields from the two sheets will cancel, giving Etot=0, again is it has
to be. !
!
The field outside a conductor arises from ALL charges everywhere. The
formula E = / "0 outside a conductor just tells me what the total E field
outside a conductor is, taking into account all charges everywhere. It's really
pretty cool!
!