Vous êtes sur la page 1sur 40

10TheMagneticMomentandtheBlochEquations

"In mathematics you don't understand things. You just get used to
them." - Johann von Neumann
10.1ClassicalAngularMomentum
Now we can begin to use some of the fundamental topics discussed
to this point. We look here at the classical pictures of angular
momentum, magnetic moment and the classical description of nmr
spectroscopy via the Bloch equations. You will need some familiarity
with vector algebra and matrix algebra to proceed with this chapter.
There is no quantum mechanical interpretation used here so a full
picture of the subject is not really possible and certainly, some
aspects of nmr spectroscopy simply cannot be described using the
following ideas. However, these topics provide a solid base from
which we can begin to develop an image in our minds about how things
work.
A quick bit of introductory physics follows. Generally, there
are two types of spacial motion that a body can undertake,
translational and rotational. Translational motion is straight line
motion and of course, rotational motion is, well, rotational. Happily
for us, many of the equations that relate to translational motion
have analogues that describe rotational motion:
Translation Rotation
Displacement s
0
Velocity
v=
ds
dt
=
d
dt
Acceleration
a=
d v
dt
=
d
dt
Force and
torque
F
Mass and
moment of
inertia
m I
Kinetic Energy
E
k
=
1
2
mv
2
E
k
=
1
2
I
2
Momentum mv L=I
Table 10-1:Classical linear and rotational quantities
These analogues make it easy to make the transition from thinking
about translational motion to rotational motion. As nmr
spectroscopists, we shall be especially interested in the angular
force or torque, and the angular momentum, L .
Starting very simply, a vector r makes an angle q with the x-axis and
traces out a path or displacement of s units from the horizontal
axis. The angle q is in radians which is defined by the ratio of s to
r; ie. one radian is the angle created when s = r:
0=
s
r
s=r 0 [10-1]
Thus allows us to relate the translational velocity of the tip of r
to the angular velocity:
v=
ds
dt
=r
d0
dt
=r u [10-2]
as well as the translational acceleration:
a=
dv
dt
=r
du
dt
=r o [10-3]
The kinetic energy of motion is a bit more complicated (but not
overly so):
v=r u
E
k
=
1
2
mv
2
=
1
2
mr
2
u
2
Figure 10-1:Relationship
between radians and
circular displacement
r
q
s
This represents the kinetic energy of a single particle in the
rotating body; the total energy of all the particles is:
E
k
=
1
2

mr
2
u
2
The angular velocity, w, is constant for all particles and the
summation of

mr
2
is called the moment of inertia and is analogous
to our familiar concept of mass:
I=

mr
2
So:
E
k
=
1
2
I u
2
[10-4]
which now looks much like the translational kinetic energy equation.
When considering translational motion and work we generally use
force, F, and distance (or displacement), s, to calculate the work
done. Again, the rotational version of this calculation is very
similar. Distance, s, is replaced by an angle, q, and force is
replaced by torque, G:
W=I0
[10-5]
The net work done is the change in kinetic energy:
dW=Id0=d(
1
2
I u
2
)=
1
2
I du
2
I=
I
2
du
2
d0
=I u
du
d0
=I
d0
dt
du
d0
=I
du
dt
=I o
I=I o
This leads us to angular momentum:
I=I o=I
du
dt
=
d( I u)
dt
The product of I and w is the rotational analog of momentum for
translational motion and is given the symbol L.
I=
d( I u)
dt
=
dL
dt
I dt=dL
[10-6]
Ok .. so far there should not have been any great leaps in logic (as
long as you have your calculus wits about you).
Now, let's look at some vector quantities. First, consider a
tangential force vector, F at a distance r from the center of
rotation and a distance vector r as in the following figure. It is
apparent that this force will change the rotational velocity and the
angular momentum. Again, this is analogous to the translational
motion situation.
Now consider what would happen if F were to be directed not
tangentially but orthogonally. In other words, if F were to be
directed parallel and co-linear with r there would be no change in
the torque (however, see below). This is just what one would expect
of a vector relationship between r and F. Thus we write the vector
cross product:

I=

F
[10-7]
Here, the magnitude of the torque is:
I=rF sin()
where f is the angle between r and F. A 3D picture helps us to see
the relationships.
Figure 10-2:Tangential
force vector, F
F
r
When F is tangential then sin(f) is 1 and the torque is simply rF.
When F is co-linear with r, sin(f) is 0 and the torque is zero. Note
that here the torque is directed along the axis of rotation (however,
see below).
Similar considerations for linear and angular momenta lead us to:

L=

r m

v
=

p
[10-8]
where p is linear momentum. As with the torque, L is directed along
the axis of rotation. We can also represent L analogously to p in
linear mechanics:
p=mv

L=I u
This of course implies that w lies along the axis of rotation. A
Figure 10-3:Vector
relationships, radius,
force and torque
F
r
I=rF
I
Figure 10-4:Tangential
momentum
mv
r
moment's thought will convince you that this must be so. For a
rotating body with no external forces acting on it, w is constant.
If this is so then it must be directed along the rotation axis since
any other direction would imply that it changes direction with the
rotating body and therefore is not constant. A similar argument
applies to a, the angular acceleration and angular momentum, L. It
seems strange at first that these vectors are not pointed in the
direction of movement but it must be so for them to be constant.
We have seen, above, what happens when there is a tangential force
applied to the rotating body. A torque proportional to the force and
its distance from the rotation axis is generated and is directed
along the axis of rotation. In other words, the magnitude of the
torque changes, but not its direction (unless the force changes
direction).
It is possible to have two forces applied simultaneously to the
rotating body.
Figure 10-5:Vector
angular velocity and
torque
w a
L G
r
In this case the two forces are of equal magnitude and opposite
direction and are called a couple of moment. One encounters this
situation for example, when using a compass. What is the torque on
the system?
I=2

F
=2rF
[10-9]
Since the two forces in the diagram are tangential the angle between
r and F is 90 degrees and the magnitude of the torque is at a
maximum. Also, from the right hand rule, the torque vector is coming
out of the page towards you. If the angle that F makes with r changes
from 90 degrees the torque decreases as in the compass example.
What happens if the force, F, is directed along a line that
passes through the axis of rotation? We have already seen that this
must not change the magnitude of the torque. There is, however,
another effect which is of pivotal (pun intended) importance to us,
called precession.
Understanding this concept requires some pretty complex 3D
visualizations. Here is the basic (somewhat crowded!) picture:
Figure 10-6:Simultaneous
tangential forces
F
r
r
F
This is the example of the spinning top. There is a force, mg, on the
rotating body that goes through the axis of rotation. The torque is:

I=

F
=

rm

g
This vector cross product will be orthogonal to the plane defined by
r and mg. From [10-5], the torque is related to the angular momentum
or rather, the rate of change of the angular momentum. This being the
case, we see that dL (not L!) is in the same direction as the torque
and is at right angles to the angular momentum, L. As the top spins
about its axis of rotation the torque causes it to precess about an
axis of precession. This is the source of the wobbling motion of a
top. Note that in the case of a force at right angles to the axis of
rotation and not passing through the axis of rotation causes a change
in the magnitude of the angular momentum but a force passing through
the axis of rotation causes a change in direction of the angular
momentum.
Figure 10-7:Torque, force and precession
mg
r
L = Iw
q
q
dL
Lsin(0)
d
I
10.2TheMagneticMoment
The idea of the magnetic moment is pretty simple and comes from
some basic observations of the physical nature of electric currents
and magnetic fields. We wish here to describe the interaction between
a magnetic moment and a static magnetic field and also with other
magnetic moments. The most basic observation is that a charge, q,
moving through a magnetic field, B, with velocity, v experiences a
force, F, which is directed at right angles to both v and B:
The vector equation for this is:

F=q

B
This is, of course, a vector cross-product from which we can get
the magnitude of F quite easily (see eq. [3-16]):
F=qvBsin()
where f is the angle between v and B as in figure 9-8. We can write
the vector equation in terms of infinitesimals:
d F=dq(

B)
and since v represents the linear velocity of charge per unit time,
dl/dt, we can write:
d

F=
dq
dt
( d

B)
Figure 10-8:Force on a charged
particle
f
B
qv
F
If we integrate over the length of the wire to get the total force of
the current we get:

F=i

B
[10-10]
where i = dq/dt.
Now, if l and B are at right angles to each other then we can
calculate the magnitude of F directly:
F=ilBsin(

2
)
=ilB
where i is the current, l is the length of the wire and B is the
magnitude of the magnetic field. We assume that B is constant over
the length of the wire.
Now, take things a step further and imagine a current conducting
coil in a magnetic field at an angle f, to the field, B:
The first diagram is a side view of the coil which is at an angle to
Figure 10-9:Electric current loop
in a magnetic field
B
F = iaB
F = iaB
x x x x x x
x x x x x x
x x x x x x
x x x x x x
x x x x x x
a
F
F
b
F' F'
side
view
left
view
f
o
o

the field, B, and is able to rotate about an axis coming out towards
the viewer. The second is a left view. Here, the field, B, in the
second diagram is directed away from the viewer into the page as
indicated by the x's. In other words, the top diagram has been turned
counterclockwise as seen from the top to get the bottom diagram. The
current carrying coil is rectangular in shape with sides of length a
and b. From [10-10] the force, F, is then given by:
F=iaB
assuming a current of i amperes moving through the coil and that the
field, B, is at right angles to the movement of current. The force on
sides b is:
F'=ibBsin()
Forces F' on sides b cancel each other. Forces F are equal in
magnitude but are not co-linear (top diagram) so that they constitute
a couple of moment. From equation [10-9] the torque that this
produces is:
I=2

F
=2rFsin(o)
=2
1
2
bFsin(o)
=bFsin(o)
=iaBbsin(o)
which will cause the coil to begin to rotate. The torque vector will
be along the axis of rotation which, in the diagram, is co-linear
with F'. a times b is the area inside the coil so:
I=iABsin(o)
where A represents the area. iA is defined as the magnetic moment, m
where A is the vector area. The vector area is the usual area of a
surface with direction perpendicular to the surface. Thus the
magnetic moment of our coil lies along an axis perpendicular to the
coil.
I= Bsin(o)
where sin(a) is the angle between the magnetic moment and the applied
field. In vector notation:

I=

B
=
d

L
dt
[10-11]
A commonly used term is the gyromagnetic ratio (or the
magnetogyric ratio depending to whom you are talking). For a
classically spinning (periodic) system we can calculate what this is:
=i A
=
q
t
r
2
=
qv
2r
r
2
(t=
2r
v
)
=
q
2m
mvr
=
q
2m
L (eq.88|)
q/2m is the gyromagnetic ratio.
=
q
2m
L
=yL
[10-12]
or in vector and differential form:
=y

L
d
dt
=y
d

L
dt
=y(

B)
[10-13]
We see from this that the magnetic moment is proportional to the
angular momentum with the constant of proportionality being g.
Finally, we are in a position to discuss the magnetic moment of
a spin which, as nmr spectroscopists, we are naturally very
interested in. The really fascinating feature of spins is that they
contain intrinsic angular momentum as a fundamental property in the
same way that mass and charge are fundamental physical properties.
Hence the use of the word spin. This is a property of matter that
was wholly unknown less than a hundred years ago.
Our spin is labeled with its magnetic moment, and the
external magnetic field as B. The torque that experiences is (eq.
[10-11]):

I=
d

L
dt
=

B
or:
d

L=

I dt
=

Bdt
we can use the gyromagnetic ratio:
d

L=

Bdt
d =y(

B) dt
and
du=y Bsin(0) dt (from figure 8-10)
Figure 10-10:Magnetic moment of a
'spin' in an external magnetic
field
sin(0)
=y

L

B
0
d
d
Also from the figure (and eq. [10-1]):
d=sin(0)d
=y Bsin(0) dt
and therefore:
d
dt
=y B
u=
d
dt
u=y B
[10-14]
where w is the angular precession frequency. Why the negative sign?
From figure 8-10 the direction of precession is left-handed compared
to the direction of B. Using the right hand rule and aligning the
thumb with B, we see that the 'positive' direction is opposite to
that of df/dt. w is the Larmor frequency of precession of a spin and
equation 9-14 is the Larmor equation. Thus, if we know the values of
the gyromagnetic ratio and the applied field we can calculate the
precession frequency. For a proton in a 11.7 Tesla field this turns
out to be 500 MHz. Just to confuse things, the preceding applies to
nuclei with a positive g .. things are opposite for nuclei with a
negative g.
The energy of a magnetic moment in an external field may be
calculated using the differential version of equation [10-5]:
dW=I d0
and doing the indefinite integral, we get:
W=

I d0
=

Bsin(0) d0
= Bcos(0)+C
If we arbitrarily take the zero point of energy to be at q = 90
o
then
we can write for the energy:
E= Bcos(0)
or, in dot product vector notation:
E=

B
[10-15]
This is a fundamental equation in nmr spectroscopy, needed for
building the quantum mechanical Hamiltonian operator. Amazingly it is
usually simply stated and is not derived in many (if any) nmr texts.
The definite integral of the above energy equation [10-15] done
between q = 0 and q = p gives the energy difference between the
lowest and highest energy of the spinning magnetic moment:
W=

Bsin(0)d0
= Bcos(0) |
0

=2 B
Equation 9-15 expresses the classical energy associated with the
interaction of a magnetic moment and a static magnetic field. We now
are tempted to ask what is the classical energy of interaction of
two magnetic moments, m1 and m2?. One might suspect that the the
required expression is similar to equation 9-15 since each magnetic
moment 'feels' the field of the other. This is so. The derivation of
this expression is somewhat tedious so we will simple state the
expression here. The curious reader may, however, consult the
literature in the references; in particular, Weil, Cheston or Atkins
discuss this in varying levels of detail. The energy expression is:
E=

0
4
3cos
2
(0)1
r
3

1

2
10.3TheBlochEquations
This topic is one of the worst developed in many nmr texts.
Almost invariably the text will more or less simply state 'here they
are' without much logical development and perhaps refer the reader to
Bloch's original paper. The problem with this is that Bloch did a
not-so-good job of explaining them too! Fortunately, others (see
references) have published papers in which Bloch's equations are
rather more readably solved for various situations. We will also try
to produce them in a logical fashion below.
Using the ideas just developed concerning magnetic moments and
angular momentum, we can develop a description of the time evolution
of nuclear magnetic moments or just the nuclear magnetisation. These
equations are named after Felix Bloch, one of the giants of early nmr
spectroscopy along with Edward Purcell and Isador Rabi.
So far, we have considered only a single spin and its magnetic
moment. What about a large collection of spins that have been placed
in a magnetic field? We define now a quantity called the bulk
magnetic moment. This is the macroscopic version of the microscopic
magnetic moments of the nuclei in the sample and is the vector sum of
all magnetic moments in the bulk of the material of interest:

M=

i
N

i
[10-16]
Pictorially this looks something like:
Of course, at the instant that they are put into the field there will
be a completely random distribution of magnetic moments in all
directions. The bulk magnetic moment (or net magnetisation or net
polarisation in nmr-speak) is therefore zero. After some time however
this changes and a net magnetic moment, M, appears, aligned with the
external magnetic field, B0, and the spin system is said to be at
equilibrium. Chapter 14 deals with the mechanism of this process in
more detail.
Figure 10-11:Randomly oriented
magnetic moments
x
y
z
Now we look closely at this net magnetisation. At equilibrium
this magnetisation vector is aligned parallel to the z-axis as is the
external magnetic field, B0. Note that we are now subscripting B
since we will shortly be dealing with two different magnetic fields
and must keep track of which one we are talking about. We will label
the magnetisation vector in the equilibrium state as M0. Imagine that
we now move this vector away from its equilibrium position in some as
yet unspecified way and that the vector, now labeled M since it is
not in the equilibrium position, is not co-linear with the x, y or z
axes. We can break M down into its component vectors along the axes:
Figure 10-12:Excess spins oriented
with the external field and net
magnetic moment, M0
x
y
z

M
0
Vector M, being at an angle with respect to the z-axis and therefore
the applied field B0, will precess about the z-axis, just as the
individual magnetic moments will. Assuming that there will no
relaxation of M back to the equilibrium position along the z-axis and
that it will precess indefinitely we can write:
d

M
dt
=y

M

B
0
[10-17]
which is the macroscopic version of equation [10-13]. Using our
determinant method (eq. [2-3]) to expand this we get:
d

M
dt
=ydet

u
x
u
y
u
z
M
x
M
y
M
z
B
0x
B
0y
B
0z
|
=y( M
y
B
0z
M
z
B
0y
) u
x
y (M
x
B
0z
M
z
B
0x
)

u
y
+y( M
x
B
0y
M
y
B
0x
) u
z
=y( M
y
B
0z
)

u
x
( M
x
B
0z
)

u
y
+0

u
z
|
=y (M
y
B
0z
) u
x
(M
x
B
0z
) u
y
|
[10-18]
remembering that the field B0 is parallel to the z-axis and has no
components in the x-y plane. ux, uy and uz are of course unit basis
Figure 10-13 Component vectors of M

M
x

M
z

M
y
x
y
z
vectors. This result tells us that the magnetisation changes in the
x-y plane but not along the z-axis. In other words the magnetisation
vector, M, precesses about the z-axis as shown in the figure. Since
the projection of M onto the z-axis will (obviously) not precess
about the z-axis we can write:
d

M
z
dt
=0
or:
dM
z
dt
=0
[10-19a]
This should also be evident from the fact that dM/dt is at right
angles to the plane defined by M and B0 and will therefore have no
component on the z-axis. Mz is called the longitudinal magnetisation.
What about the magnetisation in the transverse (the x-y) plane?
Equation [10-18] indicates that all of the change in M is in the
transverse plane. From [10-18] it is reasonable to write:
dM
x
dt
=y M
y
B
0z
=M
y
u
0
[10-19b]
using the magnitude of the x-component of M and equation [10-14].
Also:
dM
y
dt
=y M
x
B
0z
=M
x
u
0
[10-19c]
Combining equations [10-19] into a vector equation:
d

M
dt
= u
0


M
or
d

M
dt
=y

M

B
0
[10-20]
where we have used the relation (eq. [10-14]):
u
0
=y

B
0
or

B
0
=
u
0
y
[10-21]
You might be forgiven for asking where did the negative sign
come from?. This is a bit of a puzzle since, yet again, many books
treat this very loosely or not at all. However, it is really very
simple:
d

M
dt
= u
0


M
=

M u
0
=

M( u
0
)
=

My

B
0
=y

M

B
0
Equations [10-19] and [10-20] refer to the special (and artificial)
case in which there is a magnetisation M in an external field B0 and
there is no relaxation taking place. They do however serve as a
starting point for our further deliberations.
Before we take into account relaxation let's look at the
situation when a linearly oscillating electromagnetic field, Bx,
moving back and forth along the x-axis is applied at right angles to
the external field, B0. Conceptually, the Bx field may then be broken
up into two counter-rotating vectors the sum of which always is Bx.
We will call these new vectors B1L and B1R for the moment and assign a
rotational frequency of +w1 and -w1 to them. The x and y components
of these two counter-rotating vectors are contained in:

B
1R
=B
1


u
x
cos(u
1
t)+

u
y
sin(u
1
t)|

B
1L
=B
1
u
x
cos(u
1
t) u
y
sin(u
1
t)|

B
1R
+

B
1L
=2B
1
u
x
cos(u
1
t)=

B
x
[10-22]
where B1R uses +w1 and B1L uses -w1.
Only one of these two B1 fields will concern us since the other
is rotating with a negative frequency relatively speaking and does
not affect any spins. We choose one and ignore the other. Let's
choose B1R and simply call it B1 for the rest of the discussion.
So, now we have two fields to consider, B0 and B1. B0 is, of
course, static and aligned along the z-axis (or more accurately, the
z-axis is aligned with B0) and B1 is at 90 degrees to the z-axis and
rotating about the z-axis at w1 radians per second. We can call the
sum of these two vectors Beff:
Figure 10-14:Counter-rotating applied
magnetic fields

B
1R
x
y
z

B
1L

B
x

B
0
Obviously, since B1 is rotating about the z-axis, Beff is
rotating about the z-axis as well. The size of B1 is much exaggerated
in the figure. Generally, B1 << B0 so that Beff is only slightly off
of the z-axis. Our new simple Bloch equations (still without
relaxation taken into account) are:
d

M
dt
=y

M

B
eff
=y

M(

B
0
+

B
1
)
=y

M

B
0
+y

M

B
1
=ydet

u
x
u
y
u
z
M
x
M
y
M
z
B
0x
B
0y
B
0z
|
+ydet

u
x
u
y
u
z
M
x
M
y
M
z
B
1x
B
1y
B
1z
|
=y( M
y
B
0z
M
z
B
0y
) u
x
+y (M
y
B
1z
M
z
B
1y
) u
x
y( M
x
B
0z
M
z
B
0x
) u
y
y( M
x
B
1z
M
z
B
1x
) u
y
+y( M
x
B
0y
M
y
B
0x
) u
z
+y( M
x
B
1y
M
y
B
1x
) u
z
=y( M
y
B
0z
M
z
B
1y
) u
x
y (M
x
B
0z
M
z
B
1x
) u
y
+y( M
x
B
1y
M
y
B
1x
) u
z
remembering that B0 is along the z-axis and B1 is in the transverse
plane and therefore B0x, B0y and B1z are zero. The rates of change of
each component of M are:
Figure 10-15:Effective magnetic field

B
1
x
y
z

B
0

B
eff
dM
x
dt
=y( M
y
B
0
M
z
B
1y
)
dM
y
dt
=y( M
x
B
0
M
z
B
1x
)
dM
z
dt
=y (M
x
B
1y
M
y
B
1x
)
[10-23]
B0 and B0z are equal since B0 is co-linear with the z-axis ...
hence the change from B0z to B0 in [10-23]. Using equation [10-22] we
can modify [10-23]:
dM
x
dt
=y M
y
B
0
M
z
B
1
sin(u
1
t)|
dM
y
dt
=y M
x
B
0
M
z
B
1
cos(u
1
t)|
dM
z
dt
=y M
x
B
1
sin(u
1
t)M
y
B
1
cos(u
1
t)|
[10-24]
Again, note that these equations do not include relaxation and assume
the presence of B0 and B1.
We have introduced the oscillating B1 field without saying why.
Something completely astounding and counterintuitive happens when we
do this! First, let's invoke the rotating frame of reference.
Specifically, let's use equation [3-44] and our magnetisation vector:
(
d

M
dt
)
LAB
=(
d

M
dt
)
ROT
+( u
1


M)
or
(
d

M
dt
)
ROT
=(
d

M
dt
)
LAB
( u
1


M)
Recall from equation [3-44] that (w x M) refers to the rotation of
the x,y and z-axes at vector frequency w. We are completely free to
choose what that frequency is so we choose the frequency of the
rotating B1 field, w1.
We also realize that the LAB portion of this equation can be
replaced by [10-20]:
(
d

M
dt
)
ROT
=( u
o


M)( u
1


M)
or
(
d

M
dt
)
ROT
=y(

M

B
0
)+(

M u
1
)
=

M( y

B
0
+ u
1
)
=y

M(

B
0
+
u
1
y
)
[10-25]
Note that, since the order of M and w1 were reversed, the sign of
their cross product changes.
Let's put B1 into the equation now:
(
d

M
dt
)
ROT
=y

M(

B
0
+
u
1
y
+

B
1
)
and define a 'fictitious field' (using [10-21]):

B
fic
=

B
0
+
u
1
y
=
u
0
y
+
u
1
y
=
u
1
u
0
y
[10-26]
and put it into the equation:
(
d

M
dt
)
ROT
=y

M(
u
1
u
0
y
+

B
1
)
=y

M(

B
fic
+

B
1
)
=y

M

B
r
[10-27]
where Br is the resultant field. This is the equation of motion of
the magnetisation in the rotating frame and has exactly the same form
as our previously developed equation for the laboratory frame,
equation [10-20].
So, we have w0 and w1 and M and B1. w0 is the Larmor frequency
of a spin in the static field, B0 and is itself constant. M is the
magnetisation due to the sum of the magnetic moments in the
collection of (many, many) spins in the sample. B1 is the rotating
magnetic component of the applied electromagnetic field. w1 is the
frequency of the rotating frame of reference which is also the
frequency of the B1 field.
What will happen if we vary w1 from zero up to w0? The term
u
1
u
0
y
corresponds to our 'fictitious field', Bfic, and will at first
be equal to B0 when w1 is zero. As w1 increases the value of Bfic
decreases and the vector sum, Br, of Bfic and B1 begins to tilt slowly
towards the x-y plane. As w1 approaches w0 the rate of tilt
increases until at w1 - w0 = 0, Br is entirely in the x-y plane. In
other words, Bfic vanishes! Thus when the frequency of irradiation is
equivalent to the Larmor frequency of the spin the external field
vanishes from the rotating frame of reference. Amazing!
Furthermore, if we vary w1 slowly the magnetisation, M, will
precess about Br and end up precessing about B1 in the transverse
plane. When w1 - w0 = 0 we say that we are at (or on) resonance ...
hence the name, nuclear magnetic resonance. Also, since M now sees no
external field in the rotating frame its internal precession
frequency about the z-axis is zero relative to the rotating frame.
You can see this in equation [10-26] when w1 w0 = 0.
Our equations so far have assumed the projections of M onto the
laboratory frame axes, Mx, My, and Mz (figure 9-13). We ask now, what
are the projections of M and Br onto the rotating frame? We assume
that the rotating frame frequency is that of B1 and that B1 is
aligned along the rotating x'-axis. We will use x' and y' to denote
the rotating axes and x and y for the static axes. Of course, the z-
axis is the same in both frames.
Figure 10-16:Resultant
magnetic field

B
1

B
fic

B
r
Figure 10-17
x
y
z

B
fic
x'
y'

B
1
Figure 10-18
x
y
z
x'
y'

M
x'

M
y '

M
z
Figure 9-17 shows that B1 is aligned along the x'-axis and Bfic
is aligned along the z-axis. Figure 4-18 shows the vector components
of M in the rotating frame. With reference to figure 9-17 we realize
that the magnitude of Br along the rotating frame axes is B1 along
the x'-axis, zero along the y'-axis and Bfic or
u
1
u
0
y
along the z-
axis. Equation [10-27] can be expanded using these components:
(
d

M
dt
)
ROT
=y

M

B
r
=y det

u
x '
u
y '
u
z
M
x
' M
y
' M
z
B
1
0
(u
1
u
0
)
y
|
=y( M
y'
(u
1
u
0
)
y
) u
x'
y( M
x'
(u
1
u
0
)
y
M
z
B
1
) u
y '
y M
y'
B
1
u
z
[10-
28]
The components of
d

M
dt
along the (rotating) coordinate axes are:
dM
x'
dt
=M
y'
(u
1
u
0
)
[10-29a]
dM
y'
dt
=M
x'
(u
1
u
0
)+y M
z
B
1
[10-29b]
dM
z
dt
=y M
y'
B
1
[10-29c]
These are the Bloch equations 'anchored' to the rotating frame
as it were, rather than the laboratory or static frame (equation 4-
24). Note that the sine and cosine terms are gone. This is reasonable
since B1 is now 'motionless' in the rotating frame and therefore
should naturally be expected to be time-invariant in this frame of
reference.
Now, finally, we will include relaxation. What is relaxation? As
it turns out, there are two types that will be of interest to us.
First, if you do an inversion recovery experiment you can watch the
change of Mz from non-equilibrium back to equilibrium where it is
equal to M0. The description of this experiment is beyond the scope
of this document but there are many good references to it. The change
from non-equilibrium to equilibrium is, as it turns out, an
exponential one and we can model it as:
M
0
M
z
=Ae
t
T
1
or
ln( M
0
M
z
)=A'
t
T
1
where M0 is the equilibrium magnetisation (and is constant) and Mz is
the magnetisation at time t (and is therefore variable). We model it
this way so that we have positive quantity that decays towards zero
in an entirely analogous way to radioactive decay. A and A' are
constants whose values we need not be concerned about here since they
will disappear. Now, differentiating with respect to time:
1
M
0
M
z
dM
z
dt
=
1
T
1
or
dM
z
dt
=
M
0
M
z
T
1
[10-30a]
Don't confuse this with equation [10-24]! Equation [10-30a] refers to
the rate of change of Mz due to relaxation processes only. What about
Mx and My? If we think a bit about what must happen when M is tipped
away from the z-axis.
In the x-y plane there is a similar (mathematically) decay.
Magnetisation here decays according to:
M
x
=Ae
t
T
2
or
ln( M
x
)=A'
t
T
2
1
M
x
dM
x
dt
=
1
T
2
dM
x
dt
=
M
x
T
2
[10-30b]
and similarly:
dM
y
dt
=
M
y
T
2
[10-30c]
If we add these in to our current Bloch equations ([10-29])we
get:
dM
x'
dt
=M
y'
(u
1
u
0
)
M
x'
T
2
[10-31a]
dM
y'
dt
=M
x'
(u
1
u
0
)+y M
z
B
1

M
y'
T
2
[10-31b]
dM
z
dt
=y M
y'
B
1

M
z
M
0
T
1
[10-31c]
In the old CW (continuous wave) spectrometers, one swept either
the frequency w1 or the field B1 until the resonance condition was
reached, as discussed above. This was done slowly so as to maintain
steady-state conditions. Under these circumstances, Mx', My' and Mz
remained constant in the rotating frame. Therefore, we write:
0=M
y'
(u
1
u
0
)
M
x'
T
2
0=M
x'
(u
1
u
0
)+y M
z
B
1

M
y'
T
2
0=y M
y'
B
1

M
z
M
0
T
1
This being three equations in three variables, Mx', My' and Mz, we can
solve for each of the magnetisation components using our matrix
algebra. First, to simplify things a bit we will rewrite our Bloch
equations:
M
x'
+T
2
(u
1
u
0
) M
y'
+0M
z
=0
T
2
(u
1
u
0
) M
x'
M
y'
+yT
2
B
1
M
z
=0
0M
x
'yT
1
B
1
M
y'
M
z
=M
0
Here we have multiplied by T1 or T2 as needed to get rid of the
fractions and added 0Mx' and 0Mz for clarity. Also,
u
1
u
0
y
has been
replaced by Bfic as it was defined in equation 9-26. It is still
cluttered so let's simplify a bit further by letting:
a=T
2
(u
1
u
0
)
b=y T
2
B
1
c=yT
1
B
1
The set of equations now looks like:
M
x'
+aM
y'
+0M
z
=0
aM
x'
M
y'
+bM
z
=0
0M
x
'c M
y'
M
z
=M
0
and when we put them into matrix form:

1 a 0
a 1 b
0 c 1
|

M
x '
M
y '
M
z
|
=

0
0
M
0
|
Recalling the information on solving equations via matrices we
construct the inverse matrix of the coefficients of M, being very
careful about signs of cofactors and transposing to get the adjoint
matrix:
adj

1 a 0
a 1 b
0 c 1
|
det

1 a 0
a 1 b
0 c 1
|

1 a 0
a 1 b
0 c 1
|

M
x'
M
y'
M
z
|
=
adj

1 a 0
a 1 b
0 c 1
|
det

1 a 0
a 1 b
0 c 1
|

0
0
M
0
|
which becomes:

M
x'
M
y'
M
z
|
=
adj

1 a 0
a 1 b
0 c 1
|
det

1 a 0
a 1 b
0 c 1
|

0
0
M
0
|
det

1 a 0
a 1 b
0 c 1
|
=1(1+bc)a(a0)
=(bc+1+a
2
)
adj

1 a 0
a 1 b
0 c 1
|
=

1+bc a ab
a 1 b
ac c 1+a
2|
So we have:

M
x'
M
y'
M
z
|
=

1+bc a ab
a 1 b
ac c 1+a
2|
(bc+1+a
2
)

0
0
M
0
|
=

M
0
ab
M
0
b
M
0
(1+a
2
)
|
(bc+1+a
2
)
and we finally arrive at our solutions (after substituting for a, b
and c from our definitions and factoring -1 from the numerator and
denominator):
M
x
'=
y M
0
T
2
2
B
1
(u
1
u
0
)
y
2
T
1
T
2
B
1
2
+1+T
2
2
(u
1
u
0
)
2
M
y
'=
y M
0
T
2
B
1
y
2
T
1
T
2
B
1
2
+1+T
2
2
(u
1
u
0
)
2
M
z
=
M
0
(1+T
2
2
(u
1
u
0
)
2
)
y
2
T
1
T
2
B
1
2
+1+T
2
2
(u
1
u
0
)
2
It is generally the case that B1 is only few milligauss and if
T1 and T2 are only a few seconds each:
y
2
B
1
T
1
T
2
1
and our equations in the x-y plane can be approximated:
M
x
'=
y M
0
T
2
2
B
1
(u
1
u
0
)
1+T
2
2
(u
1
u
0
)
2
M
y
'=
y M
0
T
2
B
1
1+T
2
2
(u
1
u
0
)
2
A plot of My' vs w1 looks like:
This is the Lorenzian or absorbtion line shape. Mx' vs w1 plots as:
and is the dispersion line shape.
10.3TheVectorModelofNMRSpectroscopy
Figure 10-19
Figure 10-20
Strictly speaking this topic is not mathematical as much as it
is pictorial however it follows directly from the development of the
Bloch equations and is of much utility in introductory nmr
spectroscopy.
We have spoken of the equilibrium magnetisation and pictured it
as a vector in cartesian space in figure 8-12. We have also spoken of
how it is that this vector can move through this cartesian space via
the effect of an applied linearly oscillating magnetic field, B1
(figure 8-14).
Let's suppose that we have a spin system at equilibrium in an
external magnetic field and we apply a B1 field pulse. How can we
picture what will happen given what we have discussed so far? We have
seen that in the rotating frame of reference if the B1 frequency is
at or very near w0, the intrinsic Larmor frequency of the spin, the
external field B0 effectively collapses to zero (equation 8-26 with
w0 = w1 ). Thus, what the equilibrium magnetisation 'sees' is a
static B1 field oriented along an axis in the x-y plane:
What will happen? The equilibrium magnetisation will precess about
B1 of course.
Figure 10-21
x
y
z

M
0

B
1
In Fourier Transform nmr spectroscopy, turning on the B1 field for
enough time that the magnetisation rotates by 90 degrees into the x-y
plane is commonly referred to as issuing a 90 degree pulse.
Similarly, a 180 degree pulse effects a rotation of the magnetisation
by 180 degrees.
Figure 10-22
x
y
z

B
1
When we turn off the B1 field what happens? If the magnetisation
has been rotated into the x-y plane it will now precess about Bfic,
which will be zero if the pulse is 'on resonance' and non-zero
otherwise. 'On-resonance' generally means w0 = w1 in equation 8-26.
For a set of spins with differing chemical shift frequencies only one
spin can experience an on-resonance pulse while the others will
experience varying degrees of Bfic. It is easiest to begin describing
pulse sequences with on-resonance pulses. So, we have a spin whose
magnetisation is in the x-y plane as a result of an on-resonance 90
degree pulse. This being the case, in the rotating frame there will
be no apparent chemical shift evolution since the Larmor frequency of
the spin is the same as the rotating frame frequency. It will be
motionless.
Scalar coupling evolution can be represented vectorially as a
splitting of the vector into two or more vectors with differing
rotational frequencies in the rotating frame. If we imagine an on-
resonance 90 degree pulse issued to a spin that is split into a
doublet our vector model will have the vector in the x-y plane split
into two 'sub vectors', one rotating clockwise in the rotating frame
and one rotating counterclockwise. Why is this? Well, an on-resonance
pulse is issued at the chemical shift frequency of the spin. Thus the
'zero' frequency is the chemical shift frequency; that is, in the
rotating frame the chemical shift of the spin appears to be zero.
However, for a doublet in an nmr spectrum the chemical shift is half-
Figure 10-23
x
y
z

B
1
way between the two peaks observed. Thus one peak is slightly behind
the rotating frame frequency and one is slightly ahead. Therefore, in
the rotating frame we picture two counter-rotating vectors:
Note that we have changed our perspective here for convenience. We
are now looking down from 'above' the x-y plane. The coupling
constant, usually measured in hertz, is the frequency difference
between the peaks of the multiplet in the spectrum. For a doublet
this is J hz and in our picture, one vector rotates with relative
frequency +J/2 hz and one with relative frequency -J/2 hz. Why J/2?
Remember that we are thinking in the rotating frame of reference and
that the chemical shift frequency is zero. Since the frequency
difference between the doublet peaks is J hz, one of the two doublet
peaks will rotate at +J hz and the other at -J hz relative to the
chemical shift frequency.
Now, suppose we wait for a period of time after the B1 field is
turned off. Furthermore, let's assume an on-resonance pulse for
simplicity and clarity and a doublet with coupling constant J hz. If
the time period is 1/(2J) seconds where will the doublet vectors be?
We need to calculate how far the vector will have rotated:
0=ut
=2 J
1
2J
=
This gives the angle between the two counter-rotating vectors. Half
of this then will be the angle that one of the vectors will have
rotated through:
Figure 10-24
x
y
0=ut
= J
1
2J
=

2
Thus, one vector which is rotating clockwise will rotate +90 degrees
and the other -90 degrees:
We can apply these simple principles to several pulse sequences. The
APT (Attached Proton Test) experiment consists of pulses on two
channels, usually
1
H and
13
C:
The effect of the
1
H decoupler is to collapse multiplets into
singlets. In other words, multiplet vectors collapse to a single
vector located at the chemical shift frequency. So, in this pulse
sequence there is coupling evolution during the first delay period of
1/J seconds but not during the second delay period. How is this
useful? During the first delay period the two counter-rotating
vectors of the doublet will each rotate by 180 degrees and will be
Figure 10-25
x
y

2

H
1
C
13
decoupler
acquisition
1
J
1
J
co-linear along (say) the -y-axis. The 180 degree pulse on
13
C
rotates them back to where they started. The decoupler collapses them
to a single vector which will now exhibit only chemical shift
evolution. Assuming on-resonance pulses, this will be zero so that
the vector is motionless in the rotating frame during the second
delay period. Acquisition then gives a spectrum with a (decoupled)
13
C peak in it.

The interesting part of all of this is what happens when we
consider now a triplet, again with coupling constant J Hz. Still
assumming an on-resonance pulse, the vector representing the center
peak will be motionless in the rotating frame since it is also at the
chemical shift of the multiplet and the other two vectors will rotate
at +J hz and -J hz with respect to the chemical shift frequency. This
is just twice that of the doublet so each of the 'outer' vectors will
rotate twice as far in 1/J seconds. Using our APT pulse sequence
analysis from above the vector diagrams will look like:
Figure 10-26
1
J
1
J

x
{ H
1
}
The vector at acquisition time is now pointed 180 degrees away from
where the double vector was. This will result in a peak in the
13
C
spectrum that is 180 degrees out of phase with the doublet peak. The
APT experiment is a very useful way to feret out the multiplicity of
13
C peaks.
The vector model is very useful for introducing the analysis of
pulse sequences but there comes a point at which it breaks down.
Specifically, when multiple quantum coherence is generated it is not
possible to follow the evolution of the magnetisation using the
vector model. Fear not however. There are other ways to do so.
10.4Problems
10.5References
1. F. Bloch, Physics Review 70, 460-473 (1946).
2. H.C. Torrey, Phys. Rev., 76, 1059-1068, (1949).
3. G.A. Morris and P.B. Chilvers, J. Magn. Res. A, 107, 236-238,
(1994).
4. P.K. Madhu and A. Kumar, J. Magn. Res. A, 114, 201-212, (1995).
5. P.K. Madhu and A. Kumar, Concepts in Magnetic Resonance, 9, 1-
12, (1997).
Figure 10-27
1
J
1
J

x
{ H
1
}
6. F.A. Bovey, L Jelinski and P.A. Mirau, Nuclear Magnetic
Resonance Spectroscopy, Academic Press, 1988.
7. M.L. Martin, J.J. Delpuech and G.J. Martin, Practical NMR
Spectroscopy, Heyden and Son Ltd., 1980.
8. J.H. Nelson, Nuclear Magnetic Resonance Spectroscopy, Pearson
Education Inc., 2003.
9. J.A. Pople, W.G. Schneider and H.J. Bernstein, High-resolution
Nuclear Magnetic Resonance, McGraw-Hill Book Company, 1959.
10.C.P. Slichter, Principles of Magnetic Resonance, Springer-
Verlag, 1992.
11.D. Shaw, Fourier Tranform N.M.R. Spectroscopy, Elsevier, 1984.
12.J.A. Weil and J.R. Bolton, Electron Paramagnetic Resonance, John
Wiley and Sons Ltd., 2007.
13.W.B. Cheston, Elementary Theory of Electric and Magnetic
Fields, John Wiley and Sons Ltd., 1964.
14.P.W. Atkins and R.S. Friedman, Molecular Quantum Mechanics,
Oxford University Press, Fourth Edition, 2005.

Vous aimerez peut-être aussi