Pulsed Nuclear Magnetic Resonance: Spin Echoes

MIT Department of Physics

(Dated: October 26, 2017)
This experiment explores nuclear magnetic resonance (NMR) both as a physical phenomenon
concerning atomic nuclei and as a versatile laboratory technique for exploring the properties of
nuclei, the structure of molecules, and certain other properties of substances. Using radio frequency
bursts tuned to resonance, pulsed NMR perturbs a thermal-equilibrium spin ensemble, which then
behaves on average much like a classical magnetic dipole. One consequence is the ability to measure
the magnetic moments of certain nuclei such as that of hydrogen, i.e., the proton, and of fluorine.
The former is of particular interest in nuclear physics.
In addition, the experiment involves a variety of pulse sequences that, in turn, utilize spin echoes
for the determination of spin-lattice and spin-spin relaxation times of substances. Investigation
of the strong dependences of these times on viscosity and on paramagnetic ion concentrations in
samples containing glycerin and water illustrates how pulsed NMR may be used for identifying and
characterizing substances.

I. PREPARATORY QUESTIONS observed in the time intervals between the applied RF

bursts. It is experimentally much easier to detect the ex-
Please visit the Pulsed NMR chapter on the 8.13r web- tremely small effects of the energy-level transitions if the
site at lms.mitx.mit.edu to review the background ma- detection phase is separated in time from the RF burst
terial for this experiment. Answer all questions found in phase. The techniques of pulsed NMR are particularly
the chapter. Work out the solutions in your laboratory advantageous in sorting out various relaxation effects.
notebook; submit your answers on the web site. The present experiment demonstrates an essential pro-
cess common to all NMR techniques: the detection and
interpretation of the effects of a known perturbation on
II. PROGRESS CHECK a system of magnetic dipoles embedded in a solid or liq-
uid. As we shall see, analysis of the system’s response to
By the end of your second lab session you should have what is essentially a macroscopic perturbation may yield
a determination of the nuclear magnetic moment of fluo- interesting information about the microscopic structure
rine. You should also have a preliminary value of T2 for of the material.
100% glycerine.
IV. THEORY
III. BACKGROUND
IV.1. Precession of a Classical Magnetic Moment
and Free Induction
The NMR method for measuring nuclear magnetic mo-
ments was conceived independently in the late 1940s
by Felix Bloch and Edward Purcell, who were jointly In classical electromagnetism, a charged body with
awarded the Nobel Prize in 1952 for their work [1–3]. nonzero angular momentum L ⃗ possesses a magnetic mo-
Both investigators, applying somewhat different tech- ment µ
⃗ , defined by
niques, developed methods for determining the magnetic
moments of nuclei in solid and liquid samples by measur- µ ⃗
⃗ = γL (1)
ing the frequencies of oscillating electromagnetic fields
that induced transitions among the nuclear magnetic where γ is the body’s gyromagnetic ratio, a constant that
substates resulting in the transfer of energy between the depends on the mass and charge distributions. For a clas-
sample and the measuring device. Although the amounts sical body with identical spatial distributions and mo-
of energy transferred are extremely small, the fact that tions of its mass m and charge q, the gyromagnetic ratio
the energy transfer is a resonance phenomenon enabled is given by
it to be measured. q
Bloch and Purcell exposed their samples to a contin- γcl = . (2)
2m
uous wave (CW) electromagnetic field of constant ra-
diofrequency while simultaneously sweeping a superposed The magnetic moment is an interesting quantity be-
“DC” magnetic field through the resonance condition. cause when the body is placed in a static magnetic field
CW methods are rarely used in modern NMR experi- ⃗ 0 , it experiences a torque
B
ments. Rather, radio frequency (RF) fields are usually
applied in the form of short bursts (“pulsed NMR”) and ⃗
dL
=µ ⃗ 0.
⃗ ×B (3)
the effects of the induced energy level transitions are dt

gitinfo: amlevine 2017-10-26 —

 gitinfo: amlevine 2017-10-26 — 2

This equation of motion implies that if there is some There is no reason we should expect such a system
nonzero angle α between L ⃗ and B ⃗ 0 , the axis of rotation to behave anything like a classical particle with angular
⃗ precesses about B⃗0 at the rate
L momentum. Yet, as shown in the references, the wave-
function for any two-state system can be visualized as
⃗0
⃗ L = −γcl B
ω (4) a vector in the Bloch sphere, where the angular coordi-
nates are obtained from relative magnitudes and phases
independently of the value of α, much like the behavior of in the superposition. It turns out that, in this picture,
a gyroscope in a uniform gravitational field. The minus when a charged quantum spin such as an atomic nucleus
sign indicates that for a body with positive γ that is char- is placed in a static magnetic field, the Bloch sphere rep-
acteristic of, e.g., a positively charged sphere, the preces- resentation of the wavefunction does indeed exhibit an
sion is opposite the sense given by the right-hand rule analog of Larmor precession. Of course, there are key
with respect to B⃗0 . This phenomenon is called Larmor differences. For one thing, the precession frequency is
precession, and we call ωL = |ω⃗L | the Larmor frequency. not the same, owing to quantum effects. The difference
Suppose B ⃗ 0 = B0 ẑ; in this case, the x-y plane is is captured by changing the gyromagnetic ratio:
called the transverse plane. Next, suppose we place a
solenoid around the magnetic moment, with the axis of γ = gγcl . (6)
the solenoid in the transverse plane (e.g., aligned with
x̂). If α is nonzero, there is a nonzero transverse compo-
nent of µ, which generates an oscillating magnetic field at This corrective g-factor is analogous to the Landé g-
the Larmor frequency. By Faraday’s law, this transverse factor in atomic spectroscopy and varies among types of
component induces an emf in the solenoid nuclei. A corresponding change applies to the magnetic
moment, i.e., in an eigenstate of one component of angu-
V (t) = V0 (α) sin (ωL t + ϕ0 ) . (5) lar momentum of a spin- 12 particle, the magnitude of the
magnetic moment along the direction of quantization is
The phase ϕ0 is simply the initial angle between the µ = γℏ/2.
transverse part of µ ⃗ and the solenoid axis, while V0 is More significantly, however, precession of the wave-
an overall factor that incorporates the magnitude of the function in the Bloch sphere is not quite the same as
transverse part of µ⃗ , the number of turns of wire in the a precession in real space. For example, with a classical
solenoid, the solenoid dimensions, etc. It is worth noting magnetic moment µ ⃗ aligned at some nonzero angle α with
that V0 is a function of α: V0 is zero when α = 0 or ẑ, it is possible to measure precisely both µx = µ⃗ · x̂ and
α = π. In between these extremes, the signal reaches a µz = µ ⃗ · ẑ. But Heisenberg’s uncertainty principle for-
maximum at α = π/2, i.e., when µ ⃗ lies in the transverse bids simultaneous precise measurements of both µx and
plane. µz , even though the wavefunction vector in the Bloch
Regardless of its magnitude, the induced emf always sphere has well-defined projections onto x̂ and ẑ.
oscillates at the characteristic Larmor frequency. We call
this detected voltage oscillation the free induction NMR Nevertheless, the expectation of the quantum magnetic
signal. It is this signal with which NMR is primarily moment does behave very nearly like the classical mag-
concerned – we manipulate the magnetic moments in a netic moment, in the limit of a large number of repeated
sample and monitor the behavior of the resulting free measurements, per Ehrenfest’s theorem. It is precisely
induction signal to understand the sample’s bulk material this correspondence that NMR relies on. But rather than
properties. repeating the measurement of a single quantum spin, we
may make one measurement of a large number of spins at
once; this is an ensemble measurement. If the spin wave-
IV.2. Nuclear Magnetism functions of the spinning particles are approximately the
same, i.e., the spins are ”coherent”, then the macroscopic
magnetization of the ensemble should behave classically.
You have seen in quantum mechanics that particles
like electrons, protons, and composite nuclei each pos- It is worth emphasizing that what we measure in NMR
sess an intrinsic quantity of angular momentum known is M ⃗ = N ⟨⃗ µ⟩, or the expectation of the magnetic mo-
as spin, which cannot quite be understood as any form ment averaged over the bulk ensemble, multiplied by the
of classical rotation. The particle’s spin angular momen- number of spins which are coherent. Since the volume
tum along any given direction is quantized, and, for a of our sample is fixed, this is proportional to the mag-
spin 12 particle, takes on the values of +ℏ/2 and −ℏ/2, netization, and we will simply refer to M ⃗ as the ”mag-
corresponding to two states we usually refer to as ”spin netization vector” of the sample. Our conclusion is that
up” and ”spin down”. The general state (wavefunction) M⃗ exhibits dynamics similar to a classical magnetic mo-
of any such two-state system is a complex superposition ment µ ⃗ , generating an NMR free induction signal at the
of these two spin eigenstates. We sketch a rough picture Larmor frequency ωL = γB0 . It is worth noting that this
below of how macroscopic nuclear magnetism comes out macroscopic ensemble does indeed capture the quantum
of this microscopic framework; for more accurate details, nature of the spins, by way of the non-classical gyromag-
see [4] and other reference works. netic ratio γ.
 gitinfo: amlevine 2017-10-26 — 3

IV.3. Pulsed NMR The total magnetic field in the rotating frame is, of
course, B⃗ =B ⃗0 + B
⃗ 1 . This magnetic field should, even
The equilibrium state for a system of spins in a static in the rotating frame, induce a Larmor precession. The
magnetic field B⃗ 0 = B0 ẑ will have a small magnetization precession angular velocity of the magnetization vector
⃗
M0 = M0 ẑ in the same direction as the magnetic field. in this frame is
Such a configuration, in which α = 0, does not produce ⃗ = −γ B
⃗ + ωẑ ′ ,
a free induction signal. We need a way of perturbing the Ω (15)
spins out of equilibrium in order to generate a transverse
component of the magnetization. We use the methods of where the extra term comes from the kinematic motion of
pulsed NMR in order to achieve this. the rotating frame as if there were a fictitious magnetic
field opposing B⃗ 0 . Written more explicitly in terms of
In pulsed NMR, the same solenoid that is used to pick
up a transverse magnetization can also be used to gener- components, we have
ate an RF field around and within the sample. We focus
on the generation of a magnetic field ⃗ · x̂′ = − γB1 − γB1 cos 2ωt
Ω (16)
2 2
⃗ 1 (t) = B1 cos ωt x̂.
B (7) γB1
⃗ ′
Ω · ŷ = − sin 2ωt (17)
2
It is standard to take B1 ≪ B0 , so that the field B ⃗1 ⃗ · ẑ ′ = −γB0 + ω.
Ω (18)
can be treated as a perturbation. We then proceed to
examine the behavior of M⃗ in response to this perturbing
Now the crucial point: when the frequency ω of the per-
field. turbing field satisfies ω = γB0 = ωL , i.e., is on resonance
⃗ 1 as a superposition
We first note that we can write B with the natural Larmor frequency, the rapid precession
of two counter-rotating magnetic fields: due to B⃗ 0 vanishes in the rotating frame, and almost all
that remains is a constant, slow precession about x̂′ at
⃗ r (t) = B1 (cos ωt x̂ + sin ωt ŷ)
B (8) the rate γB1 /2. The sine and cosine terms produce a tiny
2
B1 high-frequency (2ω) fluttering whose effects average out
⃗
Bl (t) = (cos ωt x̂ − sin ωt ŷ) . (9) on time scales longer than the reciprocal of the Larmor
2 frequency.
Thus, B⃗1 = B ⃗r + B⃗ l . For nuclei with positive gyromag- ⃗ is initially parallel to B
If M ⃗ 0 , then application of the
⃗
netic ratios, Bl rotates in the same direction as the mag- perturbing field B ⃗ 1 for a time
netization (clockwise), while B ⃗ r rotates in the opposite
π
direction (counter-clockwise). t90 = (19)
We now consider the situation from the point of view of γB1
an observer in a reference frame rotating in the direction
evidently rotates M ⃗ by 90◦ about x̂′ , placing M⃗ in the
of precession, that is, clockwise, with angular velocity ω.
The unit vectors in this rotating frame are ⃗
transverse plane perpendicular to B0 . If the pulse is now
turned off, M ⃗ is left in the transverse plane, and from
x̂′ = cos ωt x̂ − sin ωt ŷ (10) the point of view of an observer in the laboratory frame,
ŷ ′ = sin ωt x̂ + cos ωt ŷ (11) it will precess at the Larmor frequency γB0 about ẑ. By
ẑ ′ = ẑ. (12) a similar argument, application of the perturbing pulse
for a time t180 = 2t90 rotates M ⃗ by 180◦ , inverting the
A moment’s thought will confirm that in the x-y coordi- spin population. In practice, the value of B1 is not well-
nate system, the x′ -y ′ system is indeed rotating clockwise known, so t90 is usually found by trial and error, usually
with angular velocity ω. by looking for the pulse width which yields the greatest
In this rotating frame, the field B ⃗ l appears to be transverse magnetization.
stationary, while B ⃗ r appears to be rotating counter-
clockwise at a rate 2ω. This can be shown directly by
solving for x̂ and ŷ in terms of x̂′ and ŷ ′ in the equa- IV.4. Relaxation
tions above and substituting. The result is that the two
rotating components become Owing to the microscopic nature of nuclear magnetism,
the free induction signal does not persist for very long.
⃗ r = B1 (cos 2ωt x̂′ + sin 2ωt ŷ ′ )
B (13) Once perturbed, the spins continue to precess around B ⃗0
2
1 and, at the same time, the bulk magnetization proceeds
⃗
Bl = B1 x̂′ . (14) to return towards equilibrium. The processes involved
2
in returning to equilibrium are referred to as relaxation,
On the other hand, the static magnetic field does not and are key to the utility of pulsed NMR: different sub-
⃗ 0 = B0 ẑ ′ .
appear any different, and B stances return to equilibrium at different rates and in
 gitinfo: amlevine 2017-10-26 — 4

different ways; analysis of the relaxation times of a sam- The second mechanism is typically called spin-spin re-
ple gives significant insight into its chemical composition laxation and the time constant governing its rate is de-
and structure. noted T2 . Its name derives from the idea that the pre-
Relaxation mechanisms (and other effects, as discussed cession of the spin of a particular particle may be af-
in the following section) result in an exponential decay fected by the magnetic fields from the magnetic dipole
in the free induction NMR signal, which manifests itself moments of neighboring spins, perhaps other nuclei in the
as the ubiquitous free induction decay (FID) signal, a same molecule. These fields may be constant, in which
sketch of which is shown in Figure 1. case they may affect the precession rate and direction, or
they may fluctuate and thereby tend to randomize the
90◦ RF Pulse spin’s precessional motion. The differences in the field at
the locations of different particles causes the ensemble of
envelope = exp(−t/T2∗ ) spins, which were initially precessing in phase and con-
structively contributing to the transverse magnetization,
to decohere in precession phase and thereby diminish the
observed transverse magnetization. This spin-spin relax-
ation process, like the spin-lattice relaxation, is an irre-
versible process, and it can be described by saying that
the transverse magnetization Mxy decays approximately
according to

Mxy (t) = M0 e−t/T2 , (21)

Free Induction Decay (FID)
where M0 is the initial transverse magnetization at t = 0
right after a 90◦ pulse.
FIG. 1. An idealized scope trace of a free induction decay
Most of the measurement techniques used in this lab
signal, showing also the decay envelope. The thick black line
indicates the 90◦ perturbing pulse that puts the magnetiza- center on the goal of obtaining these relaxation times for
tion into the transverse plane. The decay constant T2∗ is de- various substances. In particular, we are interested in
termined by both the T2 effect discussed in the text as well how these relaxation times change as we vary the prop-
as the effects of field inhomogeneities discussed in the next erties of the sample such as concentration or viscosity.
section. Due to the latter effects, T1 ≫ T2∗ in the real NMR Typically, experiments intended to determine T1 perturb
setup. the system, let it relax, and then attempt to measure
the recovered longitudinal magnetization. On the other
There are two relaxation mechanisms which are of hand, experiments to find T2 rely on making observations
physical interest in this lab. The first is the eventual of the decay rates of the free induction signals.
recovery of longitudinal magnetization, that is, magneti-
zation along B ⃗ 0 , due to rethermalization of the system.
The second, which occurs even in the absence of the first, IV.5. Spin Echoes
is the loss of transverse magnetization due to decoherence
in precession phase of the spins. Both of these mecha- Although we are primarily interested in relaxation,
nisms contribute to the decay observed in the FID, and there are other effects that contribute to the decay in
the rates at which they occur depend on the substance the FID. The most important of these is inhomogene-
in question. ity in the magnetic field. A global inhomogeneity in the
The first mechanism is typically called spin-lattice re- ⃗ 0 can cause different parts of the
static magnetic field B
laxation and the time constant governing its rate is de-
sample to precess at different rates, leading to preces-
noted T1 . Its name derives from the fact that rethermal-
sion phase differences and a loss of the ensemble-averaged
ization is caused by the redistribution of energy from the
transverse magnetization much more quickly than would
spins to their surrounding environment (the ”lattice”).
be expected from just spin-spin interactions alone.
This has the effect of dissipating the energy of the pulse
In fact, for simple NMR setups such as the one used in
until the entire sample has returned to its original ther-
this lab where the static magnetic field is maintained by
mal state. If we use a 90◦ pulse to send the magnetization
permanent magnets, such field inhomogeneities dominate
into the transverse plane at t = 0, then the spin-lattice
relaxation. The observed decay constant of the FID is
relaxation process can be described by saying Mz recov-
typically denoted T2∗ , and it consists of two components:
ers according to
1/T2∗ = 1/T2 + γ∆B0 ,
 
Mz (t) = M0 1 − e−t/T1 , (20) (22)

where M0 is the magnitude of the longitudinal magneti- where T2 is the spin-spin relaxation time and ∆B0 is a
zation at thermal equilibrium. Of course, as Mz recovers, measure of the inhomogeneity of the magnetic field over
the transverse magnetization correspondingly decreases. the sample volume.
 gitinfo: amlevine 2017-10-26 — 5

90◦ RF Pulse 180◦ RF Pulse Spin Echo

Free Induction Decay (FID)

FIG. 2. An idealized scope trace for a spin-echo sequence, in a case where T2 ≫ T2∗ . The thick black lines indicate the
perturbing pulses used to implement the pulse sequence. Notice the spin echo is produced at time 2τ .

Fortunately, however, the effects of field inhomo- possible for spins in one region of uniform magnetic field
geneities may be reversed to some extent. Even after the to diffuse randomly to another. If this diffusion happens
FID has decayed away, it is possible to recover the trans- within the duration 2τ required to execute the spin echo
verse magnetization, up to whatever amount has been pulse sequence, then the precise dephasing process we
irreversibly lost in relaxation. This recovery was discov- described would no longer hold, and the spin echo ampli-
ered by Erwin Hahn in 1950 and is known as a spin echo tude would be reduced beyond just relaxation. In 1954,
([5, 6]). Carr and Purcell showed that when the effects of diffu-
The spin echo pulse sequence can be described as 90◦ - sion are considered, the echo amplitude produced goes
τ - 180◦ , i.e., a 90◦ pulse, after which the FID is allowed as
to decay away for time τ , at which point a 180◦ pulse 
2τ 2 2 2 3

is applied. A spin echo forms at a time τ after this last E(2τ ) = E0 exp − − γ G Dτ , (23)
T2 3
pulse, as shown in Figure 2.
To see how the spin echo is produced, consider a typical where E0 is the echo amplitude in the absence of both
sample which has small regions of uniform magnetic field, spin-spin and diffusion effects, i.e., the initial FID am-
but such that the field differs from one region to the next. plitude, see [7]. Here, G is the gradient of the inhomo-
Following a 90◦ pulse, spins in a region of relatively high geneous field and D is the diffusion constant. Thus, the
magnetic field precess relatively fast, while those in a effect worsens the longer we wait to produce a spin echo.
region of relatively low magnetic field precess relatively
slowly. By a time τ later, the phases of the magnetization
across different regions disagree sufficiently to degrade V. APPARATUS
the overall magnetization.
The spins within each individual region are still pre- V.1. Apparatus Overview
cessing coherently in the transverse plane. The appli-
cation of a 180◦ pulse has the effect of reflecting these In addition to a permanent magnet, the experimental
transverse spins about an axis in the transverse plane. apparatus consists of a gated radio frequency (RF) pulse
The spins continue to precess, but their relative motion generator that features variable pulse widths and spac-
is now precisely reversed. Thus, those regions which were ings, a probe circuit that delivers RF power to the sample
precessing faster and accumulated more phase difference and picks up the signal from the sample, a preamp that
now undo their phase accumulation at a faster rate. The amplifies the signal, and a phase detector that produces
result is that a time τ after the 180◦ pulse, all the regions an audio signal whose frequency corresponds to the dif-
are back in phase and the total magnetization reaches a ference between the Larmor frequency and the frequency
maximum, producing a spin echo. of the signal generator, see the
The use of a spin echo allows the experimenter to re-
duce the effects of phase dispersion on the transverse
magnetization. Looking at the height of the spin echo ef- V.2. Permanent Magnet
fectively gives the amplitude of the FID as it would have
been at time 2τ if field inhomogeneity were not present. This experiment uses a permanent magnet whose field
If T2∗ ≪ T2 , which is usually the case in setups like ours, has a strength of about 1770 gauss (0.177 Tesla). Care
this is information that would have been difficult to ob- should be taken to avoid bringing any magnetizable ma-
tain without the spin echo technique. terial, such as iron or steel, near the magnet as this may
Of course, there are limitations to the technique. It is be pulled in and damage the magnet.
 gitinfo: amlevine 2017-10-26 — 6

well as a considerable amount of leakage during pulses, is

presented to a sensitive preamp (Tron-Tech W110F) for
amplification. The signal then goes into a phase detector
(Mini-Circuits ZRPD-1) where it is mixed with the ref-
erence signal coming out of the other port of the power
splitter. The mixer forms the product of the two signals,
which is equivalent to a superposition of two signals, one
at the sum and one at the difference of the frequencies of
the resonance signal and the applied RF. The sum fre-
quency is typically much higher than the difference fre-
quency and is easily filtered out. When the NMR signal
is not precisely at the frequency of the transmitter, the
difference frequency will not be precisely zero. Since we
are looking at NMR signals in the vicinity of 1-8 MHz,
mixing them down to a low but non-zero frequency makes
it easier to see the structure of the signal.
FIG. 3. Block diagram of the experimental apparatus.

V.4. Digital Pulse Programmer

When performing the experiment, you should try to
find a region where the magnetic field is most uniform to
insert your sample and label the position of the probe for Most of the controls that you will manipulate are on
reproducibility from one experiment run to another. the digital pulse programmer, the oscilloscope, or the
function generator. The keypad of the Digital Pulse Pro-
grammer is shown in Figure 5. Press any of the four
buttons on the right to select a parameter (1st Pulse
V.3. RF Signal Chain
Width (PW1), 2nd Pulse Width (PW2), Tau (τ ), or Re-
peat Time). Then use the arrow buttons to set the corre-
The experiment-specific electronics, not including the sponding time for that parameter. The default times are:
probe head, is shown in Fig. 4. PW1 = 24 µs, PW2 = 48 µs, τ = 2 ms, and Repeat Time
Although it is the policy in Junior Lab to discour- = 100 ms. Note that when either the repeat time or τ is
age the use of pre-wired experiments, there are two long, the pulse programmer responds slowly as it needs
reasons why the present set-up should not be (lightly) to complete one cycle to change the settings. The top
changed. Several of the components, particularly the two buttons on the left determine whether a two-pulse
double-balanced mixers (DBM) and the low-level TRON- sequence occurs only once (the ”Single Pair of Pulses”
TECH pre-amplifier, are easily damaged if exposed to button), or repeats continuously (the ”Repeated Pairs
RF power levels that exceed their specified limits. Fur- of Pulses” button) with a pause between sequences of a
thermore, the lengths of some of the cables have been length set by the Repeat Time parameter. The third but-
specifically selected to fix the relative phase relationship ton, labeled ”Carr-Purcell”, will create a series of pulses
of different signals. corresponding to the Carr-Purcell technique described in
The RF pulse generating system is made up of a 15 the Measurements section. Finally, the fourth button,
MHz frequency synthesizer (Agilent 33120A), a digital ”Three Pulse”, yields 180◦ - τ - 90◦ - 180◦ pulses. For
pulse programmer based on a STAMP micro-controller, this pulse sequence, the 90◦ pulse time should be set us-
a double-balanced mixer used as an RF switch (Mini- ing PW1 and the 180◦ pulse time should be set using
Circuits ZAS-3), and an RF power amplifier capable of 2 PW2.
watts output. Set the delay τ to the minimum value and observe the
The frequency synthesizer feeds a sine wave of the de- amplified RF pulses from the port marked ”transmitter”
sired RF frequency to the power splitter. The power on channel 2 of the oscilloscope. The pulses should be
splitter keeps all impedances appropriately matched approximately 20 to 30 volts peak-to-peak (note that the
while feeding one half of the RF power to a double- settings on the function generator should be set to give
balanced mixer (DBM) used as a gate, or RF switch, for signals with peak-to-peak amplitudes of 2 3 volts, since
the RF. The other half is used as a reference signal in the the function generator output gets amplified). Choose
phase detector. The gate is opened and closed by TTL the slowest possible sweep speed; this will enable both
pulses provided by the digital pulse programmer. When pulses to be viewed simultaneously. A good starting pair
the gate is open, the RF pulses pass into a constant-gain of pulse widths might be 24 µs and 48 µs, corresponding
RF power amplifier. The amplified pulsed RF is fed into to approximately 90◦ and 180◦ . Now switch to channel 1,
the probe circuit. which displays the output of the phase detector (through
Any changes in the magnetic flux from the sample gen- the low-pass filter). Incidentally, there is another low-
erate a signal in the probe coil and circuit. That signal, as pass filter which is part of the scope itself. On the Tek-
 gitinfo: amlevine 2017-10-26 — 7

FIG. 4. View of the experiment support electronics. The magnet, probe circuit, function generator, oscilloscope, etc., are not
shown.

V.5. The Probe Circuit

Single 1st
Pair of Pulse
Pulses Width

Repeated 2nd
Pair of Pulse
Pulses Width

Carr
Purcell
τ

Three Repeat
Pulse N Time

FIG. 5. The Pulse Programmer Interface. FIG. 6. Schematic diagram of the probe head electrical circuit
and its connections to the RF source and to the RF detection
electronics.

The probe circuit is a tuned LC circuit with an

tronix analog scope there is a button marked ”BW limit impedance matched to 50 ohms at the resonant frequency
20 MHz”, which limits the allowed bandwidth. This but- for efficient power transmission to the sample; see Fig-
ton should be pressed in (active). On the HP digital ure 6. The inductor L in the circuit is the sample coil, a
scope the BW limit is set by one of the soft keys. On an coil of thin copper wire wound to accommodate a stan-
Agilent scope, this is set in the channel 1 or channel 2 dard 10-mm NMR sample tube. The coil is connected
menu. Set the y-sensitivity to about 10 mV/div at first. to ground through a tunable capacitor Ct , to allow fre-
Channel 1 will display the NMR signal. Place the glycer- quency and impedance matching. Power in and signal
ine vial in the probe and place the probe in the magnet. out pass through the same point on the resonant circuit,
Now the fun begins! so that both the power amplifier and the signal preamp
have properly matched loads. The connection between
Refer to Figure 2, which is a highly stylized version of the power amplifier and the sample contains a pair of
the signals you might obtain. The form of the voltage dis- crossed diodes in series with the probe circuit from the
played during the two applied RF bursts is unimportant. point of view of the power amplifier. By becoming non-
You will be focusing your attention on the FID signals conducting at low applied voltages, these serve to isolate
that appear after each burst, and on the echo. For five or the probe circuit and preamp from the power amplifier
ten microseconds after the RF pulse the amplifier is still between pulses, reducing the problems associated with
in the recovery phase, so this part of the signal should be power amplifier noise. The crossed diodes however, will
ignored. pass the high RF voltages that arrive when the trans-
 gitinfo: amlevine 2017-10-26 — 8

mitter is on. The signal out of the probe circuit passes the value of T2 . Look for interesting trends in T2
through a coaxial line to reach another pair of crossed across the sample set.
diodes at the input of the preamp. These diodes connect
the signal line to ground in order to short the preamp 4. Pick a set of samples on which to example spin-
end of the cable when the transmitter is on, causing that lattice relaxation times. For each sample in the
end of the cable to act like a short circuit. This helps set, use an appropriate NMR pulse sequence to de-
to protect the delicate preamp from the high RF power termine the value of T1 . Look for interesting trends
put out by the power amplifier. The line from the probe in T1 across the sample set.
to the preamp input is, ideally, a quarter-wavelength in
The space of possible samples that are amenable to
length at the Larmor frequency (note that the actual ca-
NMR analysis is obviously enormous, but there are sev-
ble in use in the Junior Lab apparatus may not have that
eral samples that have traditionally been used in this
length). Any quarter-wave transmission line transforms
lab and that are available for your use. Except for the
impedance according to the following relation:
fluorine-based samples, which are used primarily to mea-
Z02 sure the magnetic moment of the fluorine nucleus, most
Zin = (24) of these samples are based on the 1 H nucleus. The sam-
Zout ples used in this lab include:
where Z0 is the characteristic impedance of the line. • Glycerin-water mixtures: Various mixtures of glyc-
Therefore during the RF pulse, the preamp circuit at erin with water, with proportions given in percent-
the far end of the quarter-wave line looks like an open ages by weight. Spin-spin interactions generally in-
circuit to the probe and does not load it down. Between crease with liquid viscosity. Thus, measurements
pulses, the voltage across the diodes is too small to turn of T2 as a function of glycerin to water ratio are of
them on, and they act like an open circuit, allowing the particular interest.
small NMR signal to pass undiminished to the preamp.
• Paramagnetic ion solutions: Two ten-fold serial di-
lutions of 0.830M and 0.166 M starting solutions
VI. MEASUREMENTS of Fe3+ ions. The presence of paramagnetic ions
greatly facilitates the dissipation of energy from the
VI.1. Measurements Overview spins to their surroundings. Thus, measurements of
T1 as a function of ion concentration are of partic-
The nature of this experiment allows for considerable ular interest.
variety in the investigations which may be performed. • Fluorine samples: There are samples of both triflu-
We present in the following sections several well-known oric acetic acid and hexafluorobenzene at the setup.
techniques in pulsed NMR used to determine relaxation The former is a strong acid and should be treated
times, nuclear magnetic moments, and so on. These mea- with care.
surement techniques may be applied to various sets of
samples in order to construct measurement sets that meet • Water: There are a number of potentially interest-
experimental objectives. Generally, in the Junior Lab a ing but somewhat difficult measurements that can
basic set of measurements involves the following proce- be done with water. These are discussed later for
dures: those interested.
1. Using appropriate samples, determine the Larmor
frequencies for 1 H and 19 F by dialing the function VI.2. Suggested Progress Check
generator frequency and looking for resonance. Use
a Hall effect magnetometer to measure the mag-
The optimal schedule for this lab is highly dependent
netic field in the sample coil. Keep in mind that it
on what measurements are planned. It is a good idea
may be necessary to retune the probehead circuit
to start by utilizing one technique to measure both T1
and redo the search for signal when moving from
1 and T2 on sets of samples (say, glycerin for T2 and para-
H to 19 F or vice versa.
magnetic ions for T1 ) in order to look for trends. Once
2. Determine t90 , the pulse width that rotates the z- those measurements are complete, additional measure-
magnetization into the transverse plane. Note that ments can then be made on other sample sets or with
this value can change each time the setup is altered, other pulse sequences (or even variations of the pulse se-
so it should be reassessed from session to session or quences).
within each session whenever the setup is altered. The first lab session or two should be used to famil-
iarize yourself with the equipment and to determine the
3. Pick a set of samples on which to examine spin-spin magnetic moments of the hydrogen and fluorine nuclei.
relaxation times. For each sample in the set, use The measurements on each sample set generally take one
an appropriate NMR pulse sequence to determine to two lab sessions, so approximately three additional
 gitinfo: amlevine 2017-10-26 — 9

lab sessions should be dedicated to performing relaxation VI.4. Finding Pulse Widths
measurements.
Delays sometimes occur when signal is lost due to To obtain an FID signal, it is unnecessary to use the
equipment changes or subtle changes in oscilloscope set- exact value of t90 when pulsing the sample. When first
tings. In such cases, after obvious debugging has been searching for a signal or finding the Larmor frequency,
done, it is best to obtain the help of lab staff rather than almost any reasonable initial guess for PW1 will result in
to spend too much time tracking down a problem. an observable FID signal (since α ̸= 0). However, when
utilizing an established NMR pulse sequence to measure
a relaxation time, it is important to use accurate values
VI.3. Finding Larmor Frequencies of t90 and t180 .
As we mentioned before, it is difficult to know the
The signal seen at the oscilloscope is the FID signal magnitude of the perturbing field B1 , so the formula
from the sample at frequency γB0 mixed with a constant t90 = π/γB1 is not very helpful. However, we do know
amplitude signal of frequency ω from the function gen- the amplitude of the FID should be maximum after a 90◦
erator. This produces a beat signal which has the same pulse and minimum or zero after a 180◦ pulse. Thus, one
decay envelope as the FID but which has a comparatively easy way to obtain the pulse widths is to find the setting
low frequency |γB0 − ω| that can easily be displayed on of PW1 which minimizes the FID. This gives t180 while
an oscilloscope. half of the value gives t90 .
It follows that in order to determine the Larmor fre- Another method is to use spin echoes. Set PW1 to
quency, we need to tune the function generator frequency some initial value and set PW2 to be exactly twice PW1.
ω until the beat frequency vanishes: Then adjust PW1, while keeping PW2 twice PW1, until
a spin echo is visible and of maximum amplitude. Then
γB0 − ω = 0. (25) PW1 gives t90 while PW2 gives t180 . Obviously, there
are many other ways in which the pulse widths can be
The value of ω read off from the function generator is obtained.
then a measurement of the Larmor frequency of the stim- Experiment with these or other techniques in order to
ulated nuclei in the sample. On the scope, the approach estimate the 90◦ and 180◦ pulse widths as accurately as
to resonance should look like a decaying sinusoid as its possible. However, it is important also to realize that the
frequency goes to zero (or its period goes to infinity). pulse programmer is only capable of setting PW1 and
When the resonance condition is precisely fulfilled, the PW2 to integer multiples of 1 µs. Thus, it is only prof-
result is simply an exponential decay. itable to narrow down the pulse widths to within one or
Another way to think about this measurement is to two microseconds. Generally, pulse sequences and mea-
consider the mixing as a ”stroboscopic” view, with strobe surements of relaxation times work well even if the pulse
frequency ω, of the precessing magnetization which has widths are slightly off. When in doubt, it may be helpful
natural frequency γB0 . When the strobe frequency to measure the effects of changes in PW1 and PW2 on the
matches the natural frequency, the magnetization vector spin echoes or other products of specific pulse sequences.
appears to stay stationary in the transverse plane, and Note that the best values for t90 and t180 may change
it simply appears to decay away with the time constant from one session to the next, becaause of changes in the
T2∗ . tuning of the probehead circuit, the power output of the
In order to measure the magnetic moment of a type function generator, the exact placement of the sample in
of nucleus with a known spin, it is necessary to measure the magnetic field, and so on. Thus, it is a good idea to
both the Larmor frequency ωL = γB0 and the magnetic quickly check the pulse widths at the beginning of every
field strength B0 seen by the sample. Measurement of session or within a session whenever relevant changes are
the latter is accomplished using a Hall effect magnetome- made to the apparatus or instrument settings.
ter, which is shared by several other experiments and so It is evident from our discussion of the spin echo se-
is not a part of the permanent setup. Make certain to quence that it can be used to make measurements of spin-
zero and calibrate the magnetometer before use and make spin relaxation and hence T2 . In this context, we call the
measurements perpendicular to the magnetic field lines. spin echo sequence the 90◦ − 180◦ pulse sequence (the
Estimate the variations in the magnetic field over the delay time τ between the pulses is assumed). Measuring
sample. the degradation of the spin echo as a function of τ reveals
Once the resonance frequency has been found, it is the effect of spin-spin relaxation, as if the FID were not
useful to return the oscillator frequency back to being affected by field inhomogeneities.
slightly off resonance. Being able to observe a beat signal This sequence can be configured by setting PW1 equal
carried by the exponential decay envelope makes identi- to t90 and PW2 equal to t180 , using the two-pulse-
fying and assessing pulse sequences easier. As long as repeated program on the pulse programmer. The spin
ω ≈ γB0 , it is still possible to manipulate the orienta- echo is produced a time 2τ after the initial 90◦ pulse.
tion of the magnetization. Adjust ω around resonance to An interesting scope technique applicable to this and
obtain a satisfactory FID signal. other pulse sequences is the use of infinite persist. With
 gitinfo: amlevine 2017-10-26 — 10

suitable trigger settings, it is possible to see the spin echo The idea of the 90◦ - 90◦ pulse sequence is to first rotate
moving towards the right end of the scope’s screen as the thermalized magnetization, i.e., magnetization in the
we increase τ on the pulse programmer. As the spin z direction, into the transverse plane, and then wait for
echo moves, its amplitude exponentially decays, and the some delay time τ , during which some of the longitudi-
result of viewing the whole process under infinite persist nal magnetization will recover via spin-lattice relaxation.
is a ”decay envelope” traced out by the peak of the spin The second 90◦ pulse then rotates this recovered mag-
echo. It is up to each group to decide whether this is an netization into the transverse plane. If, during the ap-
approach useful for making measurements; in any case, plication of the second pulse, the component of the net
it provides a good visualization. magnetization parallel to x̂′ has a negligible value, then
the amplitude of the following FID indicates the strength
of the recovered longitudinal magnetization.
VI.5. The Carr-Purcell Sequence The 180◦ - 90◦ pulse sequence, a variation of the 90◦ -
90◦ sequence, may also be used to measure T1 . It consists
As mentioned before, the loss of spin echo amplitude of a 180◦ pulse, followed by a delay of time τ , followed by
after a 90◦ - 180◦ pulse sequence is due not only to spin- a 90◦ pulse. The amplitude of the ensuing FID is mea-
spin relaxation, but also to the effects of diffusion. The sured. This pulse sequence can be configured by setting
effects of diffusion are particularly pronounced for sam- PW1 to t180 and PW2 to t90 , and using the two-pulse-
ples with large T2 , since such samples require a corre- repeated program on the pulse programmer.
spondingly large τ when using the 90◦ - 180◦ method. In the 180◦ - 90◦ pulse sequence, the first pulse serves
Diffusion effects scale as τ 3 in the exponent, so at large τ to flip, i.e., invert, the thermalized longitudinal (z) mag-
they dominate the effects of spin-spin relaxation (which netization. As relaxation then proceeds, the magnetiza-
go as τ ), and therefore must be taken into account when tion first moves back towards zero after which it increases
estimating T2 . until equilibrium is reestablished. This pulse sequence
This problem was addressed in 1954 by Carr and Pur- is sometimes called an ”inversion recovery sequence” for
cell, who introduced a sequence, now called the Carr- this reason. After allowing this process to occur for a
Purcell sequence, which is much less susceptible to dif- time τ , a 90◦ pulse is applied, bringing the recovered
fusion effects [7]. Rather than repeating a pulse se- magnetization into the transverse plane, where it gener-
quence with ever-increasing delay times, the Carr-Purcell ates an FID.
method uses a fixed τ , which can be chosen small enough Unlike in the 90◦ - 90◦ sequence, the magnetization in
to reduce the effects of diffusion to negligible levels. this case actually reverses, going through zero at time
The Carr-Purcell method prescribes the following se- T1 ln 2. However, the amplitude of the FID is insensitive
quence: to the sign of the magnetization prior to the 90◦ pulse,
which would show up merely as a phase shift in the si-
90◦ − τ − 180◦ − 2τ − 180◦ − 2τ − 180◦ − 2τ − · · · , nusoid, so the FID amplitude as seen through the scope
will appear to shrink with small enough τ , go through
continuing for as long as the repeat time allotted. The
zero, and then exponentially recover at large τ .
first two pulses are exactly the 90◦ − 180◦ sequence.
Therefore a spin echo is produced at time τ after the first
180◦ pulse. At a time of 2τ after the first 180◦ pulse,
when that first spin echo has subsided, a 180◦ pulse is VI.6. The Three-Pulse Sequence
again applied to the sample, causing yet another spin
echo to appear, and so on. The 90◦ - 90◦ and 180◦ - 90◦ pulse sequences have the
Thus, the Carr-Purcell method produces a train of spin disadvantage that they require reading the amplitude of
echoes. With each iteration, the spin echo amplitude de- the FID which occurs immediately after an RF pulse.
cays away due to spin-spin relaxation, but, if τ is chosen This problem was addressed several years ago by two Ju-
small enough, neither field inhomogeneities nor diffusion nior Lab students, who proposed a variation, the ”three-
effects play a role in that decay. pulse” sequence. (These students, Rahul Sarpeshkar and
The Carr-Purcell sequence can be configured by setting Isaac Chuang, are now MIT professors.)
PW1 equal to t90 and PW2 equal to t180 , and by using The three-pulse sequence consists of a usual inversion
the Carr-Purcell program on the pulse programmer. The recovery sequence. The first 180◦ pulse inverts the ther-
repeat time effectively determines how many iterations malized z-magnetization, and after a delay of τ , the sec-
are applied before the sequence repeats. ond 90◦ pulse rotates the recovered z-magnetization into
The 90◦ - 90◦ pulse sequence is the simplest pulse se- the transverse plane, where it generates an FID. Instead
quence used to measure spin-lattice or T1 relaxation. It of measuring this FID, this sequence waits an additional
consists of a 90◦ pulse, followed by a delay of time τ non-variable time ϵ before applying another 180◦ pulse.
and another 90◦ pulse. The amplitude of the second FID This last pulse yields a spin echo, whose amplitude re-
is then measured. It can be configured by setting PW1 flects the amplitude of the FID, but since the spin echo
and PW2 both equal to t90 , using the two-pulse-repeated is separated in time from the RF burst, its amplitude is
program on the pulse programmer. much easier to measure. The amplitude of the spin echo
 gitinfo: amlevine 2017-10-26 — 11

after the third pulse as a function of τ thus follows the tilled water and see if there is any difference. One way
same trend as in the 180◦ - 90◦ sequence. this could be done is by bubbling pure nitrogen through
The three-pulse sequence can be configured by setting the water, as N2 is diamagnetic.
PW1 to t90 and PW2 to t180 , and then using the three-
pulse program on the pulse programmer. The second
time delay ϵ cannot be set manually, and has been pro- VII. ANALYSIS
grammed to be about 1 ms. The time ϵ is kept small to
minimize spin-spin effects that might occur. Due to the wide range of possible measurements in this
experiment, there is a corresponding variety of possibil-
ities for the particular analysis approach that best suits
VI.7. Relaxation in Water your data set. However, some results that are often pre-
sented include the following:
In describing the various pulse sequences, we assumed
for the most part that the two relaxation mechanisms • The Larmor frequencies and gyromagnetic ratios of
are not simultaneously important. For example, when the proton and 19 F nucleus.
measuring T2 , we assume that a negligible amount of
the spin echo decay is due to actual recovery of the z- • Demonstration of the successful use of a pulse se-
magnetization via spin-lattice relaxation. For the most quence to measure the spin-lattice relaxation time
part, this assumption is valid when working with samples T1 across a range of samples (e.g., paramagnetic
in which one relaxation constant is drastically smaller ion solutions).
than the other (e.g., T2 ≪ T1 in viscous liquids and T1 ≪
T2 in paramagnetic solutions). • Demonstration of the successful use of a pulse se-
Of course, whether this assumption is valid for any par- quence to measure the spin-spin relaxation time
ticular sample is something which deserves consideration T2 across a range of samples (e.g., diluted glycerin
in analyzing your results. In fact, this assumption is not mixtures).
valid for many samples used in NMR, where T1 and T2
are usually comparable. An important example is water, The idea is to structure the measurement sets and anal-
where both of the relaxation constants are on the order ysis so that you can make a case for the effectiveness of
of several seconds, making measurements of both quite NMR as a way of probing the microscopic structure of
difficult. your samples, by demonstrating measurements that are
There is, however, at least one interesting way to mea- sensitive to changes in the material structure. You may
sure T1 in water which deserves mention. As discussed find that some samples are easier to work with than oth-
previously, if we want each pulse sequence to yield an in- ers. Time constraints can also dictate which measure-
dependent measurement, we must set the repeat time on ment sets to use and what analysis approach to take.
the pulse programmer sufficiently large to allow rether- One interesting idea often pursued is to reproduce the
malization. For example, executing two spin echo se- early results by Bloembergen and Purcell when working
quences too close together will make the second spin echo with water-glycerin mixtures, such as those in [2] and
appear smaller. This suggests that we can actually take Bloembergen’s thesis [8]. They found that plotting the
advantage of this fact to make a measurement of T1 us- relaxation constants against logarithm of viscosity (or
ing spin echoes and by varying not the delay time but concentration) resulted in interesting curves as the vis-
the repeat time. cocity (concentration) varied over a large range. Tables
Use the standard spin-echo sequence (90◦ - 180◦ ) with relating percent weight of glycerin to viscosity can be
some fixed time constant τ . Record the spin echo height found in published tables.
produced when repeating the sequence using a variable Remember as always to address sources of error or un-
repeat time; the spin echo height as a function of the certainty in your results, quantitatively whenever possi-
repeat time should give the exponential return of the ble. Each pulse sequence is susceptible to different sys-
magnetization with time constant T1 . For repeat times tematic effects, e.g., diffusion in the 90◦ - 180◦ sequence,
higher than 3 s, you can use the two-pulse-single program so interpretation of the results needs to take into account
on the pulse programmer and a watch, rather than the an understanding of the physics behind the technique. In
two-pulse-repeated program. As an optional experiment, some ways, this is one of the central ideas behind NMR
perform these measurements with both tap water and spectroscopy.
distilled water. Is it possible to detect a difference?
The first measurements of T1 in distilled water stood
for about thirty years. Since then, careful measurements VIII. SUPPLEMENTAL QUESTIONS
have produced a number which is about 50% higher. The
difference is due to the effect of dissolved oxygen in the Each of the magnetic moments in a sample is influ-
water, as O2 is paramagnetic. As an optional experiment, enced by the magnetic fields of other moments in its
try to remove the dissolved oxygen from a sample of dis- neighborhood. These differ from location to location in
 gitinfo: amlevine 2017-10-26 — 12

the sample, depending on the relative distance and ori- ing, e.g., a 90◦ pulse, decreases as the preces-
entation of neighbor moments to one another. An ap- sion phases of, e.g., the protons at different loca-
proximate measure of the magnetic field variation expe- tions and, thus, with different precession frequen-
rienced by the proton moments in the water molecule is cies spread out. Estimate the transverse relaxation
the range corresponding to parallel alignment of two in- time T2 for the water sample by finding the time
teracting protons at one extreme to opposite alignment required for two moments, differing by the average
at the other. ∆ω calculated above, to move from in-phase to π-
out-of-phase positions.
1. Using µ/r3 for the field of the neighbor moment,
show that the half-range in the Larmor precession As you’ve probably guessed, this lab is merely a step-
frequencies is given by ∆ω ≈ (gµn )2 /hr3 . ping off point for an incredibly varied set of potential
investigations. Some good general references for this lab,
2. The transverse magnetization of a sample follow- beyond the ones already cited, are [9–15].

[1] F. Bloch, Phys. Rev. 70, 460 (1946). a two-element column matrix called a spinor :
[2] N. Bloembergen, E. Purcell, and R. Pound, Phys. Rev.  
73, 679 (1948). u
|ψ⟩ = u|0⟩ + d|1⟩ = . (A1)
[3] Nobel Lecture for Felix Bloch and Edward Mills Purcell d
(1952).
[4] A. Abragam, Principles of Nuclear Magnetism (Oxford Normalization imposes the constraint |u|2 + |d|2 = 1.
University Press, 1961) physics Department Reading The system is governed by the Schrödinger equation
Room.
[5] E. Hahn, Phys.Rev. 80, 580 (1950). d
[6] E. Hahn, Physics Today Nov. 1953, 4 (1953), a some- iℏ |ψ⟩ = H|ψ⟩ (A2)
dt
what simplified version of Hahn’s 1950 paper.
[7] H. Carr and E. Purcell, Phys. Rev. 94, 630 (1954). that has the solution |ψ(t)⟩ = U |ψ(0)⟩, where U =
[8] N. Bloembergen, Nuclear Magnetic Relaxation (W.A. e−iHt/ℏ is unitary. In pulsed NMR, the Hamiltonian
Benjamin, 1961) physics Department Reading Room.
[9] G. Pake, American Journal of Physics 18, 438 (1950). H = −⃗ ⃗ = −µ [σx Bx + σy By + σz Bz ]
µ·B (A3)
[10] T. Farrar and E. Becker, Pulse and Fourier Transform
NMR: Introduction to Theory and Methods (Acad. Press, represents the potential energy of a magnetic moment
1971) physics Department Reading Room. placed in an external magnetic field. The σi ’s are the
[11] Feynman, Leighton, and Sands, Lectures on Physics, Vol.
Pauli spin matrices:
Volume II, Chapter 35 (Addison-Wesley, 1965) interest-
ing discussions of angular momentum, the Stern-Gerlach 
0 1

Experiment and NMR, Physics Department Reading σx ≡ ,
1 0
Room.  
[12] E. Fukushima and S. Roeder, Experimental Pulse NMR 0 −i
(Addison-Wesley, 1981) an excellent practical reference, σy ≡ ,
i 0
Science Library Stacks.  
[13] R. Freeman, A Handbook of Nuclear Magnetic Reso- 1 0
σz ≡ . (A4)
nance, 2nd ed. (Farragut Press, 1997) spin-Lattice Re- 0 −1
laxation, Science Library Stacks.
[14] G. Pake, Annual Review of Nuclear Science 19, 33 (1954). Upon insertion of equations (A4), (A1), and (A3) into
[15] G. Pake, Radio frequency and Microwave Spectroscopy of equation (A2), we get
Nuclei (W.A. Benjamin, 1961).
u̇ = (µ/ℏ) [iBx + By ] d + i(µ/ℏ)Bz u,
d˙ = (µ/ℏ) [iBx − By ] u − i(µ/ℏ)Bz d. (A5)

Appendix A: Quantum Mechanical Description of If Bx = By = 0, the equations reduce to

NMR
u̇ = i(µ/ℏ)Bz u,

Recall that for a spin-1/2 particle such as a pro- d˙ = −i(µ/ℏ)Bz d. (A6)

ton, neutron, electron, quark, or leptons, there are just Integration with respect to time yields
two spin eigenstates that may be chosen to be spin up
|S, Sz ⟩ = | 21 , 12 ⟩ ≡ |0⟩ and spin down |S, Sz ⟩ = | 12 , − 12 ⟩ ≡ u = u0 ei(µ/ℏ)Bz t = u0 eiω0 t/2 ,
|1⟩. When these states are used as basis vectors, the gen-
eral spin state of a spin-1/2 particle can be expressed as d = d0 e−i(µ/ℏ)Bz t = d0 e−iω0 t/2 , (A7)
 gitinfo: amlevine 2017-10-26 — 13

where ω0 = 2µBz /ℏ is the Larmor precession frequency. called the Bloch sphere representation, and the vector
If an atom undergoes a spin-flip transition from the spin (cos ϕ sin θ, sin ϕ sin θ, cos θ) is called the Bloch vector.
up state to the spin down state, the emitted photon has The Pauli matrices give rise to three useful classes of
energy E = ℏω0 . unitary matrices when they are exponentiated, the rota-
If a small external magnetic field Bx is imposed while tion operators about the x̂, ŷ, and ẑ axes, defined by the
keeping By = 0, and if Bx ≪ Bz , then equations (A5) equations:
become
θ θ
Rx (θ) ≡ e−iθσx /2 = cos I − i sin σx
u̇ = i(µ/ℏ)Bx d + i(µ/ℏ)Bz u, 2  2
cos θ2 −i sin θ2

d˙ = i(µ/ℏ)Bx u − i(µ/ℏ)Bz d. (A8) = , (B1)
−i sin θ2 cos θ2
For a time varying magnetic field of the type produced θ θ
by an RF burst as Ry (θ) ≡ e−iθσy /2 = cos I − i sin σy

in pulsed NMR, Bx = 2Bx0 cos ωt = 2 2
Bx0 eiωt + e−iωt . Define ωx ≡ 2µBx0 /ℏ. We see that
cos θ2 − sin θ2
 
= , (B2)
u̇ = i(ω0 /2)u + i(ωx /2) eiωt + e−iωt d,

sin θ2 cos θ2
θ θ
d˙ = −i(ω0 /2)d + i(ωx /2) eiωt + e−iωt u. Rz (θ) ≡ e−iθσz /2 = cos I − i sin σz


(A9)
 −iθ/2 2 2
Using ωx ≪ ω0 , since Bx ≪ B0 , we can try for a e 0
= . (B3)
solution of the form 0 eiθ/2

u = Cu (t)eiω0 t/2 , One reason why the Rn̂ (θ) operators are referred to as
−iω0 t/2 rotation operators is the following fact. Suppose a spin-
d = Cd (t)e . (A10)
1/2 particle has a state represented by the Bloch vector
Inserting equations (A10) into the differential equations ⃗λ. Then the effect of the rotation Rn̂ (θ) on the state is
(A9) for u and d, we get to rotate it by an angle θ about the n̂ axis of the Bloch
sphere.
iωx h i
An arbitrary unitary operator operating on a single
C˙u = Cd ei(ω−ω0 )t + e−i(ω+ω0 )t ,
2 spin-1/2 spinor can be written in many ways as a combi-
˙ iωx h i nation of rotations together with global phase shifts. A
Cd = Cu ei(ω+ω0 )t + e−i(ω−ω0 )t . (A11) useful theorem to remember is the following: Suppose U
2
is such a unitary operator. Then there exist real numbers
If the system is run near resonance (ω ≈ ω0 ), the terms α, β, γ and δ such that
e±i(ω+ω0 )t may be neglected. Then equations (A11) be-
come U = eiα Rx (β)Ry (γ)Rx (δ). (B4)
iωx
C˙u = Cd ,
2
iωx
C˙d = Cu . (A12)
2
Taking the derivatives of these equations, we see that Cu
and Cd act like harmonic oscillators of frequency ωx /2.
These have the general solution

Cu = a cos (ωx t/2) + b sin (ωx t/2),

Cd = ia sin (ωx t/2) − ib cos (ωx t/2). (A13)

Putting these in Equation (A10), we get the solution for

u and d. These are called Rabi oscillations, valid for
ωx ≪ ω0 .

Appendix B: Bloch Sphere Representation

A spin-1/2 state u|0⟩ + d|1⟩ can be visualized as a

point (θ, ϕ) on the unit sphere, where u = cos(θ/2),
d = eiϕ sin(θ/2), and u can be taken to be real because
the overall phase of the state is unobservable. This is