You are on page 1of 43

Nuclear Magnetic Resonance of Protons

Introduction

The principle of nuclear magnetic resonance (NMR) has developed into an approach
widely used to non-invasively perform analyses in materials science, solid-state physics,
biophysics, medicine, structural biology (complementing X-ray diffraction), analytical chemistry
(including for structural studies of organic molecules), etc. The principle can be used for
imaging, as for instance in magnetic resonance imaging (MRI) in medicine and neuroscience.
NMR has developed and branched into many related approaches, and keeps being an interesting
research subject. For instance, approaches from NMR are presently used to manipulate spins for
the purpose of using spin as the physical basis for quantum computers.

In this experiment you use an NMR spectrometer study magnetic resonance, free-
induction-decay, spin rotation, and spin relaxation times. These topics lie at the basis of why
NMR is such a powerful analytical technique for any problem that can be tackled using
knowledge of nuclear spins. These topics also relate to the quantum mechanics of spin.

The Nobel Prize in Physics 1952 was awarded to Felix Bloch and Edward Mills
Purcell for NMR, with as citation: “for their development of new methods for nuclear
magnetic precision measurements and discoveries in connection therewith.” This was a few
years after they first reported NMR in materials (1946, Bloch at Stanford, and Purcell, Torrey,
and Pound at Harvard). Their work was an offshoot of research on radar technology and
physics in WW2. It was quickly realized that NMR was important. A boost also came from
the early realization by Russell Varian that Varian Associates should produce commercial
NMR systems based on newly-developed homogeneous electromagnet technology (Russell
and Sigurd Varian founded Varian Associates as an early high-tech company in Silicon
Valley). From those early days, as mentioned, NMR has become an ever-evolving
indispensable tool and keeps spurring innovation in many branches of science.

1
BACKGROUND
For electrons in atomic orbitals we distinguish between electron spin and electron orbital
angular momentum, which combine to give the total angular momentum. Nuclei are composites
of protons and neutrons (both fermions), but concerning angular momentum properties the
composite often behaves as a single unit with intrinsic angular momentum, represented by I and
referred to as "nuclear spin". Associated with each nuclear spin is a nuclear magnetic moment, μ,
which produces magnetic interactions with its environment and also allows Nuclear Magnetic
Resonance Spectroscopy (NMR) to be performed on the nucleus. The ratio of the magnetic
moment to angular momentum of a particle (here nuclear spin) is called the gyromagnetic ratio,
γ, as in μ = γħI. Individual protons and neutrons have spin 1/2. A nucleus of odd nucleon number
A will have a half-integer spin and a nucleus of even A will have integer spin. Nuclei with even
numbers of neutrons (N) and even numbers of protons (Z) have I = 0 and hence μ = 0. If a
nucleus has μ = 0, then it is not NMR-active. For the nuclear spin to be observed by NMR, I can
be any non-zero integer or half-integer, e.g. 1/2, 1, 3/2, 2, etc. It turns out the terrestrially
common nucleus 12C, has I = 0 and is not NMR-active. But another common nucleus, 1H, a
proton, is NMR-active, with I = 1/2. The equipment you will use is tuned to protons, as are many
NMR spectrometers. It turns out that γ as defined above has units of frequency / magnetic field,
so Hz/T. This is handy for proton NMR, because the resonant frequency you observe at a given
magnetic field can simply be read off from γ. For protons, γ = 42.5775 MHz/T. Our magnet
generates a magnetic field of about 0.3640 T. So a good starting point, as you will see later, is to
explore frequencies around 15.50 MHz.

What happens when spins are placed in a magnetic field?


The Classical Picture. When placed in an external magnetic field (B0), the magnetic moment
tries to align with it like a compass needle in the earth’s magnetic field (Fig. 1a). Because the
nucleus has angular momentum, the torque τ exerted by the magnetic field results in precession
(like a spinning top in the earth’s gravitational field, Fig. 1b).

Fig. 1a: Left, magnetic moments in an iron bar


magnet randomly oriented. Middle; in a magnetized Fig. 1b: Classical picture:
bar, magnetic moments are aligned. Right; an precession of a spin in an
external magnetic field creates a torque on the bar external magnetic field.
magnet causing it to align along the field, τ = µ x B

2
The rate of this precession is proportional to the external magnetic field strength and is
characteristic of each nuclear spin. In NMR, the precession frequency is in the radio-frequency
(RF) range. The precession frequency is called the Larmor frequency.

The Quantum Model (see also Appendix 1). The classical view where the nucleus is treated as
a macroscopic object is not sufficient. Let us consider the proton nucleus, which has spin 1/2.
For a spin 1/2 nucleus, there are two quantum states which can be visualized as having the spin
axis pointing “up” or “down” (see Figs. 2a and 2b). In the absence of an external magnetic field,
these two states have the same energy and at thermal equilibrium exactly one-half of a large
population of nuclei will be in the “up” state and one-half will be in the “down” state. In a
magnetic field, however, the “up” state, which is aligned with the magnetic field, is lower in
energy that the “down” state, which is opposite to the magnetic field. Because this is a quantum
phenomenon for spin 1/2, there are no possible states in between. The energy separation, ∆E,
between the two quantum states is proportional to the strength of the external magnetic field B0,
and increases as the field strength increases. In a large population of nuclei in thermal
equilibrium, slightly more than half will reside in the “up” (lower energy) state, and slightly less
than half will reside in the “down” (higher energy) state. The state “up” is lower energy because
the nuclear spin direction is parallel to the direction of B0. The “down” state is higher in energy
because the nuclear spin direction is antiparallel to B0. As in any spectroscopy, if sufficient
energy is provided to the lower energy state, then transition to the next and higher energy state
occurs. If the energy of an RF pulse (more about this later) delivered in NMR experiment is the
same as the energy difference between the two levels, then nuclei in the lower state absorb
energy and transition to the higher state. In pulsed NMR, you apply the RF pulse and then wait
and observe. When the pulse is off, the nuclei in the higher energy state will tend to transition to
the lower state, and precess at the same time, giving rise to the NMR signal.

Fig 2a Fig 2b

Fig. 2a and b: Quantum picture; energy states created by an external magnetic, and
transitions between these states, lower to higher, by absorption of electromagnetic
radiation (RF pulse), and emission of energy transitioning back to the lower energy state.
At resonance, the RF pulse and the detected frequency are at the Larmor frequency.

3
Let us now do a virtual NMR experiment. Consider a specific collection of magnetic
nuclei, e.g. the protons in a liquid mineral oil sample. In the absence of a magnetic field, the spin
axes of the protons are randomly oriented and there is no net magnetization, M, of the sample.
We will begin our NMR experiment by placing our sample into a magnetic field, B0, the field of
the electromagnet (Fig. 3a). When the sample is just placed in the magnet, there is no
instantaneous net magnetization. In time a net magnetization develops which is collinear with
the direction of the B0 field. This is the thermal-equilibrium condition, and this initial
equilibrium magnetization is called M0. Figure 3b shows the various vectors discussed.

The sample of mineral oil is in the magnet and thermal equilibrium is attained. Figure 3a
shows a side view sketch of the basic system, without the sample. The sample sits in the sample
coil, a tightly wound solenoid (inside the equipment) which has its axis perpendicular to the
constant magnetic field, B0, of the permanent magnet. This coil serves two purposes. In the role
of a transmitter coil, it supplies a rotating magnetic field which can change the orientation of
the net magnetization in the sample. In the role of receiver coil, it picks up (by induction) the
magnetization due to the collection of spins in the sample of interest.

B0 

Fig. 3a: Schematic of NMR set-up. Fig. 3b: The vectors B0 and M0 and B1. The
The sample is not shown, and in RF field B1 is perpendicular to B0, and at
experiments, the sample is inserted resonance the rotation frequency of B1 equals
into the sample coil. the Larmor frequency of the specific nucleus.

Figure 3b shows the relative directions of the magnetic fields we will need: 1) B0 (along z),
the magnetic field created by the magnet as discussed above (in our equipment, a permanent
magnet), and 2) B1, a field which is perpendicular to B0 and which effectively rotates in the x-y
plane, so rotates around B0. B1 will be generated by a sinusoidal oscillatory RF signal applied to
the sample coil. The RF signal will only be applied for a short duration (~µs), hence we call it an
RF pulse (a burst of oscillations with an amplitude envelope that looks like a square pulse in the

4
time domain: the amplitude of the sinusoidal when plotted vs time looks like a square pulse).
The frequency of the sinusoidal will be dialed in on the RF generator, and should be set to
resonance, namely equal to the Larmor spin precession frequency appropriate for B0. In Fig. 3b,
the cylinder represents the sample coil. The thick arrows represent B0, created by the permanent
magnet. The thermal equilibrium magnetization of the nuclear spins is represented by the thinner
arrow labeled M0. The magnetization along the z-axis is called Mz (not shown here, but appears
in later figures). The dashed double ended arrow represents B1, and it shows the direction of the
oscillating RF field (look at this arrow as the projection along y of B1, given that B1 rotates in the
x-y plane).

It turns out that the resonant RF pulse “tips the spins” away from the B0 direction (see Fig. 4
i). The longer the duration of the RF pulse, the larger the tipping angle. Why this is happening
is explained in Appendix 1, quantum-mechanically and also referring to a classical analog.
After such an RF pulse, the net magnetization vector M (due to all the spins) is no longer aligned
along z but forms an angle with z, and hence this M now precesses freely around B0 (see Fig. 1b
and Fig. 4 ii). The resonant frequency (needed for the RF pulse) is characteristic of each
nucleus; for example for proton in a 1 T B0 field it is 42.5781 MHz. As shown in Appendix 1,
this resonant frequency effectively allows for the transitions between the energy levels shown in
Fig. 2b. Once the spins have been tipped away from z, M develops a component in the x-y plane.
This component is referred to as the transverse magnetization. This transverse magnetization
vector whirls around in the x-y plane, at the precession frequency. The sample coil acts as a
receiver or pickup coil and by induction (Lenz’ law) produces an induced oscillating voltage
(emf) at the precession frequency of this transverse magnetization. By the coil’s position in the x-
y plane and because Mz does not precess, the coil is insensitive to the component Mz.

Let us now look at the precession process in more detail. Optimized absorption of the RF
pulse happens when the RF pulse frequency is the same as the precession or Larmor frequency of
the specific nuclear spin. The duration of the RF pulse can be adjusted to cause the
magnetization that is initially along the z-axis, Mz, to be tipped into the x-y plane. We call this a
90° pulse or π/2 pulse (see Fig. 4 top ii). With the pulse off, the nuclear spins will start to precess
around B0 .

The 90° pulse is special as it tips the spins into the transverse plane. But by varying the pulse
width (duration), we can in fact tip the spins by any angle. When in resonance (with the RF pulse
frequency and the Larmor precession frequency equal), if a pulse lasting time t tips the spins into
the x-y plane, a pulse lasting 2t, will tip the spins until the net magnetization is pointing in the
direction opposite to B0. This 2t pulse is then a 180° pulse or π pulse, which takes the spins from
the +z direction to –z. The pickup coil will then show no signal (again no transverse
magnetization). A 3t pulse will take the spins to 270°, which this time can be observed by the
pickup coil signal.

5
Fig. 4: Initially, the magnetization lies along the z-axis (top i), and thus no transverse
magnetization is observed (bottom i) by the pickup coil. A 90° pulse tips the nuclear
spins into the x-y plane, and now the net magnetization vector is transverse, in the x-y
plane (top ii), and we see a signal for transverse magnetization (bottom ii). With the
RF pulse off, the transverse magnetization starts to dephase in the x-y plane (top iii),
and the magnetization tries to align with the z-axis again. Both phenomena result in
a decay in the transverse magnetization until the transverse magnetization disappears
(bottom iii).

Fig. 5a: Precession of the


Fig. 5b: The induced emf resulting from the
magnetization around B0 after a
precession of the spins around B0 , giving
90° pulse tipping the spins into the
rise to an oscillatory free induction decay
x-y plane, and subsequent
signal (blue) and its envelope (red).
realignment along B0.

How is the NMR signal observed? When the RF pulse is applied, the spins, initially along z, are
tipped towards the x-y plane (Fig. 4 and Fig. 5a). A 90° pulse puts the magnetization vector (red
vector MXY in Fig. 5a) in the x-y plane. When the pulse is over, this vector precesses around B0

6
(grey spiral in Fig. 5a). This oscillatory trace induces an emf in the sample coil (see Fig. 3a). Due
to decoherence of the spins (by more than one mechanism, see later), the magnetization will tend
to fan out (dephase) in the x-y plane (Fig. 4 iii top) and realign along B0 (grey spiral in Fig. 5a)
and by both these phenomena the induced emf (blue trace in Fig. 5b) decays exponentially in
amplitude over time (Fig. 4 iii bottom). This is denoted as the free induction decay (FID) of the
spins. The envelope of the FID is indicated by the red decay curve in Fig. 5b. When the
magnetization vector is aligned back along z (initial position), the FID signal amplitude is zero
again. The FID is detected by the pick-up coil of the NMR spectrometer, and can be observed on
a scope. You will use the FID to find various relaxation times.

What are relaxation times? When the sample is in thermal equilibrium, i.e. after you placed
your sample in the magnet and waited for a while, you have your M along the direction of B0
(Fig. 4 i), the initial condition corresponding to a thermal equilibrium magnetization. Any change
in the orientation of the spins from this initial state is followed by a decay to get back to this
initial equilibrium condition. The decay is exponential in time. We can measure the associated
time constant, the T1 relaxation time. T1, also called the spin-lattice relaxation time, is the time
characteristic for establishing thermal equilibrium magnetization in the B0 direction (Fig. 6).

Fig. 6: Diagram showing the process of T1 relaxation after a 90° RF pulse is applied at
equilibrium. The z-component of the magnetization, Mz is reduced to zero, but then
recovers gradually back to its equilibrium value if no further RF pulses are applied. The
recovery of Mz is an exponential process with a time constant T1. This is the time at which
the magnetization has recovered to 63% of its value at equilibrium, M0.

7
Note that because the states with M//z or M in the x-y plane have different energy in B0, the
T1 process involves an energy relaxation. But other relaxation pathways also occur. The spin-
spin relaxation time, T2, is the time constant for the exponential loss of x-y (transverse)
magnetization due to dephasing (decoherence, fanning out) of the spins relative to each other.
The components of M in the x-y plane will also want to relax to their initial values (of zero). As
you may expect, the time that it takes for this transverse magnetization to decay is at least as fast
as T1. But, the transverse magnetization may also be lost by additional processes which cause
the components of the x-y magnetization to fan out or dephase (Fig. 4 iii and Fig. 7), such as due
to local fields and “spin flip-flops”. These additional processes do not necessarily need changes
in energy and don’t necessarily affect T1, but they do affect T2. Spin-spin (T2) relaxation is
thus faster than T1 (spin-lattice) relaxation, and typically T2 ≤ T1.

There are various factors that contribute to this spin-spin relaxation. The T2 decay process,
influenced by neighboring or “local” spins, contains useful chemical information about the
sample. This variation in the “local field” surrounding individual nuclei, which is created by
magnetic properties of nearby atoms, changes the local precession frequency for one spin
compared to another, and leads to dephasing of spins in the x-y plane (Fig. 4 iii and Fig. 7). A
term you may come across in NMR literature that is related to these changes in the local
precession, is called the “chemical shift” which provides information about nominally the same
spins in a compound. An example is protons in methanol, CH3OH where there are in fact two
types of protons with different chemical environments (the –CH3 and the –OH).

While we really need T2 relaxation times to obtain information regarding our sample, what
we observe is a T2* decay (Fig. 7), a combination of several contributions: most importantly
saturation recovery (measured by T1, trying to get the magnetization back along z), dephasing in
the presence of fluctuating magnetic fields due to the motions of spins in a lattice (this is the
actual T2), and effects of magnetic field inhomogeneities in the instrument.

Understanding T2*: If the external magnetic field B0 across the sample is not perfectly
homogeneous, spins in different physical locations will precess at different rates. This means that
the precession of all the individual spins is no longer in phase. This dephasing effect is referred
to as inhomogeneous dephasing, and it is not due to relaxation processes. Over time, the phase
difference between the precessions of the individual protons increases and the net voltage
induced in the pick-up coil decreases. If the time for this loss of signal due to inhomogeneous
dephasing is short, then it puts an upper limit to the actually observed T2*. The FID observed
after a single 90º pulse is often due primarily to this effect. If, however, the external magnetic
field is very homogeneous, then T2* ≈ T2, and the FID signal will represent a true measure of
T2. In high quality NMR instrumentation, shim coils are used to make the B0 as homogeneous as
possible. Still, it is never possible to eliminate all influences of B0 inhomogeneities , magnetic
field instability over time, or other instrumentation limitations, in any NMR experiment. So, T2*

8
Fig. 7: Diagram showing the process of T2 and T2* relaxation after a 90° RF pulse is
applied at equilibrium. At the start of the FID, the phases of the excited spins are in phase
or coherent. This coherence is lost due to systematic T2* processes (from imperfect
instrumentation), and due to true T2 processes that provide information about the sample.

effects may be minimized, whereas T2 effects are intrinsic to the sample. Experimentally, one
tries to minimize T2* contributions to be able to obtain T2 (and T1) information.

NOTE: In the setup you will use, if the T2 of the sample is less than 0.5 ms then the FID
time constant can be used as a good measure of the actual T2. However, if the actual T2 of a
sample is longer than a few ms then instrumentation limitations come in, and what you then
measure tends to be T2*. In that case we will need to use special RF pulse sequences, such as
the “spin echo experiment” or the “CPMG Echo Train” to measure T2. Advances and
extensions of NMR, e.g. in quantum information processing, often involve designing pulse
sequences that can achieve certain spin states and certain measurements such as T2.

9
EXPERIMENTAL SET-UP AND PROCEDURES
THE APPARATUS
In Fig. 8 below we have the schematic diagram of the major parts of the pulsed NMR
spectrometer. The magnetic field is provided by a permanent magnet consisting of two parallel
pole tips connected by a soft-iron flux-return yoke. The direction of the magnetic field defines
the z-axis, the quantization axis in the quantum mechanical picture.

Fig. 8

The NMR “probe” is a coil oriented with its long axis perpendicular to the z-axis and to B0,
so that the RF pulse generated by the coil, B1, is perpendicular to B0. To transmit the RF pulse to
the coil, the coil is connected to the output of an RF amplifier source. The frequency of the
source oscillator (RF synthesized oscillator) is digitally controlled, and its (pulsed) amplitude is
controlled by the pulse programmer. The most important feature of this is that the RF pulsed
field can be turned on in pulses at pre-programmed times and with pre-programmed pulse
duration. This is achieved with an electronic on-off switch that is controlled by logic signals
from the pulse programmer. The pulses are “top hat” shaped and can be varied in duration,
number of pulses, and repetition time between pulses. All these characteristics of the pulses can
be displayed on an oscilloscope.

The signal from the precessing spins (the FID) is sensed by the coil, and sent to the receiver
(input signal amplifier). In most applications where high frequency signals are measured, we do
not directly monitor at RF frequencies; rather we use a form of detector that generates a low
frequency beat by mixing the signal with a reference or carrier signal of nearly the same
frequency (heterodyning). In the pulsed NMR apparatus, a single tunable signal generator is used
to generate the B1 and the reference signal. In a second circuit, called the RF amplitude detector,
the RF signal is rectified and converted to a slowly varying measure of the signal amplitude. The
apparatus (Fig. 9) has both a mixer that produces a low frequency beat signal, and an RF detector
that produces the slowly varying signal proportional to signal amplitude.

10
Fig. 9: The front panel of the electronics modules. The Left Module is the Receiver. The
Central Module is the Pulse Programmer. The Right Module is the RF
Oscillator/Amplifier Mixer combination.

To understand the actual electronics, let’s first note that the equipment can generate two types of
pulses independently: the A pulse and the B pulse. This is so you can generate e.g. a 90° pulse
followed by a timed interval and then a 180° pulse, to design different experiments. Both A and
B pulses contain the same RF frequency and amplitude, but can be varied in in other parameters
that we will come back to later.

Let’s follow the progression of the signal. The module is in bold; the connector titles are shown
in parenthesis. Pulse Programmer (A+B Out: this will tell the Osc/Amp/Mixer how to shape the
RF pulses) ► (A+B In) Osc/Amp/Mixer (RF Out: pulsed RF goes to sample coil and coil acts
as pickup coil) ► (RF In: the tiny spin precession signal from the coil is fed into the receiver)
Receiver (RF Out: an amplified version of RF In; and Detector Out: the amplified envelope of
RF In) ► to the mixer (see later) and oscilloscope.

One role of the Receiver module is to amplify and “forward” the signal coming from the
sample coil. As the spins precess inside the coil, by Lenz’ law they induce an oscillatory voltage.
The amplified signal can be observed at the RF Out connector. The signal usually associated
with NMR, however, and shown in the lower trace of the schematic scope capture in Fig. 10,
comes from the connector labeled Detector Out (signal amplitude envelope): the signal is first
rectified so that all values become positive and then only the maxima for each cycle are selected.
It is this rectified envelope which is referred to as the free induction decay or FID.

11
The Osc/Amp/Mixer (sometimes called Synthesizer) module displays the frequency in
MHz of the RF pulse being used to tip the spins. As mentioned, for on-resonance operation, this
frequency must be adjusted to match, with high
precision, the Larmor precession frequency of the
nucleus of interest, here protons. The Larmor
frequency depends on B0 and for the set-up is
engraved at the back of the magnet.

The signal from RF Out on the Receiver


module can be used to match the RF frequency of
RF Out on the Osc/Amp/Mixer module to the
Larmor frequency (resonance). This happens by
connecting RF Out on the Receiver to Mixer In
on the Osc/Amp/Mixer. This allows the signal
from the precessing protons to be multiplied Fig 10: Upper trace, blue: Mixed signal.
(mixed, heterodyned) with a reference signal Lower trace, yellow, FID envelope.
coming from the Osc/Amp/Mixer. The result is monitored on the Mixer Out of
Osc/Amp/Mixer. This is the signal shown as the upper trace of Fig. 10. How does this signal
multiplication work and how does tell us if we are on resonance? Look at what happens when we
multiply two sine waves mathematically. If the synthesized frequency is ω1 and the free
precession frequency is ω2, a bit of trigonometry gives 2 sin(ω1) sin(ω2) = cos (ω1 - ω2) - cos
(ω1 + ω2). A low-pass filter inside the mixer allows only the difference signal cos (ω1 - ω2) to
reach Mixer Out. When the oscillator is properly tuned to the resonant frequency, then ω1 = ω2
and the signal from Mixer Out shows no “beats”. The upper trace of Fig. 10 indicates that when
this measurement was made the system was not exactly on resonance, but is very close and
requires only a little frequency tweak (you will do this tweaking too). In many permanent magnet
NMR systems, the precession frequency drifts because the temperature of the permanent magnet
is not absolutely constant. Any change in magnet temperature causes a change in B0 and thus in
the precession frequency. The set-up has a temperature and thus field stability of one part per
million over 25 minutes. Finally, note that the clever heterodyning concept is not unique to
NMR, but ever since its invention in radio technology has become ubiquitous in
telecommunications and science.

The Pulse Programmer determines the duration of individual RF pulses as well as the
number of pulses in a series, the spacing between pulses, and how often an entire series is
repeated. The PP-101 pulse programmer provides two different pulses, A and B, both containing
the same RF frequency. The duration and number of pulses, however, are independent. The
selected pulse pattern is sent from the A+B Out connectors of the Pulse Programmer to the
A+B In connector of the Osc/Amp/Mixer. For some experiments, Sync Out can connect to the
scope to provide a trigger to tell the scope when to start recording a signal. The Sync toggle

12
switch can be set to either A or B pulse. Now let’s look at how each of the pulse parameters is
controlled by the settings on the Pulse Programmer.

A-Width and B-Width control the duration in time each of these RF pulses persists. That way
you can set e.g. A to a 90° pulse and B to a 180° pulse.
Number of Pulses (N) : the pulse programmer provides only one A pulse, but there can be
anywhere from 0 to 100 B pulses in a given series.
Delay time (τ) : the time between the start of the first A and the start of the first B pulse of a
series. When more than one B pulse is used, the system adjusts subsequent delay times between
B pulses to 2τ. We will see why this matters later. Note that the delay time can only be set up to
9.99x103 s (the setting with 4 as exponent is not functional).
Repetition time (P) : sets the time between the start of entire pulse sequences. This is set by
two dials. The scale dial sets time ranges (100 ms, 1 s, ...). The variable dial sets a time from
10% to 120% of the scale. The total time for each entire pulse series itself, is determined by the
combination of N and τ. This must be less than the repetition time P so that the digital logic does
not lock up. In addition, P must be long enough so that, after the pulse series had ended, the net
magnetization has time to realign with the primary field. If P is too short, a pulse sequence will
begin before the system has returned to thermal equilibrium and the FID will make little sense.
In fact P should be such that the time after the last echo or FID, is long compared to T1. To be
safe it should be close to at least 5 T1, preferably 10 T1 or more.

Mode: this allows you to select who decides when a pulse sequence starts. On Int, the repeat
is automatic and set by Repetition Time. On Man the pulse sequence will start when you press
Man Start. On Ext the pulse sequence will start when an appropriate external trigger is fed into
Ext Start.

The Pulse Programmer has a Blanking Out output, which should be connected to Blanking
In on the Receiver. The reason for blanking is as follows. The RF pulse exciting the spins is
much larger than the tiny induction signal the spins respond with. The signal from the sample
coil during and just after the pulse would overwhelm the receiver. So, the receiver needs to wait
until the pulse is over and the coil ringdown is over, and only then record the induction signal.
The receiver input needs to blanked during that wait time.

13
PROCEDURES
I. FINDING THE INITIAL SIGNAL
To gain familiarity with the apparatus and understand each of the following functions, it is , a
good idea to use a scope as illustrated below. Follow the connections as depicted in Fig. 11
(note: the connections are mostly already made):
• pulse sequencing,
• detection,
• signal mixing to low frequency,
• RF amplitude detection.
The best way to practice is by having a sample in the magnet and trying to observe a signal.

Fig. 11: The set-up as in the lab. The permanent magnet is on the left. The spectrometer
is the box under the scope. Close-ups of the scope traces are in following figures. Your
scope may look slightly different, so you should adapt accordingly.

14
(continued: finding the signal)
Samples (vials) are described in Appendix 2. Place a sample in the probe: glycerin is a good
choice. H2O with Cu(NO3) 2 would also be a good choice because T1, the time constant for
recovery of the equilibrium nuclear magnetization, is shortened to a few ms by the paramagnetic
Cu+2 ions. In pure H2O, T1 ≈ 3 seconds so that the time between measurements must be several
times this, which would be quite inconvenient.

Make sure the connections between the modules, and between modules and probe, are as in Fig.
11. If you read the previous paragraphs, these connections might start to make sense.
The scope is connected as follows:
• RF Out from Receiver on Ch 1 (yellow), scale of 20 mV/div.
• Mixer Out from Osc/Amp/Mixer on Ch 2 (blue), scale of 5 V/div.
• Pulse A+B Out from Pulse Programmer on Ch 3 (pink), scale of 1 V/div.
• Detector Out from Receiver on Ch 4 (green), scale of 1 V/div.
• Time scale on 100 µs/div.
• Trigger set to Ch3 (A+B Out), so that the sequence that the scope displays starts exactly
when the actual pulses start. Trigger on rising slope, mode normal, coupling DC. A very
good trigger alternative is to use the scope’s external trigger, connected to Sync Out (3
V, 250 ns pulse) on the Pulse Programmer, with Sync set to A. This external trigger
method will show its usefulness later.
The following settings of the NMR equipment form a good starting point:
Osc/Amp/Mixer:
• Frequency at 15.495 MHz (coarse or fine). Note that the frequency will change
depending on the mood of the magnet. So we start with this value and will change the
frequency later, using the frequency adjust knob.
• CW-RF ON.
• M-G ON.
Pulse Programmer:
• A ON, B OFF.
• A Width: at 3rd notch (B width irrelevant as B is off). This sets A to approximately a 90°
pulse, rotating the magnetization in-plane and allowing observation of a good FID.
• Delay time (between A and B pulses): 2 ms (irrelevant as B is off).
• Number of B pulses: 1 (irrelevant as B is off).
• Repetition Time: 100 ms (knob on 100 ms, variable on ~ 100%).
• Mode: internal.
• Sync: A.
Receiver:
• Gain: 12 o’clock.

15
• Blanking ON.
• Time constant: at minimum.
• Tuning: see later, to taste, to optimize signal on the scope.

The following figures illustrate what you might see on the scope.

Fig. 12: Scope picture. Note that the time base has been shifted to the left by 400 µs to
show a more complete picture.
Ch1 is an amplified version of the actual oscillating response induced by rotating magnetization
in the coil, decaying as time goes on. The oscillations are rapid (~ 15 MHz) and fill the scope
screen. Once we know we have a signal we will be able to ask the scope not to display Ch1,
making the scope pictures clearer.
Ch2 is the mixer output. As displayed, the frequency looks tuned (no oscillations in Ch2). Later
we show a picture where the frequency is not tuned, for comparison.
Ch3 is barely visible on this time scale. It is the brief spike at the start of the Ch1 oscillations (as
well as the horizontal lines at zero before and after).
Ch4 is the most important trace. This is the actual FID, the envelope of the oscillations of Ch1.
The FID signal rises rapidly after the pulse, and then decays exponentially.

16
Fig. 13: Another scope picture depicting an initial setup. The time base (unshifted)
here is magnified to 10 µs. The A pulse is visible on Ch2 at t=0. The FID signal starts
only with the rise of the signals on Ch1 and Ch4. The signals between the pulse and the
FID are due to parasitic oscillations from the effect of the pulse, coil ringdown, etc. We
will ignore those, and focus on the actual FID.

II. TUNING THE SYSTEM

Now that you have a signal, you can tune the system. This happens in two steps: with the
Receiver tuning knob, and with the Osc/Amp/Mixer frequency setting.

Tuning knob on the Receiver: Turn this knob slightly until the signal on Ch4 (FID, Detector
Out) is maximized, or its noise minimized. This tuning knob optimizes the RF match between
the RF equipment and the sample probe, which depends on exact routing of cables etc...

Frequency setting on the Osc/Amp/Mixer: This allows you to set the RF frequency in the pulse
to exactly match the Larmor spin precession frequency, so that you are on-resonance. The field
B0 of the magnet merrily drifts with temperature and time, and hence the Larmor frequency drifts
too. From previous paragraphs you will remember that Mixer Out (Ch2) is a beat signal at a
frequency that is the difference in frequencies between the Larmor frequency (signal in) and the
RF output frequency (displayed on the Osc/Amp/Mixer). What you should do is make slight
changes to the Frequency setting until Mixer Out (Ch2) shows almost no oscillations. Then the
difference in frequencies is minimal, and you are on-resonance. Hint: the frequency changes you
make should be slight, otherwise you can land at totally wrong frequencies. Use the coarse

17
setting, then the fine setting. Figure 14 shows the difference in scope traces for Ch2 (Mixer Out)
off-resonance vs on-resonance.

Fig. 14: Scope pictures depicting Ch2 (Mixer Out), Ch3 (A+B Out) and Ch4 (Detector
Out). Ch2 (Mixer Out) tells you if you are at resonance. In the upper panel, Ch2 shows
oscillations, denoting an off-resonance condition. In the lower panel the oscillations are
minimal, denoting resonance.

18
III. FINDING THE PULSE DURATION FOR 90° AND 180° PULSES

The duration of the RF pulse determines the angle over which the spins will be tilted away from
the z-axis. We experiment with the A pulse, and the B pulse will behave identically if both
pulses are well separated in time. On the Pulse Programmer make sure A is on and B is off.
On the scope, display Ch3 (A+B Out) and Ch4 (Detector Out). Other channels are optional.
Turn the A Width to its minimal setting (counter clockwise). The Detector Out signal should
decrease in height. Slowly turn the A Width to longer durations (clockwise), until a maximum
height in Detector Out signal is reached. At this setting the A pulse is a 90° pulse. The Detector
Out signal is maximum because the spins have been rotated to lie in-plane, 90° to the z-axis, and
now induce a maximum signal in the sample coil before their orientation decays back to the z-
axis. Zoom in on time axis of the scope, and from the A+B Out trace measure the duration of the
A pulse using the scope cursors (about 5 µs).

Now increase the A Width further. The Detector Out signal should again decrease in height,
until it reaches a minimum. At this setting the A pulse is a 180° pulse. The Detector Out signal
is minimum because the spins have been rotated to lie along the negative z-axis, 180° to the z-
axis, and now induce a minimum signal in the sample coil. Zoom in on time axis of the scope,
and from the A+B Out trace measure the duration of the A pulse using the scope cursors (about
10 µs, or double the duration of the 90° pulse). Now increase the A Width further again. The
Detector Out signal should again increase in height, until it reaches a new maximum. At this
setting the A pulse is a 270° pulse. Etc...

IV. TYPICAL SETUP

A typical scope picture for the experiments described in the next paragraphs is shown below for
your reference (Fig. 15). In this example only the A pulse is used. Only Ch3 (A+B Out) and
Ch4 (Detector Out) are displayed to minimize clutter. You can of course change the scope
parameters to display the trace in the form that is easiest for you to use.

Fig. 15: Typical scope traces.

19
EXPERIMENTS

I. PRELIMINARY

You performed the previous steps using as sample e.g. glycerin or mineral oil or H2O with
Cu(NO3) 2. Now try the CCl4 vial (carbon tetrachloride, a once-widely-used solvent). Do you
observe an NMR signal from this vial at the frequencies we use in our spectrometer? Why not?
At what frequencies do you expect a signal from CCl4 in our magnet? Will this signal originate
in C or in Cl? Hint: look up γ for the various Cl isotopes.

II. MEASURE T1, THE SPIN-LATTICE RELAXATION TIME

A) Estimate T1

The repetition time can be used to estimate T1, the time characteristic of re-establishing
thermal equilibrium magnetization in the z-direction. On the Pulse Programmer make sure A is
on and B is off. On the scope, display Ch3 (A+B Out) and Ch4 (Detector Out). Other channels
are optional. Set the scope to trigger on A+B Out. Read the frequency displayed on the scope, if
any (if the frequency is too low, the scope displays “< 10 Hz”, don’t worry about it yet): this is
now the repeat frequency of the A pulse. For instance, if the repetition time was set at 50 ms, the
frequency will be 20 Hz. If you increase the time base on the scope, you will see that the signal
collapses in a series of pulse-like signals, spaced by the repetition time. Set the time base back to
about 50 µs. Set the A pulse duration (A Width) for the first maximum, starting from minimum
width (90º pulse, about the 3rd notch). Decrease the repetition time on the Pulse Programmer
until the Detector Out signal begins to shrink from its maximum height. This decrease in the
height of the FID occurs because the z-magnetization has not returned to its thermal equilibrium
value before the next 90º pulse. In the repetition time where this happens, we have found a
rough measure of T1. The decrease in FID height becomes more dramatic as the repetition time
decreases. If the frequency is displayed on the scope, you can now read off the estimate of T1
from the scope, by converting the frequency reading (e.g. 46 Hz corresponds to 22 ms). For long
T1 times the repetition time you find is so long that the scope can no longer provide it (displays
“< 10 Hz”). Then you may need to use a stopwatch or equivalent, measuring the time between
screen refresh cycles on the scope. It has to be emphasized that this measure of T1 is very
rough. The values are typically longer by about a factor 10 from a more precise measurement
(B, below). But still, they do offer a good estimate for the order of magnitude, and they show
you the effect of the repetition time. Perform this measurement of T1 for the following three
samples: glycerin, 0.5 M aqueous solution of NH4Cl , 0.125 M aqueous solution of Cu(NO3)2 .

Another way to obtain an estimate of T1 would consist of measuring the time it takes for the
FID to decay by 63%. This method relies on the exponential nature of the FID decay. As we
have emphasized above, T1 is the characteristic time for the system to return to equilibrium
under applied static magnetic field B0:

20
<MZ(t=T1)> = M0(1−e-1) where M0 is the equilibrium magnetization along z with B0 applied.
We won’t perform such measurements of T1 because again the measurements will be too
approximate. Yet note that in this particular type of T1 measurement, you should wait at least 10
x T1 between pulses (using long repetition times), to make sure the magnetization has had time
between pulses to decay back to lie along the z-axis and reach the value M0.
B) Obtain a more precise measure of T1

A more precise measurement of T1 requires a two-pulse sequence. Suppose that we initially


start in equilibrium, with the magnetization component along z, Mz = M0, the equilibrium
magnetization along z. The magnetization is then tipped by 180º to the -z direction. Left alone,
Mz will return exponentially towards its thermal equilibrium value, following the expressions:

𝑑𝑀𝑧 𝑀0 − 𝑀𝑧
= 𝑤𝑤𝑤ℎ 𝑀𝑧 (𝑡 = 0) = −𝑀0
𝑑𝑑 𝑇1

This has as solution:

𝑀𝑧 (𝑡) = 𝑀0 �1 − 2𝑒 −𝑡⁄𝑇1 �

We have to somehow probe the spins as they are moving from the –z to the +z direction,
reestablishing thermal equilibrium. But, the spectrometer cannot directly detect Mz. Only the
precession in the x-y plane which induces an emf in the sample coil can be detected. What we
can do is to tip the magnetization component along z into the x-y plane, and detect it that way.
So, we follow the initial 180º A pulse (which gives us Mz (t=0) = -M0,) with a 90º B pulse,
delayed from A by some time τ. The maximum height of the FID we measure after the B pulse
is then proportional to Mz (τ). Note that this proportionality will do fine, since to obtain T1 from
the expression of the time dependence, all we actually need is Mz (τ)/M0. We will plot the
maximum height of the FID vs τ, the delay time between A and B pulses. This height plot can
be calibrated to read Mz (τ)/M0 since for long τ, we have Mz (τ)/M0 → 1. Fitting the now-
calibrated plot of Mz (τ)/M0 allows us to extract T1.

Note two things.


(1) The expression for Mz (τ)/M0 changes sign at τ = T1 ln(2). Yet, the maximum FID height
appearing on the scope has no sign (always positive-going) since the detector gives absolute
values. Careful observation of the mixer signal would show a phase shift as the magnetization
passes through 0, corresponding the sign change. But, we will tend to ignore the sign. The data
you record vs τ will all be positive numbers. When you plot the data to perform the fit, you can
insert a minus sign by hand for the data you know should be negative, namely for the data at

21
shorter τ, before the zero-crossing occurs. This will give you plots as in Fig. 16, showing
sketches of both the actual net magnetization Mz (τ) and the maximum height of the FID just
after the 90° pulse, as a function of τ.

Fig. 16: Upper part: Net z magnetization vs delay time.


Lower part: Maximum amplitude of FID vs delay time.

(2) The repetition time between sequences of A+B pulses should be sufficient for Mz to reach
M0 again, the equilibrium magnetization along z. Repetition time ~ 10 x T1 should work.

With only the A pulse active, insert a vial (glycerin is a good choice to start with) in the
magnet and find the 180º pulse width as follows (similar to above). To tip the magnetization to -
M0 (along –z), the length of the A pulse is increased until the FID height on Ch4 (Detector Out)
has passed through the first maximum (the 90° pulse) and then reaches a minimum on the scope
(at twice the 90° A pulse width). This indicates a 180° pulse after which there is no net
magnetization in the x-y plane. Turn on the B pulse with minimal B width. Dial in an A to B
Delay Time ~ 3 times the width of the A pulse. This delay time should be much smaller than
your rough estimate of T1. Now adjust the width of the B pulse for a maximum FID height: this
B pulse tips the spins from the –z direction into the x-y plane where they can be maximally
detected. The B pulse is now a 90° pulse (Fig. 17).

Traces on the scope are clearest if you display only Ch3 (A+B Out) and Ch4 (Detector Out).
Other channels are optional. It is best to set the scope to trigger on the B pulse, by having it
trigger on the Sync Out output of the Pulse Programmer and selecting B for Sync. However
this type of trigger is optional, and you can keep the trigger to Ch3 (A+B Out). During the
measurement you may have to keep stretching the time base on the scope. Be mindful of
keeping the repetition time sufficiently long.

22
Fig. 17: The A (180º) and B
(90°) pulse sequence.
Depicted are all channels. In
particular, note Ch2 (pulses,
with A twice as long as B),
and Ch4 (the FID following
B).

Slowly and in steps increase the Delay Time (τ in our above discussion) between the A and B
pulses. Note that the delay time can only be set up to 9.99x103 s (the setting with 4 as exponent
is not functional). For each τ, record the maximum FID height (in Volts of amplitude) after the
90° B pulse, as seen on Ch4 (Detector Out). This maximum should initially decrease with
increasing τ.
When τ results in a minimum signal in the pickup coil, it means that the net magnetization
along the z-axis is zero (Mz (τ) ≈ 0, Fig. 16). When the signal again tends to a maximum at
higher τ, the spins have “relaxed” back to alignment along the z-axis, and Mz (τ)/M0 → 1.

As explained above, plot the FID amplitudes vs delay τ, at small τ as negative values, and
after the minimum is observed (this should be ~ 0 V), plot the FID amplitudes as positive
numbers. You can increase τ up to 9.99x103 s. You will obtain a curve that crosses zero (Fig.
16). You can also additionally plot your curve in all positive numbers (Fig. 16). Also plot a
normalized curve(s), representing Mz (τ)/M0, such that Mz (τ)/M0 → 1 at long τ. Fit the
normalized curve and find T1. Notice that Mz (τ)/M0 should change sign at τ = T1 ln(2).

Perform this measurement of T1 for the three samples: glycerin, 0.5 M aqueous solution
of NH4Cl , 0.125 M aqueous solution of Cu(NO3)2 . For samples with particularly long T1 (p-
xylene, 0.5 M aqueous solution of NH4Cl) the experiment can become somewhat difficult.

Comment on the differences in the values you obtained for the T1 between the approximate
method and this method, and between different samples.

(continued on next pages)

23
III. MEASURE T2, THE SPIN-SPIN RELAXATION TIME
Many of the setup features of this experiment will be similar to the experiment for obtaining
detailed values of T1. In fact, instead of using a 180º A pulse followed by a 90º B pulse, we will
use a 90º A pulse followed by a 180º B pulse. We will assume that you know how to find the
correct pulse widths, and how to set up the equipment and scope.
As explained above, T2 is the characteristic time for the spins to, by decoherence among
themselves lose their net x-y magnetization (established by a 90º RF pulse). In the x-y plane after
a 90 ° pulse, the individual proton spins begin to precess at different rates because each
individual proton spin experiences a slightly different magnetic field. This dephasing does not
involve a change in energy because the component of magnetization in the direction of the static
magnetic field is not changing. Some of the precessing protons will precess a little faster and,
therefore, begin rotating forward in a non-inertial frame rotating with the average. Some will
precess a little slower. The net result is that the individual magnetic moments fan out in the
plane and the vector sum of these individual magnetic moments goes to zero. The characteristic
time for this, again exponential, dephasing and loss of signal to occur is T2.

The dephasing has several contributions, leading to a true T2 and what we have called T2*,
the latter due to instrumental effects. We are interested in the true T2, and below we will find a
way for compensating for T2* using a spin echo effect. What is causing the true T2 are
fluctuating microscopic magnetic fields acting on each proton magnetic moment. If you sat on a
proton, you would see a small, random fluctuation of the magnetic field due to the random
motion of other protons in the neighborhood of the proton you were sitting on. These
neighboring protons have magnetic moments which create local magnetic fields. At any given
moment, the net magnetic field created at the proton of interest by the neighboring protons could
be in the direction of the static magnetic field B0. At the next moment, it may be in the opposite
direction of the static magnetic field. This small, fluctuating magnetic field causes our proton’s
precession frequency to vary and a collection of protons to dephase.

A) Measurement of T2 using the Hahn Spin Echo


Irwin Hahn found a way to compensate for the apparent decay of the x-y magnetization due
to inhomogeneity in an external magnetic field, correcting T2* and thereby obtain the true T2.
The external inhomogeneity creates a variation in the proton precession times around an average.
The introduction of a 180° pulse, or spin flip, allows the spins to regroup before again dephasing.
This creates a spin echo which allows us to measure the true T2.
We apply an initial 90° pulse. After our initial 90° pulse, the spins in areas of stronger than
average fields precess faster than average and those in weaker fields more slowly. The spins
dephase and the induced emf fades. Now we apply a second pulse, a 180° pulse. After the 180º
flip, however, the spins that were “ahead” because they are in a stronger field are now “behind.”
Because their protons are precessing faster, however, they will now “catch up” to the “average.”

24
In the same way, after the 180° pulse, “slow” spins are now “ahead” and the “average” will
overtake them. The spins rephase momentarily and dephase again. We will see that the scope
trace Ch4 (Detector Out) after this 180º pulse shows a rise to a maximum and then a decay. The
initial dephasing, and then the rephasing and dephasing wings of the echo, are depicted in Fig.
18. If you vary the delay time between the 90º pulse and the 180º pulse, the echo maximum
decays. The decay of the echo maximum with delay time is due to the actual spin-spin
processes of T2, and not due to other effects such as magnetic field inhomogeneities.
Here follows an analogy (by Jonathan Reichert) to understand the effect of the 180º spin flip.
Consider the plight of a kindergarten teacher who must devise a foot race which keeps all
children happy, no matter how fast they run. What if the race has the following rules? All
children are to line up at the starting line. At the first whistle they are to run as fast as they can
down the field. At the second whistle they are to turn around and run back toward the starting
line. First person back wins. Of course, it is a tie, except for the ones who “interfere” with one
another or fall down. As the children run away the field spreads out, with the fastest ones getting
farther and farther ahead. At some point there is no semblance of order. On the trip back, as the
faster ones overtake the slow ones now in the lead, the group comes together again, “rephasing”
as they pass the start line. This is an analogy for the effect of the 180° spin flip which creates a
spin echo. The effect of the 180º pulse is analogous to that of the kindergarten whistle. After the
180° pulse, the signal increases as the spins rephase, hitting a maximum (somewhat lower than
the initial height of the FID) and decreasing as the spins again dephase. The decay of the
maxima shows how the protons are losing their coherent x-y magnetization. In our kindergarten
analogy, this tells us the rate (T2) at which the children are actually interacting with each other.

The spin-echo sequence is schematically shown in Fig. 18. This is an elegant way to

Fig. 18: Upper part: spin-echo


sequence with relevant delays.
Lower part: the magnetization
vector in the x-y plane, observed
from above.

25
Fig. 19

minimize various instrumentation influences on the experiment.


Figure 19 shows scope traces with the echo. Displayed are all channels. Note in particular
the two pulses on Ch3 (A+B Out) and the response on either Ch4 (Detector Out) or Ch1 (RF
Out). The A pulse occurs towards the left in the figure and Ch4 (Detector Out) or Ch1 (RF Out)
responds by showing a decaying FID. Then the spins are hit by the B pulse, in this case 1.58 ms
after the A pulse. The spins rephase and an echo appears on Ch4 (Detector Out) or Ch1 (RF
Out). The echo reaches a maximum an exact 1.58 ms after the B pulse. So, the B pulse sits
midway between the A pulse and the echo, as expected. To access details of the echo, it is best
to only display Ch3 (A+B Out) and Ch4 (Detector Out) and to open the time base on the scope
appropriately.

If you apply the Hahn echo sequence of a 90º A pulse followed by a Delay Time τ followed
by a 180º B pulse, the echo forms after time 2τ. So, the spins have been dephasing due to true
T2 processes for time 2τ by the time you see the echo. Hence, the amplitude of the echo should
follow:

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑜𝑜 𝑒𝑒ℎ𝑜 ∝ 𝑒 −2𝜏⁄𝑇2


To obtain the T2 times with the Hahn echo sequence, the amplitude of the echo (in Volts
amplitude) is recorded for different Delay Time values between the 90º A pulse and the 180º B
pulse. A plot of echo amplitude vs Delay Time τ yields an exponentially decaying curve, as
noted above. The exponential decay constant for this curve equals ½ T2. From this observation
of course you can calculate T2.

26
Perform a measurement of T2 using the Hahn Echo for the three samples: glycerin, 0.5
M aqueous solution of NH4Cl , 0.125 M aqueous solution of Cu(NO3)2 .

Comment on the differences in the values you obtained for the T2 between different samples
(you should include the plots in your report).

Note: You won’t be able to increase the delay time very far before you start seeing the echo
signal vary randomly in height during each repetition cycle. This random variation in pulse
height for longer delay time is due to self-diffusion of the nuclear spins of the molecules during
the time between the 90° and 180° pulse, making the spins see different environments as they
diffuse through the magnetic field gradients. Self-diffusion is associated with translational
motion of molecules at zero gradient of the chemical potential. The solution around the problem
consists of using multiple-pulse / multiple-echo sequences, as shown below.

B) Measurement of T2 using the CPMG Echo Train sequence


Advances in modern NMR often consist of designing rather complicated pulse sequences that
can manipulate the spins to achieve certain spin states and certain measurements. Relatively
simple examples are the Car-Purcell pulse sequence, and the Car-Purcell-Meiboom-Gill (CPMG)
Echo Train sequence.
Car and Purcell designed a multiple-pulse sequence which reduces the effect of the spin
diffusion noted above. The spin self-diffusion problem occurs at long Delay Time τ. So, instead
of simply plotting the FID amplitude as function of increasing Delay Time τ , Car and Purcell
chop up the long time interval with a periodic regrouping of the spins, producing more echoes
after the first echo. The sum of the individual times between echoes functions as an effectively
long delay time while minimizing the problem of spin self-diffusion if the time interval between
individual echoes is not too long (if it is too long the spin self-diffusion problem comes up
again). The sequence consists of applying 180° pulses every 2τ after the first 180° pulse. So, the
Car-Purcell pulse sequence reads:
90° -wait τ - 180° - wait 2τ - 180° - wait 2τ - 180° - wait 2τ - 180° etc.
Yet, the Car-Purcell pulse sequence has a practical problem: it is hard to produce a pulse that
rotates the spins exactly 180°. If the pulse length is slightly off, giving a rotational error of e.g. a
few degrees, the error accumulates linearly with the number of pulses, and soon can reach
sizable angles. The effect is to make T2 appeared shorter than it actually is. So, Meiboom and
Gill devised a way to reduce the cumulative rotational error: their pulse sequence provides a
phase shift of 90° in the applied RF between the 90° pulse and the 180° pulse train, and all the
pulses are phase-coherent. This cancels the effect of the rotational error to first order. Their
sequence is called the Car-Purcell-Meiboom-Gill (CPMG) Echo Train sequence. We will use it
to measure T2 more accurately than we did above.

27
The Pulse Programmer has the Car-Purcell pulse sequence and the CPMG pulse sequence
built-in. If on the Pulse Programmer you dial “Number of B Pulses” > 1, the output will consist
of a not just one A pulse followed by one B pulse after Delay Time τ, but one A pulse followed
by one B pulse after Delay Time τ, and then a series of B pulses spaced from each other by 2τ.
This constitutes a Car-Purcell pulse sequence. The Pulse Programmer has a Meiboom-Gill
capability too, via the M-G Out connector. M-G Out on the Pulse Programmer should be
connected to M-G In on the Osc/Amp/Mixer module, and on the Osc/Amp/Mixer module the M-
G switch should be “on”. The output from the Osc/Amp/Mixer module will now be a CPMG
pulse sequence.

On the Pulse Programmer, dial “Number of B Pulses” = 10. Make sure the M-G capability is
active. Adjust the scope settings to see a display similar to Fig. 20 (pulse sequence and FID).

Fig. 20

In Fig. 20 we have used Delay Time τ = 1 ms as an example (on the glycerin sample). Note
(top trace) the initial 90° A pulse followed by one 180° B pulse after Delay Time τ = 1 ms,
followed by a series of 9 180° B pulses spaced from each other by 2τ = 2 ms (total number of
180° pulses = 10). Note (bottom trace) the FID signal: the initial Hahn echo at 2τ after the initial
90° pulse is now followed by a train of diminishing echoes each spaced by 2τ from its neighbors.
The echoes form an exponentially decaying series. Each echo in the train has an amplitude,
expressed in Volts amplitude, that is smaller than its predecessor by a factor exp(-2τ/T2). This
fact allows you to fit the series of Volts amplitude you measure for the echoes and to obtain T2
from the fit.

28
Just to appreciate the difference between a Car-Purcell pulse sequence and a CPMG pulse
sequence, for a moment turn off the M-G switch on the Osc/Amp/Mixer module. The decay of
the echoes in the train should then be much more rapid. The rapid decay of the echoes is due to
the cumulative rotational error not being corrected for. Turn the M-G switch back on for the
measurements you will perform.

Perform a measurement of T2 using the CPMG pulse sequence for the three samples:
glycerin, 0.5 M aqueous solution of NH4Cl , 0.125 M aqueous solution of Cu(NO3)2 . Choose
the Delay Time τ for individual samples with some care. It should not be so long that spin
diffusion becomes important between 180° pulses, and not so short that the exponential decay is
not clear. It is straightforward to adjust τ while gauging the effect on the scope before you take
measurements. You can also adjust the dial “Number of B Pulses” to obtain reliable data.

Comment on the differences in the values you obtained for the T2 between different samples
and also between the measurements obtained using the CPMG pulse sequence and obtained
using the Hahn Echo (you should include the plots in your report).

29
APPENDIX 2: DESCRIPTION OF THE SAMPLE VIALS

Proton NMR samples: differently bonded hydrogen

1) Yellow cap with gold band: glycerin [glycerol, or propane-1,2,3-triol, or C3H8O3].


Hydrogen here occurs in O-H type and in C-H type bonds, with several types of environment for
each bond, as the example types below show.

2) Green cap with gold band: light mineral oil [alkanes, CnH2n+2, mineral origin, i.e. petroleum
distillates].
Hydrogen in hydrocarbons occurs in several types of C-H bonds, at least e.g. in CH2 and CH3
units if the alkanes are completely linear. Branching introduces CH units also. Butane (linear)
and isobutane (branched) are examples of very light alkanes (in fact gaseous, not liquid, at STP),
for illustration purposes given as examples below.

3) Red cap with gold band: 0.5 M solution of NH4Cl [ammonium chloride] in H2O.
Hydrogen occurs in the N-H bonds, and in the O-H bonds in H2O. The T1 and T2 of the protons
(hydrogen) in pure H2O are very long (T1 ≈ T2 ≈ 3...4 s). The presence of NH4+ and Cl- does not
much affect T1 and T2. Expect rather long T1 and T2.

and
4) Blue cap with gold band: 0.125 M solution of Cu(NO3)2 [copper(II) nitrate] in H2O.
In H2O hydrogen occurs in O-H type bonds. In pure H2O the T1 and T2 of the protons
(hydrogen) are very long (T1 ≈ T2 ≈ 3...4 s). These long relaxation times are influenced by the
presence of any magnetic species, such as the Cu2+ ion. Cu2+ has electronic configuration
[Ar]3d9 and hence has an unpaired spin (leading to a relatively high magnetic moment). The
presence of Cu2+ considerably shortens both the T1 and T2 of the H2O protons. In many
environments (not in pure H2O however), T2 is shorter than T1 to begin with, and hence the
presence of Cu2+ proportionally affects T1 more strongly. This gives the opportunity to use
paramagnetic ions like Cu2+ as contrast agents in NMR for biological studies and in MRI (Mn2+
and particularly Gd3+ are typically used rather than Cu2+).

5) Green cap with double blue band: para-xylene (p-xylene, or 1,4 dimethylbenzene, or
C6H4(CH3)2 ; a starting material used in polymer production). This is benzene (C6H6) with two
methyl (CH3) substituents. At two opposite (“para”) sites, a methyl group has taken the place of
a hydrogen. The four unsubstituted hydrogens see quite different chemical environments than
the six hydrogens in the methyl groups.

6) Red cap with double blue band: CCl4 (carbon tetrachloride, a once-widely-used solvent). Do
you expect to observe an NMR signal from this vial at the frequencies we use in our
spectrometer?

You might also like