1D Pulse sequences
We now have most of the tools to understand and start
analyzing pulse sequences. Well start with the most basic
ones and build from there. The simplest one, the sequence to
record a normal 1D spectrum, will serve to define notation:
Vectors:
z z
Mo
x 90y x
pulse Mxy
y y acquisition
Shorthand:
90y 90y
n
According to the direction of the pulse, well use 90x or 90y
(or 90 if we use other phases) to indicate the relative
direction of the B1 field WRT Mo in the rotating frame.
The acquisition period will always be represented by an FID
for the nucleus under observation (the triangle).
Inversion recovery
Measurement of T1 is important, as the relaxation rate of
different nuclei in a molecule can tell us about their local
mobility. We cannot measure it directly on the signal or the
FID because T1 affects magnetization we dont detect.
We use the following pulse sequence:
180y (or x) 90y
tD
If we analyze after the pulse:
z z
x 180y (or x) x tD
y y
Since we are letting the signal decay by different amounts
exclusively under the effect of longitudinal relaxation (T1),
well see how different tDs affect the intensity of the FID and
the signal after FT.
Inversion recovery (continued)
tD = 0 z z
x 90y x FT
y y
tD > 0 z z
x 90y x FT
y y
tD >> 0 z z
x 90y x FT
y y
Depending on the tD delay we use we get signals with varying
intensity, which depends on the T1 relaxation time of the
nucleus (peak) we are looking at.
Inversion recovery (continued)
N N
at 40oC
If we plot intensity versus time (td),
we get the following:
Intensity ( )
I(t) = I * ( 1 - 2 * e - t / T1 )
time
Its an exponential with a decay constant equal to T1.
Spin echoes
In principle, to measure T2 we would only need to compute
the envelope of the FID (or peak width), because the signal
on Mxy, in theory, decays only due to transverse relaxation.
The problem is that the decay we see on Mxy is not only due
to relaxation, but also due to inhomogeneities on Bo (the
dephasing of the signal). The decay constant we see on the
FID is called T2*. To measure T2 properly we need to use
spin echoes.
The pulse sequence looks like this:
90y 180y (or x)
tD tD
Spin echos are probably the first pulse sequences developed,
even before FT NMR existed. Although they are very simple,
spin echoes are used as blocks in almost all complex pulse
sequences to refocus Mxy magnetization.
Spin echoes (continued)
We do the analysis after the 90y pulse:
z y y
x tD
x
x
y
y y dephasing
tD
x
x 180y (or x)
refocusing
Now we return to <xyz> coordinates:
y z
x
y
Spin echoes (continued)
If we acquire an FID right after the echo, the intensity of the
signal after FT will affected only by T2 relaxation and not by
dephasing due to an inhomogeneous Bo. We repeat this for
different tDs and plot the intensity against 2 * tD. In this case
its a simple exponential decay, and fitting gives us T2.
N N
at 90oC
Intensity ( )
I(t) = Io * e - t / T2
time
Applications of spin echo sequences
So far we havent discussed how chemical shift and coupling
constants behave during spin echos. Here well start seeing
how useful they are...
A pretty anoying thing we have to do in NMR spectroscopy is
phase the spectrum. Why do we have to do this? We have to
think on the effects of chemical shift on different components
of Mxy during a short time or delay.
This short delay, called the pre-acquisition delay (DE), is
needed or otherwise the remants of the high power pulse
will give us artifacts in the spectrum or burn the receivers.
y y
...
x x
During this pre-acquisition delay all the spins have the
opportunity to evolve under the effects of chemical shifts, and
when we finally turn on the receiver all of them will have a
certain phase with respect to the carrier. It will be a mixture
of absortive and dispersive signals...
Spectrum phasing
The phase of the lines appears due to the contribution of
absorptive or real (cosines) and dispersive or imaginary
(sines) components of the FID. Depending on the relative
frequencies of the lines, well have more or less sine/cosine
components:
S()x = cosines() - real spectrum
S()y = sines() - imaginary spectrum
What we want is the purely absorptive spectrum, so we
combine different amounts of the real (cosine) and the
imaginary (sine) signals obtained by the detector. The
combination depends on the frequency of the spectrum:
S() = S()x + [ o + 1() ] * S()y
o is called the zero order phase and 1 the first order
phase. The correction is usually done by hand. In some cases
(nuclei with low ), its pretty much impossible...
There is one experiment using spin-echoes that
theoretically allows us to avoid this. Well see later why it
is actually not that useful...
Spin-echoes on chemical shift
Now we go back to our spin-echoes. The effects on elements
of Mxy with an offset from the B1 frequency are analogous to
those seen for dephased Mxy after the / 2 pulse:
z y y
x tD
x
x
y
eff
After a time tD, the magnetization precesses in the <xy>
plane eff * tD () radians, were eff = - o. After the pulse
and a second period tD, the magnetization precesses the
same amount back to the x axis.
y y
180 tD
x x
eff
Apart from being upside down, we have no dephasing if we
start acquisition immediately after the second tD. In principle,
this sequence would give a purely absorptive spectrum...
Spin-echoes and heteronuclear coupling
We now start looking at more interesting cases. Consider a
13C nuclei coupled to a 1H:
13C C H J (Hz)
C H
1H
1H
C H
13C
I
C H
If we took the 13C spectrum we would see the lines split due
to coupling to 1H. The 1JCH couplings are from 50 to 250 Hz,
and make the spectrum really complicated and overlapped.
We usually decouple the 1H, which means that we saturate
1H transitions. The 13C multiplets are now single lines:
13C C H
C H
1H
1H
C H
13C
C H I
Spin-echoes and heteronuclear coupling ()
We modify a little our pulse sequence to include decoupling:
90y 180y (or x)
13C:
tD tD
{1H}
1H:
Now we analyze what this combination of pulses will do to
the 13C magnetization in different cases. We first consider a
CH (a methine carbon). After the / 2 pulse, we will have
the 13C Mxy evolving under the action of J coupling. Each
vector is said to be labeled by the states of the 1H it is
coupled to, and :
z y y -J/2
()
tD
x
x x
y ()
J/2
Remember that under J-coupling = * tD * J.
Spin-echoes and heteronuclear coupling ()
We now apply the pulse, which inverts the magnetization,
and start decoupling 1H. This removes the labels of the two
vectors, and effectively stops them. They collapse into one,
with opposing components canceling out:
y y y
x x x
In this case the second tD under decoupling of 1H is there to
refocus chemical shift and get nice phasing
Now, if we take different spectra for several tD values and plot
the intensity we get something that looks like this for a CH:
tD= 1 / 2J tD = 1 / J
tD
Spin-echoes and heteronuclear coupling ()
The signal intensity varies with the cosine of tD, is zero for tD
values equal to multiples of 1 / 2J and maximum/minimum
for multiples of 1 / J.
If we are looking at a CH2 (methylene), the analysis is
similar, and we obtain the following plot of amplitudes versus
delay times:
tD= 1 / 2J
tD
tD = 1 / J
Analogously, for CH3 (methyl), we have:
tD= 1 / 2J tD = 1 / J
tD
Spin-echoes and heteronuclear coupling ()
Now, if we make the assumption that all 1JCH couplings
are more or less the same (true to a certain degree), and
use the pulse sequence on the following molecule with a tD
of 1 / J, we get (dont take the values for granted):
OH
3 5 7
2 4 6
1 6
HO
1,4
150 100 50 0 ppm
5
2,3 7
The experiment can discriminate between C, CH, CH2, and
CH3, and we identify all the carbon types in the molecule.
This experiment is called the attached proton test (APT).
It is one of the first multiple pulse sequences, and has been
superseded by the INEPT and DEPT pulse sequences.
Spin-echoes and homonulcear coupling
Here well see why spin echoes wont work if we want to get
our perfectly phased spectrum. The problem is that so far we
have only used single lines (no homonuclear J coupling) or
systems that have heteronuclear coupling.
Lets consider a 1H that is coupled to another 1H, and that we
are exactly on resonance. After the / 2 pulse of the spin-
echo sequence and the td delay we have evolution under the
effects of J coupling. Each vector will be labeled by the state
of the 1H it is coupled to. We have:
z y y J/2
()
tD
x
x x
y ()
J/2
The problem is that if we now put the pulse, we invert the
populations of all protons in the sample. Therefore, we invert
the labels of our protons:
Spin-echoes and homonulcear coupling ()
The pulse flips the vectors and inverts the labels:
y J/2 J/2 y
() ()
180y (or x)
x x
() ()
J/2 J/2
Now, instead of refocusing, things start moving backwards,
and we will have even more separation of the lines of the
multiplet during the second evolution period. If we then take
the FID, the signal will be completely dispersive (although
this depends on the length of the tD periods):
tD FID, FT
x
Spin-echoes and homonulcear coupling ()
We see why this is not all that useful. For different td values
we get the following lineshapes for a doublet coupled with
a triplet (both have the same J value):
Despite of its patheticness, understanding how this works is
crucial to understand 2DJ spectroscopy. The phenomenon
is known as J-modulation.
Binomial pulses
Binomial pulses are examples of pulse trains which we can
explain with vectors. Among other things, we can use them
to eliminate solvent peaks (see T1).
The simplest binomial pulse is the 1:1, two / 2 pulses with
opposite signs, separated by a certain interval td, and exactly
on resonance with the peak we want to eliminate:
90y 90-y
tD
The first / 2 puts everything on <xy>. After td, signals/spins
precess to one side or the other of x. All except the signal we
are interested in eliminating from the spectrum:
z z y
Mo
90y tD
x
x x
y y
Binomial pulses (continued)
The next / 2 return everything on x to the z axis. This
includes all the signal corresponding to the peak to eliminate,
as well as the x components of the remaining signals:
y y
90-y
x x
x
y
The resulting FID only has signals corresponding to peaks
that arent in resonance with the carrier. They will all be in
phase with the receiver, but signals on each side of the
carrier will have opposite signs:
y
FID (y)
x
FT
Due to td, both peaks on resonance and those from signals
at multiples of 1 / (2*td) Hz will be nulled (any signal with that
frecuency with turn half a cycle in td).
Binomial pulses ()
As mentioned before, they are used to eliminate solvent
peaks, particularly water in cases that other secuences could
perturb protons that exchange with water (NHs, OHs, etc.).
~ 50 mM sucrose in H2O/D2O (9 to 1).
1H spectrum:
1H spectrum with 1:1 pulse (td = 200 S):
Binomial pulses ()
To avoid the sign change we can use other binomial pulse
trains, such as the 1:3:3:1:
1/ 90y 3/ 90-y 3/ 90y 1/ 90-y
8 8 8 8
tD tD tD
You also get artifacts. None of these pulse trains, nor
experiments that take advantage on T1 differences, give
results as good as those that are obtained with secuences
using gradients, such as WEFT or WATERGATE.
For the same sample this is what we get with WEFT: