NOE

The resonance is S is saturated, meaning that you pulse at low power directly on the
frequency of S so that only resonance S is affected by the pulse. If the pulse is long
(~seconds) then the populations of S will be equalized, thus S has been saturated. The
system will try to return to equilibrium, allowing for dipolar cross relaxation. If the W2
cross relaxation mechanism is favored, the intensity of resonance I will increase (a
positive NOE); if the W0 cross relaxation mechanism is favored, the intensity of
resonance I will decrease (a negative NOE). Which of the two mechanisms dominates is
related to the molecular correlation time or rate at which the molecule tumbles in
solution. Fast tumbling = short correlation time = positive NOE.
 Transient NOEs

In standard 1D NOE, one resonance is saturated, and the system must respond to return to
equilibrium by the W0 and W2 cross-relaxation pathways.

Transient NOEs occur when the population is inverted (either selectively in a 1D or all at
once in a 2D) and the system responds again through the dipolar cross-relaxation
pathways during a mixing period. The results are then read with a 90º pulse.

σIS ∝ γ4{[6/(1 + 4 ω02τc2)] - 1}τc/rIS6

σIS = cross relaxation rate

τc = molecular correlation time
rIS = distance between the nuclei
 Transient NOEs build up then decay away

A transient NOE builds up for some period of time but then decays away. The NOESY
(and 1D gradient NOE or gNOE experiment) are transient NOE experiments.

Assuming the NOE growth is linear (which is true when τm is low), then the NOE
enhancement between 2 spins is proportional to the cross-relaxation rate which is
proportional to the distance between the nuclei.

ηA{B} = kσABτ = k' rAB-6 τ

And the distance between A and B can be determined from an NOE of known distance X
and Y:

ηA{B}/ηX{Y}= rAB-6/ rXY-6

 NOESY

NOESY = Nuclear Overhauser Effect SpectroscopY

Essentially 2D NOE experiment where the crosspeaks are NOEs between the two
protons, and the volume of the crosspeak is proportional to the distance between the two
protons.

Since NOEs are transfer through space, the NOESY is the basic experiment used for
determining 3-dimensional structures of molecules with NMR.

NOESY is a technique to measure transient NOEs, which means that the steady-state
NOE enhancement has not been reached; the saturation period is short. Transient NOEs
develop from the population disturbance created by initial inversion of the resonance.
Transient NOEs are weaker than steady-state NOEs, which is a severe drawback of
NOESY versus 1D NOE.

The advantages of NOESY versus 1D:

1) All NOEs in the molecule are observed in one experiment.

2) Although transient NOEs are weaker than steady-state NOEs, signal/noise is usually
better. Although artifacts are a problem in NOESY, they are less important than in 1D
NOE experiments.

Pulse Sequence:

τm = mixing time (ms)

The NOESY pulse sequence = DQF-COSY (except mixing time is ms versus µs for
DQF-COSY)
 Mixing Time

Mixing time is when the magnetization is being transferred from one nucleus to another
by the NOE, while the magnetization is aligned along the z-axis (so NOE is affected by
mostly T1, not T2!). Shorter mixing times allow for only magnetization transfer only
over a short distance, while longer mixing times allow for transfer over longer distances.
However, what is a short mixing time for a small molecule is very long to a large
molecule.

Choosing a mixing time:

Although choosing a mixing time is mostly empirical; there are ways to use theory to
predict what is appropriate. NOEs cannot buildup faster than 1/T1, although they can
buildup slower. However, the competitive longitudinal relaxation (T2) causes loss of
signal. An additional problem is spin diffusion as discussed with respect to 1D NOE,
where there is multiple transfer; this is also a greater problem at longer mixing times, as
the possibility for multiple transfer increases. So what are we left with on choosing a
mixing time? One that is not too short so we see nothing, and not too long so we do not
lose magnetization and do not have spin diffusion. Ideally, you want your mixing times to
be in the region where the buildup of NOE is linear with mixing time. To accomplish this
correctly many mixing times would be chosen and plots like below would be done on
every crosspeak. However, what is normally done: If you are interested in 3D structure,
acquire NOESY spectra for at least 3 mixing times, short, medium and long. If you are
concerned with discriminating distances of 3 Å versus 5 Å then acquire a mixing time on
the short side. If you are concerned with longer distance discrimination, then choose a
mixing time on the long side.
So we are back to what is short, what is long?

Small Molecule (0-2000 MW)

Short - 100-200 ms
Medium- 150-400 ms
Long- 300-800 ms

Small Macromolecule (2000-10000)

Short - 50-100 ms
Medium- 75-150 ms
Long- 150-400 ms

Large Macromolecule (10,000-30000)

Short - 25-40 ms
Medium- 40-70 ms
Long- 70-150 ms
2D NOESY = 2D EXSY (Exchange Spectroscopy)

2D EXSY is used to determine which nuclei are in chemical exchange. Resonances of

nuclei in chemical exchange, will have a crosspeak between them.

Thus, crosspeaks in a NOESY could be from chemical exchange or NOEs.

However, chemical exchange peaks are ALWAYS the same sign as the diagonal, NOEs
are the opposite sign of the diagonal for small molecules (NOEs are the same sign as the
diagonal for macromolecules, cut-off is at ~1000 MW).

Sign of Diagonal Origin of Crosspeak Sign of Crosspeak

Positive NOE Negative
Positive Negative NOE Positive
Chemical Exchange Positive

Hydroxyl protons exchanging with water will have an exchange crosspeak to water in a
2D NOESY experiment.

The exchange peaks result from chemical exchange during the mixing time, and thus
their presence is correlated to the rate of chemical exchange.

Problems with NOESY:

1) Spin Diffusion- multiple transfer of magnetization. Spin diffusion is a greater problem

as the mixing time is increased, so short mixing times are preferable.

2) NOE goes to 0 at ~1000 MW, and is small as 1000 MW is approached from either
side. An alternative approach is ROESY which is essentially independent of molecular
weight.

3) Chemical exchange peaks and NOESY peaks look alike for large molecules. An
alternative approach is ROESY as ROEs are always the opposite sign to the diagonal.

4) Zero-quantum artefacts due to strongly J-coupled nuclei. Resonances that are strongly
J-coupled will have crosspeaks in the NOESY that are much more intense than they
should be for the given distance. These crosspeaks will also be somewhat out of phase
with respect to real NOEs.
 Exchange Peaks in NOESY

The blue peaks are positive crosspeaks in a NOESY of a small molecule.

Positive peaks in a NOESY are either from resonances in chemical exchange or from
slowly tumbling molecules.

The above spectrum is a small molecule in CDCl3, so these are all exchange peaks.

What kind of exchange?

Note the chemical shift of 1.6 ppm.

III. NMR Water Signals

Signals for water occur at different frequencies in 1H NMR spectra depending on the
solvent used. Listed below are the chemical shift
positions of the water signal in several common solvents. Note that H2O is seen in
aprotic solvents, while HOD is seen in protic solvents due
to exchange with the solvent deuteriums.
Solvent
Chemical Shift of H2O (or HOD)
Acetone
2.8
Acetonitrile
2.1
Benzene
0.4
Chloroform
1.6
Dimethyl Sulfoxide
3.3
Methanol
4.8
Methylene Chloride
1.5
Pyridine
4.9
Water (D2O)
4.8
 NOESY

Set up appropriate 1H parameters. Normally should acquire 1D 1H spectrum first.

pw = 1H pulse, should be 90º pulse

tpwr = high power for 1H; should be set according to pw, or is correlated to pw.

nt = number of scans, should be multiple of 16, but 8 is acceptable

ni = number of t1 increments, points in t1 dimension, at least 200, 256 is preferable

np = number of points in t2 dimension, must be even number, 2048 is standard

d1 = relaxation delay; normally should be longer than in gCOSY to decrease t1 noise, and
make ROEs between different types of protons more equivalent, so 2.0-2.5 seconds is
normal, but aromatic (or vinyl or methyl) protons usually need 3.0-3.5 seconds for better
spectra

mix = mixing time; the larger the mixing time the more NOEs will be observed, but the
less difference in intensity by distance. In addition, the mixing time is correlated to the
size of the molecule, small molecule, long mixing time; generally 200-300 ms (so mix =
.2 or mix = .3) for MW ~400-2000 for longer mixing time 100-150 ms for short mixing
time; 300-500 ms for long mixing time for MW < ~400

Steps:

1) Acquire 1 scan of 1H 1D experiment. Set cursors ~0.5 ppm beyond last proton
resonance on both sides of spectrum, type command:
movesw

2) Re-acquire 1H 1D experiment (this is only required for 1H 1D on side of 2D spectrum)

3) Move parameters to another experiment. To move parameters from experiment 1 to

experiment 3 type command:
mp(1,3)

4) Change to experiment 3, type command:

jexp3

5) Set NOESY parameters with setup macro, type command:

NOESY
6) Spin will likely turn off, if it does not, open Acqi window and manually turn it off.
Lock Level should not drop more than ~5 %-10 %. If it does, you should reshim non-spin
shims- X1, Y1, XZ, and YZ, and on higher field (500/600) X2Y2 and XY

7) Set the number of scans (nt) to appropriate value, generally 8 (it is somewhat
preferable for nt to be a multiple of 16 but nt = 8 is acceptable; nt might need to be
increased for sufficient signal-to-noise), type:
nt=8

8) Set the number of points in t1 dimension (second dimension) (ni), generally 200-400
are suggested, the greater the points, the better the resolution, and the less t1 noise
observed, type:
ni=256

9) Set d1 time, normally should be longer than in gCOSY to decrease t1 noise, and make
NOEs between different types of protons more equivalent, so 2.0-2.5 seconds is normal,
but aromatic (or vinyl or methyl) protons usually need 3.0-3.5 seconds for better spectra

10) Set the mixing time (parameter = mix); the larger the mixing time the more NOEs
will be observed, but the less difference in intensity by distance. In addition, the mixing
time is correlated to the size of the molecule, small molecule, long mixing time; generally
200-300 ms (so mix = .2 or mix = .3) for MW ~400-2000 for longer mixing time 100-150
ms for short mixing time; 300-500 ms for long mixing time for MW < ~400

11) Start acquisition, type command:

go

12) At end of experiment, set appropriate weighting functions and linear prediction
parameters, type commands:
setLP1
sqcosine

13) If fn parameter now equals 4096, processing will be slow, then type:
fn = np
fn1 = fn

14) Process 2D type command:

wft2da

15) Switch f1 and f2 axes (make f2 the x-axis), type

trace = 'f2'

16) Display as contour plot, type command:

dpn10
17) If spectrum has yellow streaks and purple streaks horizontally across the spectrum,
then the spectrum should be phased; set cursor on peak, type:
ds
then manually phase, be careful to only click once on spectrum during phasing, only rp
(or rp1 phase correct should be changed)
redraw 2D type:
dpn10
If spectrum still has streaks in horizontal direction more phasing is required including lp
(or lp1) phase correct. If spectrum has yellow/purple streaks in vertical direction, then the
other dimension needs to be phased, so switch the axes, type:
trace = 'f1' (or 'f2', depending upon which axis is currently the x-axis, then type:
dpn10

18) Spectrum should be appropriately referenced already, but you should confirm this; if
necessary re-reference by putting cursor on appropriate diagonal peak and type (assuming
CDCl3):
rl(7.26p)
rl1(7.26p)
dp10

19) Adjust vertical scale with vs +20% and vs -20% menu buttons or with middle mouse
button or manually changing the parameter, vs2d, so vs2d = 100 (the lower the number,
the less noise displayed), then redraw 2D with dp10 command

20) Print with 1D on side with 1D in experiment 1:

plcosy(10,1.2,1)

21) or Print with full rectangle, type:

full
dp10
pcon(10,1.2) page

22) Save data, type:

svf('filename')
 GOESY

Cyclenoe is a steady-state saturation NOE experiment for Inova spectrometers; goesy is a

1D NOE technique that detects transient NOEs (as in NOESY) with gradients.

The advantage of the goesy experiment is that the NOEs observed are cleaner, less
antiphase peaks, less multiple transfer events (indirect NOEs) are observed. The signal-
to-noise is generally lower than the steady-state experiment as transient NOEs are weaker
in intensity than steady-state NOEs.

The main difference then is goesy requires a mixing time like the NOESY experiment.

the setup macro for goesy is: goesy

Normally, a 1D would be acquired, then the sw, pw, tof, solvent, tpwr would be set
appropriately. Typing goesy would set the remainder of appropriate parameters to acquire
the truncated NOE experiment.

Type wft.

Set parameters ref_pw90 and ref_pwr to reference values for the 90º pulse at a given
power. Ideally, use a medium power value the ref_pwr and ref_pw90, such as the values
for the slpw and slpwr for a TOCSY experiment.

Then, put cursors outside peak that you want to measure NOEs to. Type setshape. The
setshape macro optimizes the shaped pulse that will be used to invert resonance.

The only other relevant parameters are mix and nt.

nt = multiple of 16

mix = mixing time; should be set as in NOESY.

 ROESY

ROESY = ROtating frame Overhauser Effect Spectroscopy

ROESY is basically the same as NOESY in the rotating frame with respect to what is
observed, but ROESY depends upon the buildup of ROEs. ROEs are ALWAYS opposite
in sign to the diagonal (which means that ROEs are always positive). Note that chemical
exchange peaks are also observed in ROESY, and as in NOESY, same sign as the
diagonal. Thus, in large molecules, where NOEs are the same sign as the diagonal,
ROESY distinguishes between through-space interactions and chemical exchange.

ROE is cross relaxation in the transverse plane (not the longitudinal axis). The ROE is
created by spin-locking the magnetization in the x-y plane by continuous pulsing that
creates no torque on the magnetization; the pulses lock the magnetization along the y-
axis. Spin relaxation is governed by T1ρ.

The spin-lock is on the order of kHz, which means that the product of the frequency and
the correlation time << 1; thus, for all values of correlation time, the ROE is positive.

Cross relaxation for an ROE (σ) is proportional to:

σ ∝ γ4{3/(1 + ω02 τc2) + 2} τc/rIS6

For a small molecule (ω0τc << 1),

σ ∝ 5γ4τc/rIS6
which is the same as the equation for a transient NOE.

For a large molecule (ω0τc >> 1),

σROE ∝ 2γ4τc/rIS6

σNOE ∝ -γ4τc/rIS6

Thus, σROE = -2σNOE

Thus, ROE buildup is twice as fast as that of NOEs for large molecules.
Comparison of ROE buildup (ROESY) versus transient NOE buildup (NOESY)

Pulse Sequence of ROESY is essentially the same as TOCSY:

Comparison of Transient NOE (NOESY) with Rotating Frame NOE (ROESY)

Advantages of ROESY over NOESY:

1) Unaffected by molecular weight, as the experiment is done in the rotating frame; ROE
is not 0, even when NOE = 0.

2) Chemical exchange peaks always have opposite sign to ROEs.

3) Spin diffusion is less pronounced, and spin diffusion peaks are opposite in sign to
ROEs.

4) ROE build up is twice as fast that of the NOE for large molecules, mixing time (or
spin-lock time) is generally shorter. Mixing time for small molecules is generally about
the same for ROESY and NOESY.
Disadvantages of ROESY:

1) TOCSY artifacts, as the pulse sequences are essentially the same, there can be TOCSY
peaks in a ROESY (and of course ROESY peaks in a TOCSY). TOCSY peaks should be
opposite in sign to ROEs though, although combination ROESY-TOCSY and TOCSY-
ROESY peaks are also possible, and these would be out of phase with respect to the
ROEs. The TOCSY peaks can also cancel the ROESY peaks. These artifacts are
somewhat suppressed by recent versions of the two pulse sequences, but still a potential
problem.

Sign of Diagonal Origin of Crosspeak Sign of Crosspeak

Direct ROE Negative
Indirect ROE (3-spin effect) Positive (weak)
TOCSY Positive
Positive TOCSY-ROE Negative (false ROE)
ROE-TOCSY Negative (false ROE)
Chemical Exchange Positive
COSY-type Antiphase/mixed phase

2) ROEs have a theoretical limit of 67% of the corresponding NOE for large molecules,
so signal-to-noise is less in a ROESY than NOESY for large molecules, except in the
vicinity of where the NOE = 0.

3) Sample Heating- Since there is constant pulsing during the mixing time, sample
heating can be a problem. Normally lower field spin lock is used, ~3000-6000 Hz
(40-80 µs).

4) Lower field spin-lock means that it is hard to cover the entire frequency range if there
is large chemical shift dispersion, especially on high field instruments (3000 Hz only
covers ~ 4 ppm on an 800 MHz instrument). Thus, there is crosspeak attenuation for
peaks far from the center of the spectrum.

5) COSY type crosspeaks. This arises from the low power pulse acts like the last pulse on
a COSY to cause coherence between J-coupled spins. Mostly, this is not near as big of a
problem as the TOCSY artifacts (or the DQF-COSY artifacts in the NOESY as well).
 ROESY

Set up appropriate 1H parameters. Normally should acquire 1D 1H spectrum first.

pw = 1H pulse, should be 90º pulse

tpwr = high power for 1H; should be set according to pw, or is correlated to pw.

nt = number of scans, should be multiple of 16, but 8 is acceptable

ni = number of t1 increments, points in t1 dimension, at least 200, 256 is preferable

np = number of points in t2 dimension, must be even number, 2048 is standard

slpw = spin lock pulse width, must be a 90º pulse with a pulse width of ~30-35 µs

slpwr = power of spin lock pulse, cannot be more than 49 units, preferably ~43 units,
must correlate to slpw

d1 = relaxation delay; normally should be longer than in gCOSY to decrease t1 noise, and
make ROEs between different types of protons more equivalent, so 2.0-2.5 seconds is
normal, but aromatic (or vinyl or methyl) protons usually need 3.0-3.5 seconds for better
spectra

mix = mixing time; the larger the mixing time the more ROEs will be observed, but the
less difference in intensity by distance. In addition, the mixing time is correlated to the
size of the molecule, small molecule, long mixing time; generally 200-300 ms (so mix =
.2 or mix = .3) for MW ~400-2000 for longer mixing time 100-150 ms for short mixing
time; 300-500 ms for long mixing time for MW < ~400

Steps

1) Acquire 1 scan of 1H 1D experiment. Set cursors ~0.5 ppm beyond last proton
resonance on both sides of spectrum, type command:
movesw

2) Re-acquire 1H 1D experiment (this is only required for 1H 1D on side of 2D spectrum)

3) Move parameters to another experiment. To move parameters from experiment 1 to

experiment 3 type command:
mp(1,3)

4) Change to experiment 3, type command:

jexp3

5) Set ROESY parameters with setup macro, type command:

ROESY

6) Spin will likely turn off, if it does not, open Acqi window and manually turn it off.
Lock Level should not drop more than ~5 %-10 %. If it does, you should reshim non-spin
shims- X1, Y1, XZ, and YZ, and on higher field (500/600) X2Y2 and XY

7) Set the number of scans (nt) to appropriate value, generally 8 (it is somewhat
preferable for nt to be a multiple of 16 but nt = 8 is acceptable; nt might need to be
increased for sufficient signal-to-noise), type:
nt=8

8) Set the number of points in t1 dimension (second dimension) (ni), generally 200-400
are suggested, the greater the points, the better the resolution, and the less t1 noise
observed, type:
ni=256

9) Set d1 time, normally should be longer than in gCOSY to decrease t1 noise, and make
ROEs between different types of protons more equivalent, so 2.0-2.5 seconds is normal,
but aromatic (or vinyl or methyl) protons usually need 3.0-3.5 seconds for better spectra

10) Set the mixing time (parameter = mix); the larger the mixing time the more ROEs
will be observed, but the less difference in intensity by distance. In addition, the mixing
time is correlated to the size of the molecule, small molecule, long mixing time; generally
200-300 ms (so mix = .2 or mix = .3) for MW ~400-2000 for longer mixing time 100-150
ms for short mixing time; 300-500 ms for long mixing time for MW < ~400

11) Start acquisition, type command:

go

12) At end of experiment, set appropriate weighting functions and linear prediction
parameters, type commands:
setLP1
sb = -at
sbs = sb
sb1 = -ni/sw1
sbs1 = sb1

13) If fn parameter now equals 4096, processing will be slow, then type:
fn = np
fn1 = fn

14) Process 2D type command:

wft2da

15) Switch f1 and f2 axes (make f2 the x-axis), type

trace = 'f2'
16) Display as contour plot, type command:
dpn10

17) If spectrum has yellow streaks and purple streaks horizontally across the spectrum,
then the spectrum should be phased; set cursor on peak, type:
ds
then manually phase, be careful to only click once on spectrum during phasing, only rp
(or rp1 phase correct should be changed)
redraw 2D type:
dpn10
If spectrum still has streaks in horizontal direction more phasing is required including lp
(or lp1) phase correct. If spectrum has yellow/purple streaks in vertical direction, then the
other dimension needs to be phased, so switch the axes, type:
trace = 'f1' (or 'f2', depending upon which axis is currently the x-axis, then type:
dpn10

18) Spectrum should be appropriately referenced already, but you should confirm this; if
necessary re-reference by putting cursor on appropriate diagonal peak and type (assuming
CDCl3):
rl(7.26p)
rl1(7.26p)
dp10

19) Adjust vertical scale with vs +20% and vs -20% menu buttons or with middle mouse
button or manually changing the parameter, vs2d, so vs2d = 100 (the lower the number,
the less noise displayed), then redraw 2D with dp10 command

20) Print with 1D on side with 1D in experiment 1:

plcosy(10,1.2,1)

21) or Print with full rectangle, type:

full
dp10
pcon(10,1.2) page

22) Save data, type:

svf('filename')
 GOESYTR

goesytr does a 1D transient ROE experiment

the setup macro for goesytr is: goesytr

Normally, a 1D would be acquired, then the sw, pw, tof, solvent, tpwr would be set
appropriately. Typing goesytr would set the remainder of appropriate parameters to
acquire the truncated NOE experiment.

Type wft.

Set parameters ref_pw90 and ref_pwr to reference values for the 90º pulse at a given
power. Ideally, use a medium power value the ref_pwr and ref_pw90, such as the values
for the slpw and slpwr for a TOCSY experiment.

Then, put cursors outside peak that you want to measure NOEs to. Type setshape. The
setshape macro optimizes the shaped pulse that will be used to invert resonance.

The only other relevant parameters are:

nt = multiple of 16

mix = mixing time; should be set as in NOESY.

slpw = spin lock pulse at power slpwr as in ROESY

slpwr = spin lock power for pulse slpw as in ROESY