NMR Trainingfor Advanced Users

Huaping

Aug 18, 2008

Two Insider Scoops

• Receiving efficiency– conceptually, it is similar to extinction coefficient in UV spectrospcopy; it

characterizes how efficient a unit magnetization can produce a signal by a given NMR receiver

– NMR signal size is proportional to receiving efficiency– Receiving efficiency can be pre-calibrated as a function of 90° degree

pulse length– receiving efficiency is the same for all nuclei of the same type

(indifferent to chemical shifts) in the same sample

• Solvent signal offers a universal and robust concentration internal standard– Normalized NMR signal size is strictly proportional to concentration for a

given sample, regardless how concentrated or dilute the sample is– Unit magnetization generates the same amount of total NMR response,

which is indifferent to chemical shift or line-shape

Outlines

• Basic preparations for NMR: safety, sample, lock, shim and tune

• Understanding NMR: excitation and observation

• RF pulse calibration

• NMR observables:– Chemical shift, scalar couplings, NOE and relaxations

• Introduction to basic 2D's

• Simulations for spin systems, pulses and sequences

Safety• Personal safety

– Cryogens: do not lean on or push magnets– Cryoprobes: avoid contact with transfer line– Magnetic and RF hazards

• Instrument safety– Know the limits of instruments– Probe limits: avoid excessive long decoupling and long hard pulses or

their equivalents– Be conservative– Double check pulse program and parameters for any non-standard new

experiment

• Data Safety– Back up data promptly and regularly– Data processing or manipulation has no impact on the raw (FID) data– Do not change parameters after data are acquired

Sample• Rule #1: for Bruker NMR spectrometers, the

NMR tube insert cannot exceed max depth (19mm or 20mm) from the center of the RF coil

– Longer insert than recommended may present problems for the probe, as well as cause frictions during spinning

– Varian is more flexible in allowing longer insert

• Rule #2: center of NMR sample should be as close as possible to the center of RF coil.

– Normal sample needs to about 500 ul or slight more

– Too much solvent is a waste!– Too little solvent may make shim difficult, but

it does work!

• 10% deuterated solvent is sufficient for locking

20mm

18mm Coil CenterRF coil

Samples of smaller volumes

• Follow rule # 1 and then rule #2

• Shimming might be challenging due to air/glass and air/solution interfaces

• Consider Shigemi tubes

• Be careful with spinning– Non-spinning is recommended

for volume ~ 300 ul or less

300ul 400ul 500ul

Sensitivity for smaller volumes

• Volume less than 300 ul may not offer additionally sensitivity improvement over that achieved by 300 ul, if the total amount of analyte is constant

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

0 200 400 600 800 1000

volume (ul)

rela

tiv

e s

en

sti

vit

y

Tune and match the probe

• Only higher fields (500, 600 and 800 HMz) in our facility need tuning

• Most of the time only proton channel requires tuning

• Drx500-2 with BBO needs special attention– Proton always needs tuning– BB (used for 13C or 31P etc) channel

needs tuning, by first dialing the numbers to the pre-set values

Frequency

tunematch

RF

ref

lect

ion Carrier frequency

Significance of tuning/matching

• Shorter 90 degree pulse – More efficient use of RF power

• Protects transmitter

– More uniform excitation in high power

• Better sensitivity– Reciprocity: if excitation is inefficient,

then detection is equally inefficient

• Potentially quantitative:– The product of NMR signal size is

inversely proportional to the 90 degree pulse length

0.5

0.7

0.9

1.1

8 10 12 14

90 degree pulse length (ms)

NM

R s

ign

al s

ize

Recognizing Bruker probe types

Dials for broadband (BB) tuning/matching

BBO probe on drx500-2 TXI probe

magnet

BB Dialing stick

Tabulated values for BB tuning/matching

1H tuning/matching rods are labeled as yellow

side view

bottom view

side view

Do not touch those!

Lock

• Lock depends on shim: bad shim makes bad lock– Initialize shim by reading a set of good shims (i.e. rsh shims.txi)– Inheriting a shim set from previous users may present difficulties – Unusual samples (esp. small volumes) may need significant z1/z2

adjustments

• Use “lock_solvent” or “lock” command – The default (bruker) chemical shift may appear as dramatically changed

if the spectrometer assumes another solvent

• Avoid excessive lock power – Lock signal may go up and down if lock power is too high due to

saturation of deuterium signal– Apply sufficient lock power and gain so that lock does not drift to

another resonance (this may happen by autolock if multiple deuterium signals exist)

Shim• The goal of shim is to make the total magnetic field within the active

volume homogeneous (preferably <1Hz).Total magnetic field = static field (superconductor) + cryoshim (factory set) + RT shim (user adjust)

• Shim can be done either manually or by gradient, which can be very efficient and consistent if done properly

• Sample spinning may improve shim– However, spinning-side band appear

• Recommendation:– Start from a known good shim set (by rsh on bruker or rts on varian).– Do not inherit shims from other users unless you know they’re good– Non-spinning and higher order (spinning) shims should not change

dramatically from sample to sample for most applications

Lock: lock gainrecommended not recommended

higher lock gain

may easily lose lock;change in lock level (during shimming) is less visible

Lower lock level due to lower lock gain

Lock: avoid high lock powerGood lock Bad lock

unstable lock and lower level

Lock power okay

Lock power too high

Evaluate shims• Look for a sharp peak

– No clear distortion– Full width at half height should be about 1 Hz or less for small molecules– Small (1% or smaller) or free of spinning side-bands

• Check if peak distortions are individual or universal

• Make sure that phasing is not causing peak distortions

• Maximize the lock level– Higher lock level => better shim

• Lock level does not drop significantly when spinning is turned off

Shim by line-shape

Plot made by G. Pearson, U. Iowa, 1991

z4 too small z4 too big make z4 smaller first

Understanding NMR• Modern NMR spectrum is an

emission spectrum

• Equilibrium state– Magnetization is along +z axis– It is desired to have the largest

+z magnetization prior to excitation

• Excitation by a RF pulse– A projection of magnetization is

made on xy plane– It is desired to have the largest

xy plane project for observation

• Observation– Precession of the projected xy-

plane magnetization

RF pulses• RF pulse manipulates spins

– Important in excitation and decoupling– Defined by length, power and shape

• RF power is expressed in decibels– Bruker

• Power range: typically 0db (high power) to 120db (low power)

– Varian: • Coarse power: typically 60db (high) to 0db (low); 1 db increment;

absolute• Fine power: 4095 (high) to 0 (low); default is 4095; relative

– e.g. 54.5db can be roughly achieved through setting coarse power to 55 and fine power to 3854

coarse attenuator fine attenuatorin out

RF pulse calibration

• Hard pulse (high power pulse) can be calibrated directly or indirectly

• For best calibrations, pulses need to be on resonance (know the chemical shift or resonance frequency!)

• Soft or shaped pulsed can be first calculated and then fine-tuned to optimum – Shapetool (by Bruker) or Pbox (by Varian) can be used for

calculation and simulation– Be aware of possible minute phase shift (several degrees for

soft pulses), which can be critical in water flip back or watergate

Proton pulse calibration• Most hard (highest power) 90° pulses are typically from 5 us to

20 us.• Direct observation for high power proton pulse calibration (or

even for heteronuclei if sensitivity is sufficient) – 360° method (not quite sensitive to radiation damping or relaxation)– 180° method

90º

180º

360º

First pulse with 2 us; 2 us increment

90º

180º

270º

360º

270º

450º

NMR observables

• Chemical shifts– expressed in ppm

• Scalar couplings– expressed in Hz– 2D or nD bond correlations

• NOEs / relaxation / line-shapes

• Peak size– Potentially useful in quantitative analysis

Chemical shifts

• Reflects chemical environment:– Ring current effect

• Outside of ring: high ppm

• Inside: low ppm

– Effect of electron withdrawing groups

• Donating: low ppm• Withdrawing: high ppm

Chemical shifts of solvents/impurities

Gottlieb et al. JOC 1997

Example: aliasing

(from arx300) aliased from 0 ppm with phase distortion,because the peak is out of the “detection window”

sw=16ppm

okay

aliased

• Oversampled proton spectrum on higher fields (500 – 800 MHz) does not have the aliasing issue: peaks outside of sw will disappear

Spectral aliasing (cont’d)

• In direct observe dimension, spectral aliasing is generally avoided by either increasing spectral width (sw) or moving center frequency (sfo1)

• Sometimes the indirect detection dimension (in nD spectrum) may intentionally adopt aliasing to improve resolution in that dimension

Scalar coupling

AB system“roofing”

JABJAB

A B

1

1:1

1:2:1

1:3:3:1

Scalar coupling: simulation helps!

Observed (300 MHz)

simulated

P O

Ar

O

O

O

R

H

Ha

b12Hz

8Hz

8Hz

Ha and Hb are not exactly equivalent, withchemical shift difference of 0.025ppm

These are not impurities!

pro-chiral!

Example:satellites and spinning side-bands

TMS

6.6 Hz; 29Si satellites; 2.3% each

spinning sideband; 20 Hz from center

120 Hz; 13C satellites; 0.55% each

-4

-3

-2

-1

0

1

2

3

0 2 4 6 8 10

w htc

NO

E +

1H

P

C

N

Dipolar coupling: NOE

• NOE depends on correlation time (molecule size) and resonance frequency

• NOE does not always enhance the observed signal

15N

1H

31P

13C

Molecule size

Temperature

NOE implication in Quantification

• The observed nucleus should be free of interference from other nuclei

• Pre-saturation in aqueous samples may not be appropriate for accurate quantification– Small molecules tend to gain signal size due to

positive NOE from saturated water– Large molecules tend to lose signal size due to spin

diffusion

Relaxation

• T1 relaxation allows magnetization to recover back to +z axis– Nuclei with larger gyromagnetic ratios (resonance

frequencies) tend to relax faster• 1H: 0.1 – 10 s (proteins have short T1’s)• 13C, 15N, 31P: much longer than 1H

– Nuclei in a proton rich environment tend to relax faster

• T2 relaxation contributes to the observed resonance line-shape– T2~T1 for small molecules– Line-shape offers an estimate of T2

Line-shape

• Full Width at Half Maximum is 1/(T2*) Hz, with T2

* as apparent spin lattice relaxation time

• Magnetic inhomogeneity (shim) can increase FWHM (2) or distort the line-shape (reduce T2*)

• T1 > T2 > T2*

• Small molecules– 1H: T1 ~ T2 in the order of seconds– 13C: seconds to tens of seconds; even longer if no

proton attached (CO and quaternary)

• Large molecules– 1H: T1 ~ T2 hundreds of mini-seconds or shorter– 13C: seconds or sub-seconds

-20 -15 -10 -5 0 5 10 15 20

off set (Hz)

FWHM (2)

Lorentzian: A(w)= / (2 + (w-w)2)2=1/(T2

*)

multiple chemical environments:chemical or conformational exchange

• Fundamentally, chemical shift reflects chemical environment surrounding a nucleus’

• Multiple chemical environments may alter chemical shift or even cause significant peak broadening

Fast exchange

slow exchange

Jin, Phy. Chem. Chem. Phys. (1999)

(N)H line-shape: influence of relaxation and scalar coupling

JNH ~65 Hz

Slow 14N relaxation (compared to JNH)

medium14N relxation

this might be the very reason why CHCl3 proton appears as a singlet though JH-35Cl and JH-37Cl exist

In addition to chemical exchange, (N)H proton line-shape is also influenced by the coupled nucleus 14N

Fast relaxation

Improving Sensitivity

• More scans in a given amount of time

• Use Ernst angle for excitation:

cos = exp(-Tc/T1)

• Increase concentration with less solvent / salt

sensitivityTc/T1

Pulse angle (degrees)

Improving sensitivity

• Receiver gain needs to be maximized which requires good water suppression

• Avoiding excessive large receiver gain (for signal clipping)

• Excessive acquisition time end up with spending time collecting noise and down-grade signal to noise ratio

sensitivity vs receiver gain(arx300; chloform signal)

0 100 200 300 400 500receiver gain

S/N

Missing a carbonyl carbonpresumably due to insufficient

relaxation

About 5 mg in CD3OD. 2800 scans (~4 hrs)

O OR

NHOH

R'

OH

O

? Missing a carbonyl

Solution: Use H2O

Why it works:

• Carbonyl 13C is reduced due to presence of a proton rich environment in H2O.

• Potential intra-molecular hydrogen bond is weakened or broken, and decoupled from ring movement

In H2O:D2O (1:1). 1400 scans (~2 hrs).

O OR

NHOH

R'

OH

O

Direct observe: 31P, 13C or 15N

• 19F, 31P and 13C can be observed directly on all PINMRF 300 and 400 MHz instruments (please follow local PINMRF instructions)

• 13C can be observed on higher fields (500 MHz and above), without any cable change

• Drx500-2 with BBO probe offers higher sensitivity for 31P, 13C, 15N and most other heteronuclei (19F excluded)– Observed nucleus needs to be cabled to x-broadband pre-

amplifier– BBO tuning is needed for both proton and observed nucleus– Double check filters if re-cabled

Direct observe: 31P, 13C or 15N

• Satellite peaks can frequently be indirectly observed in proton spectrum (so that we know the less sensitive heteronuclei are there to be observed directly!)

• Decoupling of proton may improve signal by– Sharper peaks– NOE– Proton channel has to be tuned!

1D acquisition for very long hours

• helpful– Split long experiments into smaller blocks and save

data regularly (multiple data can always be summed if needed)

– Dissolve the compound in water (H2O) might be helpful (shorter relaxation time)

– Lower sample temperature may help

• Not helpful– Save several days’ data into one single FID– Use 300 ul or less volatile solvent

From 1D to 2D

1D

time domain frequency domain

w1

FT

t2

2D

time domains

w2

FT(t2)

t2

t1t1

FT(t1)

w2

w1

frequency domains

2D NMR• Correlate resonances through bond or space

– COSY: coupling• Magnitude mode recommended. • 1 mg or less will do• Minutes to a couple of hours

– TOCSY: coupling network• ~ 70 ms mixing time• 1 mg or less will do• An hour or longer

– NOESY / ROESY: distance / NOE• Mixing time ranging from less than 100 ms (proteins) to 500 ms (small molecules)• 1 mg or more• Hours or longer

– HSQC/HMQC: proton correlation to X, typically through one-bond scalar couplings (two or three bond correlation possible)

• 1mg or less will do• An hour or longer

– HMBC: proton correlation to X, through multiple bond scalar couplings• 1 mg or more• Hours or longer

2D NMR• Resolve overlapping peaks

– Resolution is provided largely through the indirect dimension

– No need to have highest resolution in the direct detected dimension

• Limit direct acquisition time to 100ms or less if heteronuclear decoupling is turned on

• Lower decoupling power if longer acquisition time is needed

– Change in experimental conditions may help

2D NMR essentials: acquisition• Proton tuning and matching• Calibration of proton (90 degree) pulse length

– Standard pulse lengths can be used if the solution is not highly ionic (< 50 mM NaCl equivalent)– All proton pulses are likely getting longer if the solution is ionic and/or the probe is not tuned

• Modest receiver gain– rg about half of what rga gives or less

• Carrier frequency (center of spectrum in Hz) and SW (sweep width) in both dimensions (avoid aliasing unless intended to)

• Number of scans (NS)– The pulse program recommends NS (a integer times 1, 2, 4, 8 or 16)– Needs some dummy scans, especially with decoupling / tocsy

• Number of increments in the indirect dimension (td1)– Larger td1 improves resolution in the indirect dimension– Rarely exceeds 512 (except occasionally in COSY

• Detection method in the indirect dimension– Determined by the pulse program– Typically is either states (and/or TPPI) or echo-antiecho

• Acquisition time (aq) less than 100 ms with decoupling• Modest gradients (cannot be more than the full power of 100% and typically less than 2 ms in

duration) • Go through the pulse program if you really care

2D processing• Window functions

– Allow FID approach zero at the end of the acquisition time– Sine bell functions with some shifts are recommended most of the time

• Zero filling– Typically double data points in each dimension

• Phasing– Indirect dimension zeroth and 1st order corrections are recommended in the

pulprogram. If not, use 0 for both– Direct dimension first order phase is rarely more than 50 degrees. Zeroth order

can be anywhere from 0 to 360 degrees– Phase in the 2D mode to best appearance

• Referencing– Can be done by picking a known resonance in the spectrum– Or referenced by (external) protons

INEPT

HSQC: a Block Diagram

acq

dec

H

X

J

t1/2 t1/2

• Magnetization transfer pathway:F1(H) -> F2(X) -> F2(X,t1) -> F1(H) -> F1(H,t2)

90 180

180 90

States: =x and =y are acquired for same t1 and treated as a complex pair in Fourier transform. No need to change receiver phase

TPPI: =x, y, -x and –y are acquired sequentially in t1, and receiver phase is incremented too. Real Fourier transform.

J J J

HMQC or HSQC

Magnitude HMQC (9 mins)Easy set up and slightly higher sensitivity

Phase sensitive HSQC (18 mins)Better resolution

adapted from acornnmr.com

codeine

HMQC and HSQC comparison

• HMQC– Fewer pulses– More tolerant to pulse

mis-calibrations

– Allows homonuclear (proton) coupling in the indirect dimension

• HSQC– More pulses– Less tolerant to pulse

mis-calibrations

– No homonuclear (proton) coupling in the indirect dimension

Data Presentation

• Processed data can be readily viewed, manipulated and printed by xwinplot (wysiwyg)

• Xwinplot can readily output .png, .jpg or .pdf files for publications or presentations

• Files can be transferred through secure ftp

Pulse sequence:the heart and soul of NMR

;zggpwg;this is a bruker sequenceprosol relations=<triple>#include <Avance.incl>#include <Grad.incl>"d12=20u"1 ze2 30m d1 10u pl1:f1 p1 ph1 50u UNBLKGRAD p16:gp1 d16 pl0:f1 (p11:sp1 ph2:r):f1 4u d12 pl1:f1 (p2 ph3) 4u d12 pl0:f1 (p11:sp1 ph2:r):f1 46u p16:gp1 d16 4u BLKGRAD go=2 ph31 30m mc #0 to 2 F0(zd)exitph1=0 2ph2=0 0 1 1 2 2 3 3 ph3=2 2 3 3 0 0 1 1ph31=0 2 2 0;comments for parameters…

Delay

define f1 power level

Phases

90° pulse on f1

Gradient pulse

Shaped 90° pulse

Acq. and go to label 2

1H

G

90x

90-x180x

90-x

On-res: dephased by two gradientsOff-res: refocused by two gradients

Write to disc. And go to label 2

Delay only; be very careful with critical command in a

labeled line label

Where Things are: Bruker File Structure

• User NMR data /u/data/username/nmr• Pulse programs /u/exp/stan/nmr/lists/pp • Gradient programs /u/exp/stan/nmr/lists/gp• Shaped pulses

/u/exp/stan/nmr/lists/wave• decoupling /u/exp/stan/nmr/lists/cpd• Frequency(f1) lists /u/exp/stan/nmr/lists/f1• Parameter sets /u/exp/stan/nmr/par• Shim sets /u/exp/stan/nmr/lists/bsms• Macros /u/exp/stan/nmr/mac

Gradients

• Homospoil gradients– Size of duration may not matter much– Stronger ones tend to clean up unwanted

magnetization better

• Gradient echoes:– Exact ratios between multiple gradients must

follow– Diffusion loss must be considered for small

molecules, especially during long echoes• Log of signal size is proportional to -2g22D

Simulations

• Can be easily performed for pulses, spin-systems or pulse sequences

• Save experimental time

• Enhance our understanding of NMR

• Most frequently used for shaped pulses

Shaped Pulse: What and Why

• What– Narrow sense: amplitude modulation only, while

phase is constant– Broad sense: amplitude and phase modulation

• Why– To achieve perturbation over a certain frequency

range (uniform and selective)• Narrow bandwidth: shaped pulse. e.g. Gaussian• Wide bandwidth: adiabatic pulse

How is Shaped Pulse Different• Composite pulse is typically a block of square pulses with constant

phases– Pulse integration does not correlate with pulse angle– Pulse calibration come from individual component

• Adiabatic pulse sweeps frequency (phase has strong time dependence)– Pulse integration does not correlate with pulse angle– Pulse calibration depends on sweep range, and somewhat on

adiabaticity too

• Simple shaped pulse can be calibrated by integration– Caveat: a 180° pulse is not necessarily twice of a 90° pulse– Some shaped pulses are good for 180° inversions (z -> -z) while others

are good for 90° excitations (z -> x/y)

Shaped Pulse Examples

• Square pulse: simplest shaped pulse; good for simple hard excitation

• Gaussian and Sinc: good selectivity; for proton

• Gaussian cascade: G4, G3, Q5 and Q3; for carbon– G4 for excitation– G3 for inversion – Q5 for 90°– Q3 for 180°

Gauss

Sinc1

G4

G4: four Gaussian lobes

Choosing Shaped Pulses• Define the goal

– excitation, inversion or refocusing– length or power level

• Rule of thumb: bandwidth is ~ 1/P360 or RF strength (for square pulses)– shape

• Power requirement– peak power may not exceed certain level

• Length requirement– Be aware of probe limit on length in case of high power– While longer pulses tend to have better selectivity, relaxation / scalar coupling

may limit pulse length

• Run pulse simulation and calculation– Bandwidth needs to be first satisfied– Simulated frequency profile is to have top-hat behavior– Phase needs to be linear in the region of interest

Shaped Pulse Calculation

• Rule of thumb:

– 6db change in power results two fold change in pulse length

dB = 20 log (P90/P90ref)

– e.g. 10us @0db => 20us @6db for the sample pulse angle

• For a shaped pulse with a imperfect linear amplifier,

dB = 20 log (P90*shape_integ/P90hard*comp_ratio)

Modern spectrometers have comp_ratio close to 1

• Adiabatic pulses require different treatments

Example: Setting up a Sinc Pulse

• Within xwinnmr, launch shape tool by typing “stdisp” or from menu

• Within shape tool, choose shapes -> sinc. Change lobe number to 1 and click “OK”

• On the left is the amplitude profile (sinc shape) and (constant) phase is shown on the right

1 means one sinc lobe

Example: a Sinc Pulse (cont’d)

• Within shape tool, choose analyze -> integrate pulse. Make necessary updates. In this particular case, we assume the reference is 9.5 us @1.5db and you wish to calculate for 1000us 90 degree pulse. Then click OK

• The power level is calculated as 35.8db compared with the reference. Click “seen”

• If satisfied, you can save this shaped pulse under /u/exp/nmr/stan/lists/wave/.

• Go back to xwinnmr->ased, and update the sinc1 shaped pulse as pulse length of 1ms, and power level to be 35.8 + 1.5 (since reference 9.5us is @ 1.5db) = 37.3db

• If needed, the shaped pulse power can be fine tuned by gs, or a careful calibration

Pulse Simulation

• Within shape tool, choose analyze -> simulate. Update the length as 1000us and rotation angle as 90 (for sinc1 we just set up). Click “OK”.

• A new Bloch module will show default (x,y) profile for excitation. Click on z to view z profile.

z

Pulse Simulation (cont’d)

• If you decide that the starting magnetization is x, you can click (in Bloch module) “calculate”->”excitation profile”. Change initial Mx to 1 and Mz to 0. Click “OK” and then the excitation profile will be updated.

• If you wish to examine trajectory (how a magnetization at a given frequency responds to the sinc1 pulse), you can click “time evolution”, and update initial values etc (may not allow too many steps). Click “OK”.

Advanced NMR Training

10-12 noon, 8/18 (Monday)

BROWN 3106

Contact Huaping Mo for details

Demo

• Sample preparation; Shigemi tube

• Lock and shim

• Tune and match

• Calibration of 90 degree pulse

• Calculation / simulation of pulses

• Set up 2D: COSY and HSQC

• Processing and present data (xwinplot)