Processing and Analysis of NMR Spectra

1
Processing and Analysis of NMR Spectra

• Advanced Computing in NMR Spectroscopy
• Florence, September 9-14, 2001

Dr. Christian Fischer
2
TOC

• Processing
• Fourier Transformation (FT), Data acquisition
• Window multiplications, Zerofilling
• Phasing, Baseline correction
• Postprocessing
• Mean row calculations
• Symmetrization
• Analysis
• Multiplett analysis
• Volume - distance calculation
• Backcalculation
• J pattern search
• Match Transfer
• Tripple Resonance 3D

3
Fourier transform
or
4
Fourier pairs of important functions
Rectangle functions
5
Fourier pairs of important functions
Gauss function
Gauss function
6
Fourier pairs of important functions
lorentzian
exponential
7
Data acquisition sampling
Data are sampled at discrete time intervals and
for a finite duration
Sampling interval
Folding will occur, if frequencies above
8
Data acquisition sampling
9
Data acquisition sampling
10
Data acqu. quadrature detection
Discrimination of /- frequencies
11
Relaxation
FID
12
Convolution theorem
13
Window multiplications

• Aims
• Improving S/N
• Resolution enhancements
• lineshape transformation
• (apodization) suppressing of ripple in spectrum
• deconvoluting instrument response

14
Improving S/N
FID
Filter function

• Exponential multiplication will reduce noise
• due to broadening signal intensity will be lost
• but noise decreases more

matched filter
15
Improving S/N
16
Resolution enhancement
Improvement in S/N at the cost of resolution
(linewidth)
Decreases S/N but improves resolution (linewidth)
17
Lineshape transformation
Product function
resolution enhancement
lorentz - gauss transformation
Lorentz-Gauss transformation small peak basis
gt signal separation - artifacts at peak base -
small lines can disappear - integration not
possible
18
Lineshape transformation
19
Deconvolution of instrument response
e(t) instrument s(t) signal
measure E(?) separately
20
Zero substitution
dominated by signal
dominated by noise
Window multiplications will do better !
21
Zero filling
Nk N
N
k N k1 adds information to the spectrum!
(it takes imaginary data into accont) kgt1
improvement of digital resolution
22
Zero filling
23
Signal phases
A(?)
D(?)
Detector phase is generally uncalibrated gt
mixture of absorption and dispersion
Spectrum
24
Phasing
Time or filter delays on time signal gt frequency
depended phase corr. needed
freq. independend
freq. dependend
(approximation)
25
Phasing time shift theorem
Each response (?) needs to be corrected with its
own phase correction
applies therefore only at ?
considerable changes over the linewidth of the
resonance gt lineshape distotions will result
(baseline roll)
26
Baseline corrections
Bad baselines are caused by distortions in the
initial points of S(t)
Bad points are caused by filter artifacts,
distortion or clipping of strong signals, etc.
27
Baseline corrections
uncorrupted FID
1st point corrupt
First three points corrupt
First two points corrupt