development of novel 2d nmr techniques for mixture analysis

Development of Novel 2D NMR Techniques for Mixture Analysis

A dissertation submitted to the University of Manchester for the degree of

Master of Science by Chemistry Research in the Faculty of Science and Engineering

2020

Arika Hisatsune

School of Nature Science Department of Chemistry

The University of Manchester


2

List of Contents

List of Abbreviations and Symbols ………….………………………………………….……………………………….. 4

List of Figures …………………………………………….……….…………………………………………………..………….. 5

Abstract ……………………………………………………....…………………………………………………………………….. 8

Declaration …………………….………………….……………….………………………………………………………………. 8

Copyright Statement ……………………….……….………..……….………………………………………………..……. 9

Acknowledgements ………………………….………….…………………………………………………….…………….. 10

1. Chapter 1 – Introduction ……………………………….…………………………….……………………..… 11

1.1. Thesis overview …………………………………………….………………,………………….……… 11

1.1.1. Understanding NMR ………………….….……………….….…………………………. 11

1.2. Spectral editing ………………………………………………………….……………………………… 14

1.2.1. Introduction …………………………..………….…..…..…….………..……………… 14

1.2.2. Perfect echo (PE) …………………….…………….…..….….……….……….……… 14

1.2.3. Low-pass J-filters (LPF) ……………………….….….…..….………………..……… 16

1.2.4. Zero Quantum Suppression (ZQS) ……………..…...….…………..…..……… 18

1.3. 2D NMR ……………………………………………………………………………………………………... 19

1.3.1. Introduction …………………………….……….….……………………………………… 19

1.3.2. Acquisition of spectrum ………….……………..…………...………………….…… 19

1.3.3. Pulse sequence of TOCSY ………….………….…….……………………………….. 19

1.4. Existing work, 1D DISPEL experiment ………………………………………….…………….. 20

2. Chapter 2 – 1D DISPEL using simulation ………………………..…………………………….…………. 21

2.1. Introduction ……………..……………..…..……………….…………….….…………………………. 21

2.2. Theory of Bloch equation ……………………….…..……….…..…..…………………….……… 21

2.3. DISPEL 1D simulation …….…………..…………..….…………..……………….…………………. 25

2.3.1. Investigation of 1H 90° off-resonance effect ….……….....…..……………. 25

2.3.2. Investigation of 13C 90° off-resonance effect …….….………………………. 25

2.3.3. Investigation of the influence of composite pulses on 13C ………………….

off-resonance effect ……………………………………………………………………. 26

2.3.4. Investigation of the effect of B1 miscalibration ……………………………… 28

3. Chapter 3 – 1D DISPEL experiment ……………………………………………………………………….. 31


3

3.1. Introduction ………..………………………………..…….…..…..…….……..……………………… 31

3.2. DISPEL 1D experiment ….………………………..….……………….……………..……..………. 31

3.2.1. Investigation of 13C 90° off-resonance effect using propanol ………….. 31

3.2.2. Investigation of effect of B1 miscalibration …………………………………….. 32

3.3. DISPEL 1D spectrum ……………………………………………………………………..………….. 34

3.3.1. 1D DISPEL spectrum of doped 2.5% propanol .………….……….………….. 34

3.3.2. 1D DISPEL spectrum of 100 mM quinine …………………………..….……….. 35

3.3.3. 1D DISPEL spectrum of Q-mix ……….……………………….….……..….……….. 36

4. Chapter 4 – 2D TOCSY-DISPEL experiment …………………………………………………………….. 38

4.1. Introduction …………………….………………..…………………………..………………………….. 38

4.2. 2D TOCSY-DISPEL spectra ….….…………..…………………….…….………….……………….. 38

4.2.1. 2D TOCSY-DISPEL Spectra of doped 2.5% propanol.…….……...…..……. 38

4.2.2. 2D TOCSY-DISPEL spectrum of 100 mM quinine ………….….….…..…….. 40

4.2.3. 2D TOCSY-DISPEL spectrum of Q-mix ……………….....………..…….……….. 41

5. Chapter 5 - Discussion ……………………………………………………..……………………………………. 45

5.1. Conclusions ……………………..…….……….………………..…….…..………………………….…. 45

5.2. Discussion ………………………………………………………………………………………………….. 45

5.3. Future work ……………..……….……..……………..……………….………………………………… 47

6. References ……………..……………………………………………………………………………………………… 48

Word count: 12023


4

List of Abbreviations and Symbols

1D One-Dimensional

2D Two-Dimensional

AP Anti-Phase

B0 External static magnetic field (tesla)

B1 Radiofrequency magnetic field (tesla)

COSY COrrelational SpectroscopY

CPMG Carr-Purcell-Meiboom-Gill

CW Continuous-Wave

DISPEL Destruction of Interfering Satellites by Perfect Echo Low-pass filtration

F1 (Indirect) dimension

F2 (Direct) dimension

IP In-Phase

J Scalar coupling constant (Hz)

LPF Low-Pass Filter

M Magnetization

NMR Nuclear Magnetic Resonance

PE Perfect Echo

PFG Pulsed Field Gradient

PO Product Operator

RF Radio Frequency

T Temperature (K)

t1 Incremented evolution period (s)

t2 Acquisition time (s)

TOCSY TOtal COrrelation SpectroscopY

ZQC Zero Quantum Coherence

ZQS Zero Quantum Suppression

δ Chemical shift (ppm)

τ Delay period (s)

𝛾 Gyromagnetic ratio


5

List of Figures

Fig. 1 500 MHz 1H spectrum of chloroform

Fig. 2 The pulse sequence of the spin echo (SE)

Fig. 3 Spinach simulation of spin echo pulse sequence applied to a two-spin system

Fig. 4 The pulse sequence of the Carr-Purcell-Melboom-Gill (CPMG)

Fig. 5 The pulse sequence of the perfect echo (PE)

Fig. 6 Spinach simulation of perfect echo pulse sequence applied to a two-spin system

Fig. 7 The pulse sequence of the one-staged LPF

Fig. 8 Graphical representation of low-pass filter stages applied on AX spin system.

Reproduced from P. Moutzouri et al.

Fig. 9 Graphical representation of Four-stage low-pass filter applied on AX spin

system. Reproduced from P. Moutzouri et al.

Fig. 10 500 MHz zTOCSY spectrum of menthol

Fig. 11 The pulse sequence of TOCSY

Fig. 12 The pulse sequence of DISPEL

Fig. 13 500 MHz 1H NMR spectra of CHCl3 doped with chromium tris-acetylacetonate

in DMSO-d6

Fig. 14 Spinach simulation of a spin echo pulse sequence applied to an 11 single-spin

system

Fig. 15 The effect of the normal 90° 13C pulses on signals obtained for a 13C chemical

shift range of 75000 Hz, using Spinach package

Fig. 16 The effect of the composite 13C pulses on signals obtained for a 13C chemical

shift range of 75000 Hz,using Spinach package

Fig. 17 The effect of the normal and composite 90° 13C pulses on signals obtained for a 13C chemical shift range of 75000 Hz, using Spinach package

Fig. 18 Spinach simulation of DISPEL pulse sequencewith B1 error using normal 90°

pulse

Fig. 19 The effect of the B1 miscalibration on signals obtained using normal pulses for

a 13C chemical shift range of 50000 Hz, using Spinach package


6

Fig. 20 Spinach simulation of DISPEL pulse sequencewith B1 error using composite 90°

pulse

Fig. 21 The effect of the B1 miscalibration on signals obtained using composite pulses

fora 13C chemical shift range of 50000 Hz, using Spinach package

Fig. 22 The pulse sequence of 1D DISPEL

Fig. 23 The experimental data of the effect of the normal 90° 13C pulses on signals

obtained for a 13C chemical shift range of 80000 Hz

Fig. 24 The experimental data of the effect of the normal 90° 13C pulses on signals

obtained for a 13C chemical shift range of 80000 Hz

Fig. 25 The experimental data of DISPEL pulse sequencewith normal 90° pulse with B1

error

Fig. 26 The experimental data of the effect of the B1 miscalibration on signals obtained

for a 13C chemical shift range of 80000 Hz

Fig. 27 The comparison between the experimental data and the Spinach simulation

data of the effect of the B1 miscalibration on signals obtained for a 13C chemical

shift range of 80000 Hz

Fig. 28 500 MHz 1H spectrum of 2.5 % doped propanol in DMSO-d6

Fig. 29 Expansion of 500 MHz 1H DISPEL spectrum of 100 mM quinine in DMSO-d6

Fig. 30 Expansion of 500 MHz 1H spectrum of Q-mixture in DMSO-d6

Fig. 31 The pulse sequence of 2D TOCSY-DISPEL

Fig. 32 500 MHz 2D TOCSY spectrum of n-propanol doped with chromium tris-

acetylacetonate in DMSO-d6

Fig. 33 500 MHz 2D TOCSY spectra of n-propanol doped with chromium tris-

acetylacetonate in DMSO-d6

Fig. 34 Expansion of 500 MHz 2D spectra marked by a purple square in Fig. 25

Fig. 35 Expansion of 500 MHz 2D spectra 100 mM quinine in DMSO-d6

Fig. 36 Expansion of 500 MHz 2D Q-mix spectra in DMSO-d6

Fig. 37 Expansion of 500 MHz 2D TOCSY spectra of Q-mix in DMSO-d6, region marked

by orange in Fig. 28


7

Fig. 38 Expansion of 500 MHz 2D TOCSY spectra of Q-mix in DMSO-d6, the region

marked by yellow in Fig. 28

Fig. 39 Expansion of 500 MHz 2D Q-mix spectrum in DMSO-d6


8

Abstract

The analysis of high dynamic range mixtures by 1H NMR is complicated by the presence of 13C

satellite signals. While satellites caused by long-range couplings are usually buried beneath

homonuclear multiplet structure, one-bond satellites of major mixture components often

overlap with signals from minor components, complicating both quantification and

identification. Broadband 13C decoupling can eliminate, or at least greatly reduce, interference

in 1H NMR from 13C isotopomer signals, but only at the cost of significant sample heating. This

limits spectral resolution, because longer acquisition times will increase this heating. A recent

NMR experiment, DISPEL (Destruction of Interfering Satellite by Perfect Echo Low-pass

filtration), suppresses one-bond 13C satellite signals in 1D 1H spectra without the need for

decoupling. This new approach is generally applicable, and it is possible to concatenate it with

a wide variety of multidimensional NMR methods such as COSY and TOCSY at very low cost in

signal-to-noise ratio, making it significantly easier to analyze the spectra of high dynamic range

mixtures by 1H NMR.

Chapter 1 contains an introduction to the theoretical NMR background necessary for this

thesis.

Chapter 2 contains details of 1D DISPEL simulations, investigating factors that influence the

degree of 13C-1H satellite peak suppression.

Chapter 3 contains experimental 1D DISPEL results, explaining how it can be applied to the

analysis of minor components in high dynamic range mixtures.

Chapter 4 contains experimental 2D DISPEL results, explaining how it can be applied to the

analysis of minor components in the high dynamic range mixtures.

Chapter 5 summarizes the conclusions of the research.

All the experimental and simulation data and parameters for the work described are freely

available at https://dx.doi.org/10.17632/2zvkcp3hjf.1.

Declaration

The author declares that all the work presented in this thesis has been completed at the

premises of The University of Manchester and at home due to COVID-19. Unless stated

otherwise at the beginning of each chapter, no portion of the work referred to in this thesis


9

has been submitted in support of an application for another degree or qualification of this or

any other university or other institute of learning.

Copyright Statement

I. The author of this dissertation (including any appendices and/or schedules to this

dissertation) owns certain copyright or related rights in it (the “Copyright”) and

she has given The University of Manchester certain rights to use such Copyright,

including for administrative purposes.

II. Copies of this dissertation, either in full or in extracts and whether in hard or

electronic copy, may be made only in accordance with the Copyright, Designs and

Patents Act 1988 (as amended) and regulations issued under it or, where

appropriate, in accordance with licensing agreements which the University has

from time to time. This page must form part of any such copies made.

III. The ownership of certain Copyright, patents, designs, trademarks and other

intellectual property (the “Intellectual Property”) and any reproductions of

copyright works in the dissertation, for example graphs and tables

(“Reproductions”), which may be described in this dissertation, may not be owned

by the author and may be owned by third parties. Such Intellectual Property and

Reproductions cannot and must not be made available for use without the prior

written permission of the owner(s) of the relevant Intellectual Property and/or

Reproductions.

IV. Further information on the conditions under which disclosure, publication and

commercialisation of this dissertation, the Copyright and any Intellectual Property

and/or Reproductions described in it may take place is available in the University

IP Policy, in any relevant Dissertation restriction declarations deposited in the

University Library, The University Library’s regulations and in The University’s

policy on Presentation of Dissertations


10

Acknowledgements

I would like to thank my academic advisors, Professor Gareth A. Morris and Professor

Mathias Nilsson for unconditional support and giving me this opportunity to complete my MSc

in Chemistry Research as a member of Manchester NMR Methodology Group. Every day, I saw

myself grow as an NMR spectroscopist and as a researcher. I’d also like to thank Dr Peter Kiraly,

who has always here for me, guiding me throughout my projects. My days were filled with full

of new knowledge and experiments! Thank you for your endless help, guidance and

discussions.

Thank you, Manchester NMR Methodology Group members, for honest opinions,

friendships and discussions (over pints sometimes). I could not have asked for a better group

to spend my MSc year with, and I would not have been here today without you guys!! You guys

made my time in the UK a special one. Thank you to my old supervisors from the University of

Toronto, Professor Andre Simpson and Dr Ronald Soong for supporting me in my

undergraduate year and acknowledging my growth as an NMR spectroscopist at ENC. That

meant a lot. Thank you, Ryan Anthony, for being here through ups and (many more of) downs,

listening to my complaints and always pushing me to be a better person. Without you, I am

not sure if I would be to complete the program. Merci pour tout. I am also thankful for the

School of Chemistry, the University of Manchester for providing a safe work environment and

for EPSRC for the research funding.

Last but not least, a special thanks go to my parents, Toshiyuki and Mika and my siblings,

Miki and Taiki, for unconditional love and support throughout my time in Manchester! You

guys have provided me with an opportunity to go and to live in another country, and always

had faith in me. Thank you for never leaving my side and acknowledging every single small

accomplishment of mine along the way. I LOVE YOU GUYS and ありがとう.


11

Chapter 1 – Introduction

1.1 Thesis overview

This thesis describes new experiments for the analysis of high dynamic range mixtures

by Nuclear Magnetic Resonance (NMR) spectroscopy. NMR is often used as a non-destructive

and minimum sample preparation analytical method, but problems can arise when it comes to

the analysis of a high dynamic range mixture. Mixture analysis provides rich information and

is important in fields like food chemistry and the pharmaceutical industry but can be

challenging.

The first successful accurate measurements of nuclear magnetic moments using

magnetic resonance absorption were done in 1938 by I. I. Rabi.1 In 1946, the group of F. Bloch

and E. M. Purcell independently demonstrated NMR for condensed matter, and the start of

NMR’s journey as an analytical technique began.2 In 1950, it was discovered that there are

slight changes in the atomic nucleus Larmor frequencies3 due to chemical shifts and spin

couplings. Larmor frequencies is also known as precessional frequencies, which refers to the

rate of precession of the magnetic moment (such as 1H) around the external magnetic field.

Over the past 80 years, NMR spectroscopy has bloomed into a major tool for analytical

chemistry, and it is now routinely used in the fields of chemistry, pharmaceutical science,

biology and medicine.

1.1.1 Understanding NMR

The atomic nucleus4 can be regarded as a spinning charged particle, which generates a

magnetic field. With the application of an external magnetic field (B0), the quantization of spin

energy occurs. The spin movement is controlled by the energy state of the spins, which are

known as 𝛼-spin state and 𝛽-spin state. The𝛼 is the low energy state for nuclei with a positive

gyromagnetic ratio (𝛾).5 The gyromagnetic ratio is a constant for a particular nucleus and is

directly proportional to the strength of the tiny nuclear magnetic moment.

The energy required to induce flipping depends on the strength of the magnetic field

(B0). The frequency of the nuclear transition can be written as

𝑣! = 𝛾𝐵!/2𝜋 (equation 1.1)


12

This two-state description is only applicable for those nuclei with spin quantum number of

I=1/2 such as 1H, 13C, 15N, 19F and 31P. Some spins precess about the positive z-axis and some

about the negative z-axis. The total magnetization determines an NMR signal.

The distribution of nuclei in the different energy state is given by the Boltzmann

equation6 "!""#$"%&'#$

= 𝑒#$%/'( = 𝑒#)*/'( (equation 1.2)

where Nupper and Nlower represent the population of nuclei in the upper and lower energy state.

It is important to have an excess number of nuclei in the lower energy state for a signal to be

detected. The small population difference presents a significant sensitivity problem for NMR.

As seen in equations 1.1 and 1.2, the use of a stronger magnetic field will increase the

sensitivity. The application of the magnetic field strength should be uniform across the sample

to avoid achieving the Larmor condition at a range of frequencies, leading to a broader signal.

Relaxation processes return the spin system to thermal equilibrium, in the absence of

perturbing radio frequency (RF) pulses.7,8

The most common form of NMR performed is proton (1H) NMR. In samples with natural

abundance, 99.98% is the isotope 1H.9 Proton NMR spectra of most organic compounds are

characterized by chemical shifts in the range of +14 to -4 ppm and by spin-spin couplings

between protons.10 NMR can be performed in both liquids and solids, but the simplest form is

in liquids and that generally requires a solvent. The solvent used in liquid-state NMR is

deuterated to avoid swamping of the signals, to accurately define 0 ppm and to stabilize

magnetic field strength. Deuterated water (D2O) and deuterated chloroform (CDCl3) are

common solvents.

Chemical shifts, signal integrals and splitting patterns provide rich information on the

structure of the sample being analyzed. The chemical shift9,11,12 determines the position in the

spectrum at which nuclei resonate. The exact values depend on local electron density,

determined by molecular structure, neighbouring functional groups, hybridization, solvent,

and temperature. The stronger a nucleus’ local magnetic field, the higher their chemical shifts

will be (deshielding)3, the higher their chemical shift values will be. The area under the signal9

is determined by the relative numbers of spins contributing to the signal. Multiplet patterns


13

are determined by the number and magnitude of the scalar couplings experienced by a nucleus

and by the spin quantum numbers of the nuclei to which it is coupled.

The magnitude of splitting is known as the scalar coupling constant (J).13 This J-coupling

does not change with the magnetic field, so it is quoted in Hz and not ppm. Spin-spin

couplings14 are caused by the small magnetic fields that nuclei with I > 0 possess. Due to these

magnetic fields, they influence each other, resulting in the changes in the energy and hence

Larmor frequency of nearby nuclei as they resonate. The scalar coupling is one of the most

important types of coupling and arises from the magnetic interactions between two nuclei that

are communicated through chemical bonds. It can be used to determine the conformations of

chemical species.

Small extra peaks can often be seen on either side of the main peaks on 1H NMR spectra.

These are caused not by homonuclear scalar coupling (JHH) but by heteronuclear scalar

coupling (JCH), and they are called carbon satellite peaks.13,15 Carbon satellite peaks are small

because the natural abundance of the I=1/2 13C isotope is only 1.1%, so couplings to 13C results

in two extra peaks of 0.55% intensity, one-bond 13C coupled to 1H (1JCH ) Hz apart, either side

of the main 12C-1H peak. See Fig. 1 for an example of satellite peaks in 1H NMR.

Figure 1: 500 MHz 1H spectrum of chloroform showing 13C satellite peaks either side of a parent peak.

This spectrum was acquired with 16 scans, in an experiment time of 1 min and 46 s.

Satellite peaks generally show the same homonuclear multiplet structure as parent peak, so if

the parent peak is a triplet, the satellite peaks would also normally be triplets. 13C satellite

peaks cause problems when it comes to the analysis of high dynamic range mixtures because

satellite peaks can mask or be confused with signals of components that are present at low

concentration, making the latter harder to identify and quantify. This thesis describes methods

for removing satellite peaks in 1D simulation and in 2D NMR.

0.55%


14

1.2 Spectral editing

1.2.1. Introduction

The analysis of chemical mixtures can be hard, not only because mixtures contains more than

one analyte but also due to signal overlap. With overlap, some signals could be masked, making

chemicals that are present at low concentration hard to identify and quantify. Spectral

editing16 is a technique used to suppress certain classes of NMR signals based on their NMR

properties. The perfect echo, low pass J-filters, and zero-quantum suppression are examples

of building blocks used to suppress unwanted signals.

1.2.2. Perfect echo (PE)

The spin echo18 is the single most important building block in the modern NMR. The

pulse sequence is

Figure 2: The pulse sequence of the spin echo (SE).

It contains 90 and 180 pulses shown as thin and thick rectangles with delay (𝜏) in between.

With a simple spin echo, a spin system with homonuclear coupling shows J-modulation as a

function of the time 2𝜏. J-modulation is caused by the effect of the scalar coupling remaining

while the chemical shift is refocused, resulting in distortion of the spectrum. A simulation

package (Spinach)17 can be used to demonstrate J-modulation effects in 1D 1H NMR, see Fig.

3.

Figure 3: Spinach simulation of 1H spin echo pulse sequence applied to a coupled two-spin system,

for 2𝜏 times of (a) 1ms, (b) 10 ms, and (c) 100 ms. 1JHH of 150 Hz.

a) b) c)


15

With increasing 𝜏, more J-modulation can be seen, affecting the phases of the spectral peaks.

In more complex experiments, it can result in sensitivity loss or missing spectral peaks.

To suppress J-modulation, a Carr-Purcell-Meiboom-Gill (CPMG)14 experiment can be

used. The pulse sequence is

Figure 4: The pulse sequence of the Carr-Purcell-Meiboom-Gill (CPMG) experiment. tau is usually in 𝜇𝑠.

However, it comes with the cost of high RF power deposition, which will cause sample heating,

resulting inter alia in problems such as line-shape distortions and shifts in signals. CPMG can

suppress the modulation arising from couplings between spins with chemical shift differences

∆𝜈 << 1/𝜏.14 However, J-modulation can be also refocused by a method called the “perfect

echo.”18

The perfect echo14,19,20 is used to refocus the effects of both JHH and chemical shifts. The

perfect echo pulse sequence is

Figure 5: The pulse sequence of the perfect echo (PE).

A perfect echo is made by inserting a 90° flip angle orthogonal (with phase y if the initial

excitation pulse has phase x) pulse at the midpoint of a double spin echo. During the spin echo,

the chemical shift evolves during the first delay 𝜏 and is then refocused by the 180° pulse; the

same goes for JHX. However, JHH keeps on evolving even after application of a 180° pulse,

contributing to phase distortion. The distortion is caused by the buildup of anti-phase (AP)

coherence with respect to homonuclear J-coupling. The perfect echo acts as a J-compensated


16

building block by refocusing the unwanted AP coherences, giving an improvement in spectral

quality. See Fig. 6.

Figure 6: Spinach simulation of 1H perfect spin echo pulse sequence applied to a coupled two-spin system,

for 2𝜏 times of (a) 1 ms, (b) 10 ms, and (c) 100 ms. 1JHH of 150 Hz was used.

This sequence gives spectra that are free of phase anomalies arising from undesired J-

modulation. In the next section, it will be shown that this also provides time to apply a four-

stage 13C low-pass 1JCH filter, allowing the signals of all protons directly attached to 13C to be

suppressed.

1.2.3. Low-pass J-filters (LPF)

A low-pass J-filter20,21 suppresses signals whose scalar coupling constants exceed a

lower limit. Signals with large J-couplings, for example, one-bond 13C satellite signals, which

typically have 1JCH in the range of 115-250 Hz and are to be from the spectrum while long-range

satellite signals, with nJCH of 0-30 Hz, remain unperturbed.

The simplest low-pass JCH filter pulse sequence is shown below. Instead of refocusing

AP coherences with respect to homonuclear J-coupling like the perfect echo, this method

suppresses AP coherences with respect to heteronuclear J. This is a one-stage low-pass filter,

as it only contains one carbon pulse. The tau delay in the pulse sequence is set to 1/(21JCH), of

the order of 4 ms.

Figure 7: The pulse sequence of the one-staged LPF.

The initial 1H 90°x excitation pulse generates transverse magnetization, −Hy. The product

operator is explained the section 2.2. During the τdelay, proton magnetization coupled to 13C

a) b) c)


17

evolves into −Hycos(πJCHτ), in-phase (IP), and −2HxCzsin(πJCHτ), anti-phase (AP) coherences.

The amount of AP coherence generated is dependent on the coupling constant and the

duration of the τdelay. The 90° pulse on the carbon converts AP coherence into multiple

quantum coherence. Only single-quantum coherence results in a detectable signal; multiple

quantum coherence will not be seen.

A four-stage low-pass JCH filter is used in the DISPEL sequence. The more stages are

used, the greater the opportunity to suppress signals with a wider range of 1JCH. The perfect

echo provides enough time for four pulses. See Fig. 8 and Fig. 9 for how the number of stages

influence a heteronuclear AX spin system suppression. Fig. 8 and Fig. 9 were retrieved from a

paper published in 2017 by P. Moutzouri et al. 13

Figure 8: Graphical representation of (a) one-stage, (b) two-stage, (c) three-stage

low-pass filter stages applied on AX spin system. Reproduced from P. Moutzouri et al.13 The delay τ is set to 4.05 ms.

With a one-stage filter, only a narrow range of 1JCH gives sufficient (>>10:1) suppression, but

with increasing numbers of stages, the range of 1JCH giving good suppression is widened.

Figure 9: Graphical representation of four-stage low-pass filter applied on AX spin system. The blue line is a 10x magnification of the red line. Reproduced from P. Moutzouri et al.13

τ1, τ2, τ3, and τ4 were 3.2, 1.1, 3.95, and 1.56 ms, respectively.


18

The blue line in Fig. 9 is the red line magnified by ten times to see the oscillation patterns. With

a four-stage low-pass filter, a suppression factor of 70:1 or better is achieved for heteronuclear

couplings from 110 to above 350 Hz.

1.2.4. Zero Quantum Suppression (ZQS)

Many types of coherences can be generated, but only single-quantum coherences can

be detected. Zero-quantum coherence evolves at the frequency which is the difference

between two chemical shifts. When zero quantum coherences are converted into observable

single quantum coherences, they do so with phases which depend on the evolution time. The

result is the appearance of AP dispersive signals, interfering with the spectrum acquired. See

Fig. 10 for an example. The baseline of the Fig. 10a is distorted due to zero-quantum coherence.

(a) (b)

Figure 10: 500 MHz selective 1D 1H TOCSY spectrum of menthol (a) without and (b) with ZQS. Each spectrum was acquired with 8 scans, in an experiment time of 45 s.

In order to get rid of the AP dispersive signals, zero-quantum suppression20,22,23 is used.

This uses a frequency-swept 180° pulse24 applied simultaneously with a field gradient. The field

gradient causes the Larmor frequency to become a function of position in the active volume

of the NMR tube. This ensures that spins at different positions experience the 180° pulse at

different times, which results in different parts of the sample having different zero-quantum

evolution times. The field gradient is turned off at the end of the 180° pulse or else the different

parts of the sample will give a signal at a different Larmor frequency. The signal components

that originate from zero-quantum coherence averaging to zero, giving a spectrum with no AP

dispersive signal interference. The zero-quantum suppression is followed by a further gradient

pulse, which is used to dephase all other orders of coherence. The result is a pulse sequence

element that only allows longitudinal magnetization to survive.

0.7

𝛿 𝐻/𝑝𝑝𝑚" 𝛿 𝐻/𝑝𝑝𝑚

"𝑝𝑝𝑚

3.7 2.9 2.1 1.5 3.7 2.9 2.1 1.5 0.7

a) b)


19

1.3 2D NMR

1.3.1 Introduction

The principles of 2D NMR spectroscopy were first presented in 1971, and during the 1980s the

approach found wider application. “Two-dimensional” refers to two frequency dimensions,

while 1D methods have only one.25,26 “Pseudo-2D” methods have also been developed for

which the idea of a 2D representation has been adapted to include one frequency and one

‘other’ dimension. Conventional 2D experiments find wide utility in chemical research.

1.3.2 Acquisition of spectrum

Generating the second dimension requires data acquisition with a pulse sequence containing

preparation, evolution, mixing and detection periods. The preparation and mixing periods

depend on the nature of the experiment. The simplest form of preparation or mixing period is

a single 90° pulse, which is used to excite transverse magnetizations. The evolution period

provides the key to the generation of the second dimension. The mixing part of the sequence

may transfer magnetization associated with one spin to other spins, for example to spins with

which it is coupled, a process known as coherence transfer.27 Control of the coherence transfer

pathway over the course of a pulse sequence can be achieved by phase cycling and/or field

gradient pulses. While gradient pulses can save time and avoid the need for the signal

addition/subtraction used in phase cycling, phase errors due to chemical shift evolution during

the application of a gradient pulse can cause complications. Therefore, they are often applied

within a spin echo to refocus this evolution. The result of a two-dimensional experiment is a

2D time-domain data set s(t1,t2); the free induction decays observed in t2 are modulated in

phase and/or amplitude as a function of t1, and are doubly Fourier transformed to give a two-

dimensional spectrum S(F1,F2).

1.3.3 Pulse sequence of TOCSY

TOCSY (TOtal Correlation SpectroscopY)26 is a homonuclear experiment; it shows cross-

peaks between spins which are connected by an unbroken chain of couplings, so it is useful for

identifying spins which belong to an extended network of couplings. The pulse sequence of

TOCSY is


20

Figure 11: The pulse sequence of Total Correlational SpectroscopY (TOCSY).

The key part of the TOCSY experiment is the period of isotropic mixing or spin locking,

which is shown here as a DIPSI-2 sequence element.26 The spin lock is arranged that only z-

magnetization present at beginning and the end of DIPSI-2 contributes to the spectrum. In a

two-spin system, such a period of isotropic mixing causes the following evolution of z-

magnetization:

𝐼6+,→1/2 [1 + cos(2𝜋𝐽+.𝜏𝑚𝑖𝑥)]𝐼6+, + 1/2[1 − cos(2𝜋𝐽+.𝜏𝑚𝑖𝑥)] 𝐼6.,

− sin(2𝜋𝐽+.𝜏𝑚𝑖𝑥)1/2(2𝐼6+/𝐼6.0 − 2𝐼6+0𝐼6./)

The 𝜏 mix is duration of time during DIPSI-2. The important thing here is that z-magnetization

on spin one is transferred to z-magnetization on spin two, giving rise to cross-peaks in the 2D

spectrum, to an extent that depends on the coupling, and the mixing time. The overall

intensities of the TOCSY cross-peaks depend on the mixing transfer function and the relaxation

that takes place during the isotropic mixing which leads to loss of signal intensity. To get

correlations for smaller couplings, a longer period of mixing is required, thus the loss due to

relaxation will be more severe than when observing correlation through larger couplings.

1.4 Existing work, 1D DISPEL experiment

DISPEL stands for Destruction of Interfering Satellites by Perfect Echo Low-pass filtration. This

work was first done in 2017 in 1D.13 DISPEL contains three components which were

mentioned in Section 1.2: a perfect echo, low-pass filter and zero quantum suppression. The

pulse sequence is

Figure 12: The pulse sequence of the Destruction of Interfering Satellite by Perfect Echo Low-pass filtration (DISPEL).

τ1, τ2, τ3, and τ4 were 3.2, 1.1, 3.95, and 1.56 ms, respectively.

t1 t2


21

DISPEL suppresses one-bond satellite peaks with high efficiency at negligible cost in sensitivity.

The effect of DISPEL on a 1D 1H NMR spectrum is shown in Fig. 13. To facilitate quantitative

comparison, the normal 1H spectrum of Fig. 13(a) was obtained using the DISPEL sequence

with the 13C pulses omitted. The one-bond 13C satellite peaks are suppressed with high

efficiency in Fig. 13(b), at negligible cost in signal-to-noise ratio and without the need for

broadband 13C decoupling.

a) b)

Figure 13: 500 MHz 1H NMR spectra of CHCl3 doped with chromium tris-acetylacetonate in DMSO-d6,

(a) without and (b) with the 13C pulses of a 1D DISPEL experiment. One-bond satellite peaks are marked by stars. Each spectrum was acquired with 16 scans, in an experiment time of 1 min and 46 s.

Chapter 2 – 1D DISPEL using simulation

2.1 Introduction

Spinach17 is a MATLAB software library often used for simulation of spin dynamics in all types

of magnetic resonance (NMR, EPR, MRI, etc); it has well-annotated open-source code. Spinach

provides a method to simulate large spin systems quantum mechanically and is a useful tool

to predict the outcomes of experiments in a perfect condition. This chapter contains

simulations done using Spinach and aims to investigate two factors that affect the suppression

of satellite peaks: 13C off-resonance effects and B1 miscalibration. Demonstrations using

important building blocks, the spin echo and perfect echo, were presented in chapter 1 section

2.2.

2.2 Theory of Bloch equations

Before 1966, most chemical applications of NMR used continuous-wave (CW)

spectrometry, which relies on the detection of the steady-state transverse magnetization, as

a function of static magnetic field strength at a fixed frequency. The Bloch equations28 provide

a framework for treating the simultaneous effects of relaxation, RF fields, and resonance offset

(Ω!) for an ensemble of isolated spins-1/2. Offset from resonance is defined by

Ω! = 𝛾𝐵! − 𝜔 (equation 2.1)

𝛿 𝐻"

9.48 9.30 9.12 8.94 8.76 9.48 9.30 9.12 8.94 8.76

𝛿 𝐻"

0.55%

a) b)


22

in rad s-1. For simplicity, the Bloch equations follow the magnetization dynamics in a frame of

reference that rotates about the static field direction at the radiofrequency used. The Bloch

equations are a set of three coupled differential equations for the components of the net

nuclear magnetization vector, normalised with respect to the thermal equilibrium z-

magnetization M0. The equations can be written as:29

112G𝑀0𝑀/𝑀,

I=G0 −Ω! 𝜔34 sin𝜙Ω! 0 −𝜔34 cos𝜙

−𝜔34 sin𝜙 𝜔34 cos𝜙 0IG

𝑀0𝑀/𝑀,

I +G–𝑀0 /𝑇.–𝑀/ /𝑇.

(1– 𝑀,)/𝑇+

I

(equation 2.2)

𝜔34 is the nutation frequency of the RF (its amplitude in angular frequency units), and 𝜙 is its

phase. T1 and T2 are time constants used to characterize two types of relaxation – spin-lattice

and spin-spin relaxation that return the spins to equilibrium.8,17,29 The spin-lattice relaxation

re-establishes the Boltzmann distribution of spin states responsible for net z-magnetization,6

and spin-spin relaxation causes loss of the phase coherence between transverse components

of individual spins that is responsible for net transverse magnetization.

The energy of the classical magnetic dipole in a magnetic field is given by:

E =𝜇B0 = 𝜇,Bz (equation 2.3)

where 𝜇 is the magnetic moment, 𝜇, is its z component and the magnetic field B0 is along the

z-axis. The magnetic moment is proportional to its angular momentum, which makes it

𝜇,=𝛾𝐼,.26 𝛾 is the magnetogyric ratio.29 Based on these equations, the quantum mechanical

expression can be obtained by replacing E and Iz by the corresponding quantum mechanical

operators:

𝐻P = 𝛾𝐵!ℏ𝐼6, = ℏ𝜔!𝐼6, (equation 2.4)

where 𝜔! = 𝛾𝐵! is an angular frequency, also known as the Larmor frequency and ℏ is

Planck’s constant divided by 2𝜋. For convenience, energies are often expressed in angular

frequency units in NMR, with ℏ replaced by 1.26

In a CW experiment, the RF field tends to rotate M away from the z-axis while

transverse magnetization is being destroyed by transverse relaxation and spin-lattice

relaxation is regenerating longitudinal magnetization. At a sufficiently long time, the system

settles down into a steady-state, under which these two tendencies are balanced. If the phase


23

𝜙 and resonance offset Ω! are both set equal to zero, then the equations for the time

derivatives of the magnetization components My and Mz are: 112𝑀/ = −𝑇. 𝑀/ − 𝜔34𝑀,

#+ (equation 2.5)

112𝑀, = 𝜔34𝑀/ − 𝑇+ 𝑀/(𝑀, − 1)

#+ (equation 2.6)

In the steady-state, both time derivatives vanish and allow one to solve for the two

magnetization components. The steady-state magnetization decreases as T1, T2 and the

nutation frequency (𝜔34)increase.29

However, the Bloch equation assumes that spins do not interact coherently, which is

not the case for many systems for which NMR can provide useful information. NMR, like any

quantum phenomenon, is governed by the time-dependent Schrödinger equation.30 Spins-1/2

have two eigenstates, 𝛼 and 𝛽 : the spin wavefunction (𝜓) is a variable mixture of these.

Quantum mechanics represents the states of systems by wavefunctions (𝜓) and uses

operators to deduce both the values of observable quantities and the evolution of those

wavefunctions. The two eigenfunctions of the operator for the z component of dimensionless

spin angular momentum, Iz, may be given the symbols of 𝛼 and 𝛽 and in Dirac bra-ket notation

|𝛼⟩ and |𝛽⟩:

Iz |α> = ++.|α⟩ Iz |β> = – +

.|β⟩

The state of a single spin-1/2 can be written as a general superposition state:

|𝜓⟩ = U𝐶5𝐶6W (equation 2.7)

where the left-hand side uses the Dirac bra-ket notation for the wavefunction and the right-

hand side tabulates the (complex) coefficients of the spin-1/2 basis states |α> and |β>. If an

operator 𝐴6 acting on a ket |𝜓⟩ yields a result 𝑎|𝜓⟩, then |𝜓⟩ is said to be an eigenvector of 𝐴6

with eigenvalue α.29 If the operator 𝐴6 corresponds to an observable quantity, then the

expectation value obtained from an experimental measurement of this quantity for a system

in the state |𝜓⟩ is given by the scalar product of ⟨𝜓| with 𝐴6|𝜓⟩:

⟨𝐴6⟩=⟨𝜓|𝐴|[𝜓⟩

In practical calculations, we need to know the values Aij of all possible scalar products for

different basis states i and j; these can be written as a matrix representation A, in which the

ij’th element is Aij where Aij =⟨𝜓7|𝐴|[𝜓8⟩.


24

NMR experiments measure transverse magnetization, which corresponds to a net

excess of spins with correlated coefficients such as Cα and Cβ: coherences, typically generated

by applying radiofrequency pulses. They can be tabulated in the form of a density matrix, which

summarises the quantum states of the entire ensemble of the spins, without referring to the

states of individual spins. The density matrix can be regarded as representing a density

operator; for an ensemble of non-interacting spins – 1/2:

𝜌] = ^𝜌55 𝜌56𝜌65 𝜌66_ = `

𝐶5𝐶5 𝐶5𝐶6aaaaaaa

𝐶6𝐶5 𝐶6𝐶6b

The diagonal elements of the spin density operator 𝜌55 and 𝜌66 are the populations of the

states |𝛼⟩ and |𝛽⟩.29,31 The off-diagonal elements 𝜌56 and 𝜌65 are the coherences between

states |𝛼⟩ and |𝛽⟩.

Unfortunately, density matrix analysis rapidly becomes unwieldy as the number of

spins increases. For weakly-coupled spin systems, analysis can be greatly simplified by

decomposing the density operator into a sum of product operators, whose evolution can be

calculated using simple rotations.

For a system of isolated spin −1/2 nuclei, the density operator can be decomposed into

the operators E, Ix, Iy and Iz, where the first is the identity operator and the remainder the x, y

and z components of spin angular momentum. From the vector model, it is easy to see how

these magnetizations transform under the influence of pulses with flip angle 𝛽:

𝐼069)cd 𝐼0𝐼0

69*cd 𝐼0 cos 𝛽 − 𝐼, sin 𝛽

𝐼/69)cd 𝐼/ cos 𝛽 + 𝐼, sin 𝛽 𝐼0

69*cd 𝐼/

𝐼,69)cd 𝐼, cos 𝛽 − 𝐼/ sin 𝛽 𝐼,

69*cd 𝐼, cos 𝛽 + 𝐼0 sin 𝛽

The free precession Hamiltonian is:32

𝐻P = Ω!𝐼, (equation 2.8)

It causes rotation about the z-axis at frequency Ω! . Free precession for a time t causes a

rotation through an angle 𝛼 , where 𝛼 = Ω!𝑡. Only x- and y-magnetization are directly

observable in an NMR experiment; it is the precession of the magnetization in the xy-plane

which gives rise to the free induction signal. For a system of two spins, each spin would have

three operators plus the identity, making a total of 16 combinations. As spins 1 and 2 are

coupled, they generate in-phase and anti-phase magnetization and zero-quantum, single-


25

quantum and multiple-quantum coherences. As stated earlier, only single quantum can be

observed.

2.3 DISPEL 1D Simulation

2.3.1 Investigation of 1H 90° off-resonance effect

The spin echo is the single most important building block in modern NMR. This basic building

block is used here to investigate the effect of 1H 90° pulse duration and offset from resonance

on the signals obtained. 11 single spins that are not coupled to each other are placed evenly

across the spectrum to demonstrate the off-resonance effect. Three different 1H 90° pulse

durations, 1, 10 and 50 µs, were used to investigate the effect:

a)

b)

c)

Figure 14: Spinach simulation of a 1H spin echo pulse sequence applied to an 11 single-spin system, with 1H 90° pulse durations of (a) 1µs, (b) 10µs, and (c) 50 µs for a range of 10000 Hz.

The on-resonance signal, at 0 Hz, stays at the same intensity for all three 1H 90° pulse durations.

At pulse durations of 1 µs and 10 µs, the offset has little effect. However, at a pulse duration

of 50 µs, the influence of offset is getting stronger as the spins go further away from resonance,

resulting in increased signal phase changes and decreased intensity. The signal intensity drop

is symmetric with respect to resonance.

2.3.2 Investigation of 13C 90° off-resonance effect

Carbon off-resonance effects can have some influence on the degree of 13C-1H coupled

peak suppression. Here, it is investigated using the DISPEL sequence. As Fig. 9 shows, 13C-1H

coupled peaks at 1JCH of 145 Hz is at great level of signal suppression but not perfectly cancelled.

To demonstrate the effect if 13C resonance offset in DISPEL suppression. To demonstrate the

effect of 13C resonance offset on DISPEL suppression, the 13C chemical shift is varied from -

-5000 0 5000 1H chemical shift (Hz)

a) b) c)


26

37500 Hz to 37500 Hz for a 1JCH of 145 Hz in Fig. 15. The ratio of the satellite amplitude with

DISPEL to the original satellite amplitude is shown on the right-hand scale.

Figure 15: The effect of the hard 90° 13C pulses on signals obtained for a 13C chemical shift range of 75000 Hz, using Spinach package. Spin system of AX was used. 1JCH of 150Hz was used.

The 13C-1H coupled peaks change sign from negative to positive around +/- 12500 Hz. Past that

point, the suppression is no longer efficient and is experiencing a lot of off-resonance effects.

For a wide range of chemical shift, the suppression was performed efficiently. As it was for 1H

off-resonance effect, the influence of offset is getting stronger as the spins go further away

from resonance.

It is important to note that a hard 90° pulse is already efficient at suppressing satellite

peaks. Considering a normal 90° pulse’s ability to generate transverse magnetization, it can be

said that it contains ‘in-built’ self-compensation for off-resonance effects. Increased signal

phase changes and intensity are caused by the influence of offset driving the vectors further

towards the traverse plane at the expense of frequency-dependent phase errors. These errors

are approximately a linear function of frequency and can be removed through phase correction

of the spectrum.25

2.3.3 Investigation of the influence of composite pulses on 13C off-resonance effect

Composite pulses33 are made up of a number of conventional pulses which rotate the

magnetization vectors about different axes, sometimes with free precession allowed to occur

in the intervals. They can be used to compensate for off-resonance effects. A carefully chosen

composite of imperfect pulses makes a pulse that is more perfect than just a normal pulse.

Composite pulses can be placed within a pulse sequence in place of a hard pulse; however,

composite pulses take slightly more time than normal pulses. There are many variations of

composite pulses: the composite pulse 10°X – 𝜏 – 100°-X33 is used here. To demonstrate the

13C chemical shift (Hz)

0.1

| | | | | | |

-37500 -25000 -12500 0 12500 25000 37500

0.3 0.2


27

effect of 13C resonance offset on DISPEL suppression, the 13C chemical shift is varied from -

37500 Hz to 37500 Hz for a 1JCH of 145 Hz in Fig. 16. The ratio of the satellite amplitude with

DISPEL to the original satellite amplitude is shown on the right-hand scale.

Figure 16: The effect of the composite 13C pulses on signals obtained for a 13C chemical shift range of 75000 Hz, using Spinach package. Spin system of AX was used. 1JCH of 150Hz was used.

The 13C-1H coupled peaks change their sign from negative to positive past +/- 18750 Hz instead

of at +/- 12500 Hz like normal pulse did. A wider range of chemical shift was suppressed

compared to a normal pulse. As it was seen in the 13C off-resonance effect, Fig. 15, the

influence of the offset is getting stronger as the spins go further away from resonance.

The comparison of the 13C off-resonance effect using normal and composite 90° 13C for

a 13C chemical shift range of -37500 Hz to 37500 Hz is shown in Fig. 17. The intensity ratio of

relative satellite amplitude is used as a vertical scale.

Figure 17: The effect of the hard and composite 90° 13C pulses on signals obtained for a 13C chemical shift range of 75000 Hz,

using Spinach package. Spin system of AX was used. 1JCH of 150Hz was used.

-0.05

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

-50000 -40000 -30000 -20000 -10000 0 10000 20000 30000 40000 50000

I cou

pled

/ no

t cou

pled

13C chemical shift (Hz) at 1JCH of 145 HzNormal pulse Composite pulse


0.1

0.2

-37500 -25000 -12500 0 12500 25000 37500

| | | | | | |


28

This result demonstrates that the composite pulse provides a modest increase in 13C

bandwidth. In both pulses, the suppression is no longer adequate beyond some offset.

Composite pulses are especially useful when a wider range of chemical shifts is present, such

as for the analysis of a natural compound that contains alkanes and aromatic functional groups.

With a negligible increase in experimental time, the off-resonance effect can be minimized.

2.3.4 Investigation of the effect of B1 miscalibration

B1 miscalibration has an influence on the degree of 13C-1H coupled peak suppression as

it causes imperfect flip angles, similar to off-resonance effects. B1 inhomogeneity can be easily

compensated by nulling Mz using a composite 90x90y sequence. These composite pulses allow

the magnetization vectors to be placed closer to the transverse plane than a single pulse. Fig.

18 shows the Spinach simulated result of 0%, 10% and 20% errors in B1 for a 13C chemical shift

range of 50000 Hz using hard 90° pulses. A 13C pulse duration of 15 𝜇s was used for these

simulations. The intensity ratio of relative satellite amplitude is shown on the right-hand side.

a)

b)

c)

Figure 18: Spinach simulation of DISPEL pulse sequence with B1 error of (a) 0 %, (b) 10 %, and (c) 20 % using hard 90° pulse. The intensity ratio of relative satellite amplitude is shown on the right-hand side. Duration of 90 degrees 1H pulse was 0.1 𝜇𝑠,

13C pulse was 15 𝜇𝑠, and τ1, τ2, τ3, and τ4 were 3.2, 1.1, 3.95, and 1.56 ms, respectively.

With 0 % error, the 13C-1H coupled peaks change their sign from negative to positive past +/-

12500 Hz. With 10 % and 20 % error, the heights of 13C-1H coupled peaks are bigger than they

are with 0 % error on resonance and throughout the range.

-25000 -12500 0 12500 25000


0.06 0.02

| | | | |

0.12 0.06 0.02

0.18 0.12 0.06 0.02


29

The comparison of processed B1 miscalibration data is shown in Fig. 19. The intensity

ratio of relative satellite amplitude is used as a vertical scale.

Figure 19: The effect of the B1 miscalibration on signals obtained using hard pulses for a 13C chemical shift range

of 50000 Hz, using Spinach package. The intensity ratio of relative satellite amplitude is shown on the right-hand side. Duration of 90 degrees 1H pulse was 0.1 𝜇𝑠, 13C pulse was 15 𝜇𝑠, and τ1, τ2, τ3, and τ4 were 3.2, 1.1, 3.95, and 1.56 ms,

respectively.

Fig. 19 demonstrates that with greater errors in calibration, less suppression will be achieved

on resonance and throughout the offset range. With calibration errors, the satellite peaks

would be 3 times taller than when they are without errors at -/+ 25000 Hz.

Composite pulses were used to demonstrate the compensation with B1 error. See Fig.

20. The intensity ratio of relative satellite amplitude is used as a vertical scale.

a)

b)

c)

Figure 20: Spinach simulation of DISPEL pulse sequence with B1 error of (a) 0 %, (b) 10 %, and (c) 20 % using composite 90° pulse Duration of 90 degrees 1H pulse was 0.1 𝜇𝑠, 13C

pulse was 15 𝜇𝑠, and τ1, τ2, τ3, and τ4 were 3.2, 1.1, 3.95, and 1.56 ms, respectively.

The highest amplitude of a 13C-1H coupled peak is approximately 1 % of the uncoupled peak

when there is no error in B1. Compared to hard pulses, composite pulses achieved a steady

suppression over a wider range of chemical shift. The comparison of processed B1

-0.020

0.020.040.060.08

0.10.120.140.160.18

0.2

-30000 -20000 -10000 0 10000 20000 30000

I cou

pled

/ no

tco

uple

d

13C chemical shift (Hz) at 1JCH of 145 Hz0% error 10% error 20% error


0.01

-25000 -12500 0 12500 25000

| | | | |

0.04 0.01

0.07 0.04 0.01


30

miscalibration data is shown in Fig. 21. The intensity ratio of relative satellite amplitude is used

as a vertical scale.

Figure 21: The effect of the B1 miscalibration on signals obtained using composite pulses for a 13C chemical shift range of

50000 Hz, using Spinach package. The intensity ratio of relative satellite amplitude is shown on the right-hand side. Duration of 90 degrees 1H pulse was 0.1 𝜇𝑠, 13C pulse was 15 𝜇𝑠, and τ1, τ2, τ3, and τ4 were 3.2, 1.1, 3.95, and 1.56 ms, respectively.

The figure demonstrates that with 20 % error in B1, the 13C-1H coupled peaks on resonance are

nine times as higher as they are with 0 % error at a chemical shift of -/+ 25000 Hz. With 10%

and 20% errors, the suppression around on-resonance is poorer when compared to hard 90°

pulses. The errors are well compensated by the multiple stages of the 1JCH filter, so suppression

is still more than adequate, as it was for hard pulses.

-0.02

0

0.02

0.04

0.06

0.08

0.1

-30000 -20000 -10000 0 10000 20000 30000

I cou

pled

/ no

t cou

pled

13C chemical shift (Hz) at 1JCH of 145 Hz0% error 10% error 20% error


31

Chapter 3 – 1D DISPEL experiment

3.1 Introduction

An experimental investigation of the DISPEL sequence without a z-filter was performed using

three samples: propanol sample, quinine sample and Q-mix sample. The propanol sample is

used to illustrate that DISPEL works; quinine is used to see if a sequence works efficiently on a

complicated chemical structure; Q-mix is used to see if a mixture of similar compounds can be

identified using the sequence. The z-filter is removed from the DISPEL sequence on Fig. 12 to

increase sensitivity by cutting off the duration of Z-filter. The z-filter element is not needed if

the acceptable multiplets can be acquired. The pulse sequence used for the following

experiments was

Figure 22: The pulse sequence of 1D DISPEL without z- and ZQS-filters.

3.2 DISPEL 1D experiment

3.2.1 Investigation of 13C 90° off-resonance effect using propanol

The carbon off-resonance effects can have a great influence on the degree of 13C-1H coupled

peak suppression. Fig. 15 showed the simulated data, but the experimental data have noise,

so the high S/N ratio is required for satellite peaks to be detected. The simulated and the

experimental figures look different because there are three signals in the experimental data,

a parent peak and a satellite peak on either side. The satellite peaks are small because the

natural abundance of 13C is 1.1%. Fig. 23 shows the experimental data of 13C 90° off-resonance

effect on methoxy signal of Q-mixture sample (quinine, quinidine and cinchonidine mixture)

from -40000 Hz to 40000 Hz with respect to on-resonance for the methoxy signal.

Figure 23: The experimental data of effect of the hard 90° 13C pulses on signals obtained for a 13C chemical shift range of

80000 Hz on 500 MHz. Spectrum was acquired with 128 scans and an experimental time of 2 h 36 min.

-40000 0 40000


0.6


32

The intensity ratio at 1 is when there is no suppression. The methoxy signal has 1JCH of 145 Hz,

the same 1JCH value as in the simulation. At +/- 40000 Hz, the satellite peaks can be seen clearly,

but the satellite peaks are suppressed by about 40% even at +/- 40000 Hz. The influence of the

offset is getting stronger as the spins go further away from resonance. Fig. 24 shows

comparison between simulated and experimental data for 13C satellite intensity ratio. The

intensity ratio of 1 is when there is no suppression.

Figure 24: The experimental data of the effect of the hard 90° 13C pulses on signals obtained for a 13C chemical shift range of

80000 Hz

The height of satellite peaks differed from chemical shift to chemical shift and were not seen

to be completely symmetrical, as they were in simulation. At 40000 Hz, the experimental

suppression was worse than in simulation. However, the general trend seen in experimental

data matched with simulation: the edges of the ranges have poorer suppression. Unlike

simulation, experimental data has noise causing little ups and downs but even at the edges of

the range, DISPEL worked efficiently. The DISPEL sequence is pleasingly robust with respect to

resonance offset.

3.2.2 Investigation of the effect of B1 miscalibration using propanol

B1 miscalibration has a great influence on the degree of the satellite peak suppression as it

causes imperfect flip angles. Fig. 18 showed the Spinach simulated result of 0%, 10% and 20%

errors in B1 over a wide range of the 13C chemical shift and Fig. 25 shows the experimental data.

The intensity ratio of 1 is when there is no suppression.

-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

-50000 -40000 -30000 -20000 -10000 0 10000 20000 30000 40000 50000

Inte

nsity

ratio

13C chemical shift (Hz) at CH coupling of 145 HzExperimental Simulation


33

a)

b)

c)

Figure 25: Experimental data of DISPEL pulse sequence with hard 90° pulse with B1 error of (a) 0 %, (b) 10 %, and (c) 20 %.

It is clear that the satellite peaks are getting taller with an increase in B1 error. Comparison for

processed B1 miscalibration data is shown in Fig. 26. The intensity ratio of 1 is when there is no

suppression.

Figure 26: The experimental data of the effect of the B1 miscalibration on signals obtained for a 13C chemical shift range of

80000 Hz.

The height of satellite peaks differed from chemical shift to chemical shift again with B1 error.

The general trend seemed to follow the simulation. The comparison between simulation and

the experimental data is shown in Fig. 27.

-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

-50000 -40000 -30000 -20000 -10000 0 10000 20000 30000 40000 50000

Inte

nsity

ratio

13C chemical shift (Hz) at CH coupling of 145 HzExperiment - 0 % error Experiment - 10 % error Experiment - 20 % error

-40000 0 40000


0.6

0.6

0.7


34

Figure 27: The comparison between the experimental data and the Spinach simulation data of the effect of the B1

miscalibration on signals obtained for a 13C chemical shift range of 80000 Hz.

Overall, the experimental data had worse suppression than the simulation data, which was

expected. However, the highest intensity ratio was 0.7, so DISPEL is pleasingly robust with

respect to B1 error.

3.3 DISPEL 1D spectrum

3.3.1 1D spectrum of doped 2.5% propanol

This is the structure of propanol:

This sample was made of 2.5% propanol doped with chromium(lll) acetylacetate, in DMSO-d6.

1D 1H spectra, with and without 13C pulses had been acquired. See Fig. 28.

-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

-50000 -40000 -30000 -20000 -10000 0 10000 20000 30000 40000 50000

Inte

nsity

ratio

13C chemical shift (Hz) at CH coupling of 145 HzExperiment - 0 % error Experiment - 10 % error Experiment - 20 % error

Spinach - 0% error Spinach - 10 % error Spinach - 20% error


35

a)

b)

c)

Figure 28: 500 MHz 1H spectrum of n-propanol doped with chromium tris-acetylacetonate in DMSO-d6. (a) without DISPEL (b) without and (c) with 13C pulses the satellite peaks are marked by red arrows.

The 1D spectrum without DISPEL showed four sets of satellite peaks and those are also seen

in DISPEL without 13C pulses. The satellite peaks are marked by red arrows. Fig 28(a) and 28(b)

are expected to be similar. These satellites are suppressed efficiently in Fig. 28(c). This shows

that 1D DISPEL works on a doped chemical like propanol even without a z-filter or ZQS filter.

3.3.2 1D DISPEL spectrum of 100 mM quinine

This is the structure of quinine:

Unlike propanol, quinine is a complicated chemical that contains aromatic rings and many

other functional groups, with a wide range of 13C chemical shift. 99% quinine was used here

without further purification. The 8.36 ppm to 9.0 ppm region of the 1D DISPEL 1H quinine

spectrum with and without 13C pulses is shown in Fig. 29. Satellite peak of interest is marked

𝛿 𝐻" /ppm


36

by a red arrow. A peak that overlaps with one of the satellite peaks is marked by a green arrow.

The overlapped peak is an impurity in the sample.

a)

b)

Figure 29: Expansion of 500 MHz 1H DISPEL spectrum of 100 mM quinine in DMSO-d6 (a) without and (b) with without 13C pulses. Each spectrum was acquired with 16 scans.

DISPEL spectra with an experiment time of 2 min and 11 s.

Another peak around 8.45 ppm is another satellite peak, which was suppressed efficiently

using the DISPEL sequence. This region shows where DISPEL is useful. The satellite peak around

8.8 ppm is close to the non-satellite peak, making identification and quantification challenging.

With satellite peaks removed, the peaks from chemical compound can now be identified.

3.3.3 1D DISPEL spectrum of Q-mix

The Q-mixture solution was made up of 23.5 mM of cinchonidine, 35.3 mM of quinine and 29.4

mM of quinidine in DMSO-d6. Here are the structures of cinchonidine, quinine and quinidine:

The structure of all three compounds is similar so they are expected to have overlapped signals

in a 1D 1H NMR spectrum. The 7.7 to 9.0 ppm region of the 1D 1H Q-mix spectrum with and

without 13C pulses is shown in Fig. 30. Satellite peaks of interest are marked by red arrows.

Peaks that overlap with one of the satellite peaks are marked by green arrows.

𝛿 𝐻" /ppm


37

a)

b)

Figure 30: Expansion of 500 MHz 1H spectrum of Q-mixture in DMSO-d6 (a) without and (b) with 13C pulses. Each spectrum was acquired with 16 scans. A 1H conventional spectrum was acquired with an experimental time of 1 min 8 s

and DISPEL spectrum with an experiment time of 4 min.

The green peak A was also present in the quinine spectra, but is present in higher

concentration, suggesting that the impurity present in quinine may be cinchonidine. Unlike in

the quinine sample, the cinchonidine peak is dominant here, merging a non-satellite peak with

one of the satellite peaks. This spectrum also demonstrates a case where the suppression of

satellite peaks is useful. Because one of the satellite peaks is almost masked by the dominant

peak from cinchonidine, it is easy to mistake another satellite as a non-satellite peak,

complicating the analysis of the spectrum acquired.

𝛿 𝐻" /ppm

𝐴 𝐵


38

Chapter 4 – 2D TOCSY-DISPEL experiment

4.1 Introduction

The DISPEL sequence is shown to work in 1D in the previous chapter; the pulse sequence used

there is concatenated here with a 2D TOCSY sequence. A 180° carbon pulse in the middle of t1

evolution is used to refocus the effect of couplings to carbon-13 in the indirect dimension. The

minimum phase cycle of this variant of 2D TOCSY is 8 steps, and sufficient suppression of

carbon satellites can be achieved without further phase cycling of the DISPEL part. Instead of

phase cycling, gradients31 are used for the DISPEL part. The gradient pulses can be used to

select particular coherence transfer pathways and selection using gradients offers some

advantages and disadvantages when compared to selection using phase cycling. The gradient

pulses are introduced into the pulse sequence in such a way that only the wanted signals are

refocused and observed in each experiment. Therefore, unlike phase cycling, there is no

reliance on the subtraction of unwanted signals, thus, it is expected to have reduced t1-noise.

However, switching on and off a gradient pulse induces currents called eddy currents34 in

nearby conductors. These induced currents generate magnetic fields that perturb the NMR

spectrum in high-resolution NMR probes.

Figure 31:The pulse sequence of 2D TOCSY-DISPEL. τ1, τ2, τ3, and τ4 were 3.2, 1.1, 3.95, and 1.56 ms, respectively.

4.2 2D TOCSY-DISPEL spectrum

4.2.1 2D TOCSY-DISPEL of 2.5% propanol

As seen in 1D DISPEL of 2.5 % propanol in Fig. 28, there are 8 one-bond satellite peaks in this

sample. 2D TOCSY, 2D TOCSY-DISPEL spectra with and without 13C pulses were acquired. See

Fig. 32 for the 2D TOCSY spectrum. The satellite peaks are marked using red arrows.

t1 t2

gt3 gt1 gt3 gt2 gt4 gt4 gt5 gt5


39

Figure 32: 500 MHz 2D TOCSY spectrum of n-propanol doped with chromium tris-acetylacetonate in DMSO-d6. The experiment was acquired with 8 scans and 512 increments in an experiment time of 8 h 53 min. The satellites are

marked by the red arrows.

All 8 satellite peaks (marked with red arrows) were clearly seen in the 2D TOCSY spectrum as

expected. However, the t1-noise showed up as streaks. The t1-noise is caused by instrumental

instabilities, giving (pseudo-)random variations of FID amplitudes during the data acquisition,

which then introduces random noise-like peaks into the spectrum obtained by Fourier

transforming the acquired data. See Fig. 33 for 2D TOCSY-DISPEL spectra with and without 13C

pulses.

a) b)

𝛿 𝐻" 𝛿 𝐻"

𝛿𝐻 "

0.6

1.8

3.0

4.2

0.6

1.8

3.0

4.2

0.6

1.8

3.0

4.2

𝛿𝐻 "

/ppm

4.2 3.0 1.8 0.6 4.2 3.0 1.8 0.6

4.2 3.0 1.8 0.6

𝛿𝐻 "

/ppm

𝛿 𝐻" /ppm

𝛿 𝐻" /ppm

𝛿 𝐻" /ppm

𝛿𝐻 "

/ppm


40

Figure 33: 500 MHz 2D TOCSY-DISPEL spectra of n-propanol doped with chromium tris-acetylacetonate in DMSO-d6, (a) without and (b) with 13C pulses. Each experiment was acquired with 8 scans and 512 increments in

an experiment time of 9 h 27 min. The satellite peaks are marked by red arrows.

The 2D TOCSY-DISPEL spectrum of 2.5% doped propanol shows 8 one-bond satellite peaks, as

marked by red arrows on Fig. 33(b). The 2D TOCSY-DISPEL was expected to give similar

spectrum as convectional 2D TOCSY, however, the t1-noise showed up more severely. With 13C

pulses, the satellite peaks were efficiently suppressed, as seen in Fig. 33(b). To take a look at

these spectra closely, part of the spectrum marked by a purple box is shown expanded in Fig.

34.

a) b)

Figure 34: Expansion of 500 MHz 2D TOCSY-DISPEL spectra marked by a purple square in Fig. 33. (a) without and (b) with 13C pulses. Each experiment was acquired with 8 scans and 512 increments

in an experiment time of 9 h 27 min. The diagonal satellite peaks are marked by red arrows.

The diagonal satellite peaks are marked by red arrows. There are other peaks that are not

marked called cross-peaks. The t1-noise creates the technical challenge of this 2D spectra,

however, the DISPEL suppressed satellite peaks successfully as seen in Fig. 33(b) and 34(b). The

DISPEL worked effectively not only in 1D but in 2D NMR on 2.5 % propanol.

4.2.2 2D TOCSY-DISPEL spectra of 100 mM quinine

As it is seen in Fig. 29, the satellite peaks overlap with a non-satellite peak. The 2D TOCSY-

DISPEL with and without 13C pulses was acquired. The region of interest is expanded here to

see how effectively the DISPEL has worked. The satellite peaks are marked by red arrows and

the non-satellite peak is marked by green arrows.

𝛿 𝐻" /ppm

𝛿 𝐻" /ppm

𝛿𝐻 "

/ppm

𝛿𝐻 "

/ppm


41

a) b)

Figure 35: Expansion of 500 MHz 2D TOCSY-DISPEL spectra of 100 mM quinine in DMSO-d6, (a) without and (b) with 13C pulses. Each experiment was acquired with 8 scans and 512 increments in an experiment time of

12 h 33 min. The satellite peaks are marked by red arrows and the non-satellite peak is marked by green arrows.

The satellite peaks are separated in 2D better than they were in 1D and can be distinguished

from one another. Therefore, this specific area did not need to use DISPEL to help identifying

the peak from quinine in 2D NMR. Nonetheless, the satellite peaks are suppressed; however,

Fig. 35(a) shows a problem - the satellite peaks are not identical in 2D spectrum. There is no

splitting visible in one of the satellites, which may be noise leading to the two doublet

components not being resolved. This could be due to perfect echo being imperfect due to

concatenation with TOCSY sequence. During the TOCSY period, the spins are still evolving, thus,

may not be in the condition that perfect echo is set to work. The t1 noise is seen as a streak

again, too.

4.2.3 2D TOCSY-DISPEL spectra of Q-mix

The Q-mix is made out of three chemical compounds: quinine, quinidine and cinchonidine. As

they were shown in Section 3.3.3, they are similar in molecular shape. The quinine and

quinidine are stereoisomers, whereas cinchonidine does not have methoxy. The full spectrum

of the 2D TOCSY-DISPEL was acquired using hard and composite pulses for the carbon channel.

The region of interest in Fig. 30 is shown here to see if the satellites were successfully

suppressed in 2D TOCSY-DISPEL.

8.36 8.64 8.96

8.96 8.64 8.36 8.96 8.64 8.36

8.36 8.64 8.96

𝛿𝐻 "

/ppm

𝛿𝐻 "

/ppm

𝛿 𝐻" /ppm

𝛿 𝐻" /ppm


42

a) b)

Figure 36: Expansion of 500 MHz 2D TOCSY-DISPEL spectra of Q-mix in DMSO-d6, (a) without and (b) with 13C pulses. Each experiment was acquired with 16 scans and

512 increments in an experiment time of 25 h.

The two regions of interests are marked by orange and yellow squares. To examine these

spectra closely, the portion of the spectrum marked by orange and yellow boxes is shown

expanded in Fig. 37 and in Fig. 38, respectively.

a) b)

Figure 37: Expansion of 500 MHz 2D TOCSY-DISPEL spectra of Q-mix in DMSO-d6, marked by orange square in Fig. 36. (a) without and (b) with 13C pulses. Each experiment was acquired with 16 scans and 512 increments in an experiment time

of 25 h. The satellite peak marked by red arrow is overlapping with the cinchonidine signal marked by a green arrow.

The satellite peak is now removed and this cinchonidine peak is fully resolved. Another satellite

peak seen at the right top corner of the left spectrum was suppressed with the use of DISPEL.

7.70 8.20 8.70

7.70 8.20 8.70

8.70 8.20 7.70 8.70 8.20 7.70

8.40 8.68 8.92

8.40 8.68 8.92

8.92 8.68 8.40

8.92 8.68 8.40

𝛿𝐻 "

/ppm

𝛿

𝐻 "/p

pm

𝛿 𝐻" /ppm

𝛿 𝐻" /ppm

𝛿𝐻 "

/ppm

𝛿𝐻 "

/ppm

𝛿 𝐻" /ppm

𝛿 𝐻" /ppm


43

a) b)

Figure 38: Expansion of 500 MHz 2D TOCSY-DISPEL spectra of Q-mix in DMSO-d6, marked by yellow square in Fig. 36 (a) without and (b) with 13C pulses. Each experiment was acquired with 16 scans

and 512 increments in an experiment time of 25 h. The satellite peak of interest is marked by red arrow.

The satellite peak of interest is marked by red arrow and that satellite peak and the peak one

around 7.5 ppm is not symmetrical. However, this is due to those satellite peaks having

different parent peaks. The composite pulses were used in 2D TOCSY-DISPEL to reduce off

resonance effect. See Fig. 39.

a) b)

Figure 39: Expansion of 500 MHz 2D TOCSY spectra of Q-mix in DMSO-d6, with (a) hard and (b) composite 13C pulses. Each experiment was acquired with 16 scans and 512 increments in an experiment time of 25 h.

7.54 7.68 7.82

7.54 7.68 7.82 7.82 7.68 7.54

7.82 7.68 7.54

7.70 8.20 8.70

7.70 8.20 8.70

8.70 8.20 7.70 8.70 8.20 7.70

𝛿 𝐻" /ppm 𝛿 𝐻" /ppm

𝛿𝐻 "

/ppm

𝛿

𝐻 "/p

pm

𝛿𝐻 "

/ppm

𝛿𝐻 "

/ppm

𝛿 𝐻" /ppm

𝛿 𝐻" /ppm


44

For the Q-mixture sample, the hard pulses were able suppress satellite peaks and showed

almost no difference with the spectrum that used composite pulses. The use of composite

pulses is to overcome off-resonance effects but was not required for Q-mixture.


45

Chapter 5 – Discussion

5.1 Conclusions

In summary, the DISPEL sequence worked on a simple, a complicated and mixture of samples

not only in 1D but also in 2D NMR.

Chapter 2 investigated two factors that affect the suppression of 13C-1H coupled peaks: 13C off-

resonance effects and B1 miscalibration. Simulations demonstrate that suppression over a

wider range of 13C offset can be obtained with the use of composite pulses, and even with B1

errors, efficient suppression can be achieved for a wide range of chemical shift.

Chapter 3 showed that DISPEL spectra without 13C pulses were similar to 1D 1H NMR spectra.

The satellite peaks are removed by DISPEL, helping with identifying and distinguishing non-

satellite and satellite peaks. The experimental data suggested that the suppression is sufficient

with a normal B1 level and even with significant error in B1.

Chapter 4 used three samples to show that the DISPEL works when concatenated with 2D

TOCSY. However, the t1-noise is seen as streaks, making cross-peaks harder to identify. The

composite pulses were also used; however, the sample is not affected by the off-resonance

effect enough for them to be useful.

5.2 Discussion

Composite pulses were used to overcome the 13C off-resonance and B1 miscalibration

effects, but there are other ways to overcome off-resonance effects. Off-resonance effects can

be dealt with by using higher power transmitters that are able to excite over wider bandwidths.

However, it is hard for the probe to sustain such high powers without being damaged, and

sample heating can occur. Therefore, the use of composite pulses to overcome off-resonance

effects is a better approach.33 But there is a limitation in composite pulses as well – that the

sequences which provides the best compensation tend to be the longest and most complex,

always longer than the simple pulses so may not be suitable for use within all the pulse

sequence.24 There is sufficient time in a DISPEL-TOCSY sequence to implement simple

composite pulses, but to implement something more complicated for better compensation of

off-resonance effects would be challenging. Also, composite pulses that have been designed

for a particular initial magnetization state may not perform well, or may give unexpected

results, so when applied to other states. 180° composite pulses can be designed specifically for


46

inversion or for refocusing, in which they act on longitudinal and transverse magnetization

respectively.24 A sequence such as 90y180x90y can provide offset compensation when used as

a refocusing pulse, but it introduces errors in phase that may be harmful to the overall

performance of an experiment. Thus, it is important to check whether a compensated pulse

sequence will actually show improvements over the uncompensated sequence.24 There is

another way to reduce B1 miscalibration, which is to use adiabatic pulses – these are 180° pulse

and give very effective simultaneous compensation for resonance offset and miscalibration,

but swept-frequency 90° pulses do not.35 However, DISPEL does not contain 180° pulses in 13C,

so this cannot be applied in 2D TOCSY-DISPEL.

The graphical figures in chapter 3 show that the general trend seen in experimental

data matched with simulation: the edges of the offset range have poorer suppression. The

experimental data were slightly different from the simulation, as it is expected. This is likely to

be due B1 inhomogeneity, and possibly also to the transient errors in phase and amplitudes

that occur at the starts and ends of pulses, known as phase glitch.36 B1 errors can be partially

compensated by composite pulses, but as can be seen in Fig. 26, this is less effective off

resonance. The phase glitch problem causes phase errors arising from current switching phase

transients in the transmitter coil at both leading and tailing edges of the RF pulse. The idea of

“glitches” came from Ellett et al.37 and Mehring and Waugh.38,39 The phase glitch problem can

be reduced by using longer, lower power pulses, but this would defeat the purpose of

improving performance off resonance.

The t1-noise was one of the problems seen in 2D TOCSY-DISPEL spectra, making cross-

peaks harder to identify. The t1-noise is not a problem that is seen only in DISPEL-TOCSY but

occurs in all nD NMR spectra. In case of the NOESY, co-addition of multiple spectra has been

reported to significantly reduced the t1-noise compared to conventional acquisition with the

same total acquisition time and resolution.40 One factor increasing t1-noise can be long data

acquisition time. Shortening the acquisition time and acquiring many short experiments, then

adding them together can give a reduction in t1-noise. This method could be applied to the 2D

DISPEL-TOCSY sequence. This sequence requires a minimum of 8 phase cycling steps, so 8 scans

per increment would be required. However, this method would only be advantageous when

the number of scans perincrement required to obtain adequate signal-to noise ratio exceeds

the number needed to complete the pulse sequence's phase cycle.40 Breaking a single long 2D


47

acquisition into multiple shorter ones will cost additional disk space and time to process, but

this problem is fairly easy to be solved.

Another method for suppressing t1-noise is called reference deconvolution.41 This is a

powerful processing method for removing t1-noise that affect all peaks in a spectrum in the

same way. The application of reference deconvolution to 2D COSY NMR spectrum has been

illustrated on a tetracyclic orthoamide in deuteriochloroform using TMS as the reference

signal.41 This method improved the spectrum quality, giving a much cleaner and more

informative spectrum. Reference deconvolution could also be applied in 2D DISPEL-TOCSY. The

spectrum of interest and the reference signal are acquired simultaneously, and application of

processing method can be applied. One downside of this method is lower S/N ratio, as the

noise from the experimental reference signal is included in the correction function and

therefore gets convolved into the corrected spectrum.42

5.3 Future work

DISPEL offers a way to deal with some complications in mixture analysis, however, it can be

improved. The t1-noise was seen, as expected in 2D spectra. The t1-noise was more noticeable

because the vertical scale was increased to show the 13C satellites, which are only 0.55% of the

parent peaks. t1-noise is a problem in any 2D experiment looking at such small signals, but it

complicates the cross-peak identification and hinders the identification of the structures of the

unknown chemicals. Reduction of t1-noise is an area to look into further. Also, application of

DISPEL in other 2D method could improve analysis of high dynamic range mixtures.


48

References

1 E. D. Becker, Anal. Chem., 1993, 65, 295A–302A. 2 R. Freeman, Chem. Heterocycl. Compd., 1995, 31, 1004-1005. 3 E. L. Hahn, Phys. Rev., 1950, 80, 580. 4 R. Kimmich, Diagnostic Imaging., 1984, 29, 593 – 596. 5 C. Kittel, Phys. Rev., 1949,76, 743. 6 S. Succi and R. Benzi, Phys. D Nonlinear Phenom., 1993, 69, 327-332. 7 J. Keeler, Understanding NMR Spectroscopy, John Wiley & Sons, Chichester, UK, 2005. 8 K. Dorine, T. Howe-Siang, J. Phys. Chem. A, 1999, 103, 10369-10380. 9 N. E. Jacobsen, NMR Spectroscopy Explained, John Wiley & Sons, Hoboken, USA, 2007. 10 T. J. Horne, G. A. Morris, J. Magn. Reson., 1996, 123, 246-252. 11 M. Balci, Basic 1H- and 13C-NMR Spectroscopy, Elsevier, Ankara, Turkey, 2005. 12 D. S. Wishart, Prog. Nucl. Magn. Reson. Spectrosc., 2011, 58, 62-87. 13 P. Moutzouri, P. Kiraly, A. R. Phillips, S. R. Coombes, M. Nilsson and G. A. Morris, Anal.

Chem., 2017, 89, 11898-11901. 14 J. A. Aguilar, M. Nilsson, G. Bodenhausen and G. A. Morris, Chem. Commun., 2012, 48,

811-813. 15 P. W. A. Howe, Anal. Chem., 2018, 90, 4316-3419. 16 R. A. de Graaf, Magnetic Resonance Spectroscopy: Tools for Neuroscience Research

and Emerging Clinical Applications, John Wiley & Sons, Hoboken, New Jersey, 2013. 17 K. Takegoshi, K. Ogura and K. Hikichi, J. Magn. Reson., 1989, 84, 611-615. 18 T. Parella, Magn. Reson. Chem., 2019, 57, 13-29. 19 P. Moutzouri, P. Kiraly, M. Foroozandeh, A. R. Phillips, S. R. Coombes, M. Nilsson and

G. A. Morris, J. Magn. Reson., 2018, 295, 6-11. 20 B. Bikash and C. L. Khetrapal, J. Magn. Reson., 2014, 242, 143-154. 21 M. J. Thrippleton and J. Keeler, Angew. Chemie - Int. Ed., 2003, 42, 3938–3941. 22 S. Massou, C. Nicolas, F. Letisse and J. C. Portais, Metab. Eng., 2007, 9, 252-257. 23 N. Khaneja, J. Magn. Reson., 2017, 282, 32. 24 T. D. W. Claridge, High-Resolution NMR Techniques in Organic Chemistry, Elsevier

Science, Oxford, UK, 3rd edn., 2016. 25 P. J. Hore, J. A. Jones and S. Wimperis, NMR: The Toolkit, Oxford University Press,

Oxford, UK, 2000. 26 P. J. Hore, Nuclear Magnetic Resonance, Oxford University Press, Oxford, UK, 2015. 27 H. J. Hogben, M. Krzystyniak, G. T. P. Charnock, P. J. Hore and I. Kuprov, J. Magn.

Reson., 2019, 73, 401-409. 28 W. S. Hinshaw and A. H. Lent, Proc. IEEE, 1983, 3, 338-350. 29 M. H. Levitt, Spin Dynamics : Basics of Nuclear Magnetic Resonance, John Wiley &

Sons, New York, 2001. 30 M. A. B.Whitaker & I. Singh, Physics Letters A, 1981, 87, 9-10. 31 A. Ahmet and O, Salim, J. Chem. Eng. Data., 2012, 57, 2619-2624. 32 J. Keeler, Understanding NMR Spectroscopy, John Wiley & Sons, Chichester, UK, 2nd

edn., 2011. 33 R. Freeman, S. P. Kempsell and M. H. Levitt, J. Magn. Reson., 2011, 38, 453-479. 34 O. W. Sørensen, Magn. Reson. Chem., 2006, 46, 1798-1802.


49

35 A. Tannús and M. Garwood, NMR in Biomed., 1997, 10, 423-434. 36 J. D. Ellett et al., J. Chem. Phys., 1971, 5, 117-119. 37 M. Mehring and J. S. Waugh, Rev. Sci. Instrum., 1972, 43, 649-670. 38 R. W. Vaughan, D. D. Elleman, L. M. Stacey, W. K. Rhim, and J. W. Lee, Rev. Sci.

Instrum., 1972, 43,13-56. 39 H. Mo, J. S. Harwood, D. Yang and C. B. Post, J. Magn. Reson., 2017, 276, 43-50. 40 A. Gibbs, G. A. Morris, A. G. Swanson, D. Cowburn, J. Magn. Reson., 1996, 101, 351-

356. 42 H. Barjat, G. A. Morris, A. G. Swanson, S. Smart, S. C. R. Williams, J. Magn. Reson.,

1995, 116, 206-214.

development of novel 2d nmr techniques for mixture analysis

Documents