identification and isolation of drugs in complex...

Identification and Isolation of Drugs in Complex

Samples

By

Niki Kim Burns

B.ForensicSc (Hons)

Submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy

Deakin University

May, 2017

sfol

Retracted Stamp

IV

"All our dreams can come true, if we have the courage to pursue them."

- Walt Disney

V

TABLE OF CONTENTS

TABLE OF CONTENTS ........................................................................... V

ABSTRACT .............................................................................................. IX

ACKNOWLEDGEMENTS .................................................................... XII

PUBLICATIONS ................................................................................... XIV

ABBREVIATIONS ................................................................................. XV

CHAPTER 1: ............................................................................................... 1

INTRODUCTION ....................................................................................... 1

Complex samples in context .................................................................... 2

Opiates ................................................................................................. 2

Synthetic cannabinoids ........................................................................ 6

Separation and Detection methods ........................................................ 10

High Performance Liquid Chromatography (HPLC) ......................... 11

Two-Dimensional High Performance Liquid Chromatography (2D-

HPLC) ............................................................................................................ 12

Other applied separation techniques .................................................. 13

Chemiluminescence ........................................................................... 14

Electrochemiluminescence (ECL) ..................................................... 16

Mass Spectrometry (MS) ................................................................... 17

Nuclear Magnetic Resonance Spectroscopy (NMR) ......................... 18

Data Analysis ......................................................................................... 18

Project aims ........................................................................................... 19

CHAPTER 2: ............................................................................................. 21

BLIND COLUMN SELECTION PROTOCOL FOR TWO-DIMENSIONAL HIGH PERFORMANCE LIQUID CHROMATOGRAPHY... 21

Chapter Overview .................................................................................. 22

Introduction ............................................................................................ 23

Experimental .......................................................................................... 28

VI

Chemicals and samples ...................................................................... 28

Liquid Chromatography Mass Spectrometry (LC/MS) ..................... 28

Two-Dimensional High Performance Liquid Chromatography ........ 30

Data analysis ...................................................................................... 30

Results and discussion ........................................................................... 32

Conclusions ............................................................................................ 51

CHAPTER 3: ............................................................................................. 52

OPIATE PROCESS STREAM ANALYSIS ............................................ 52

Chapter overview ................................................................................... 53

Introduction ............................................................................................ 54

Experimental .......................................................................................... 56

Chemicals and samples ...................................................................... 56

Two-Dimensional High Performance Liquid Chromatography ........ 56

Principal Components Analysis (PCA) .............................................. 56

Liquid Chromatography/ Mass Spectrometry (LC/MS) .................... 58


Principal Components Analysis (PCA) .............................................. 63

Mass Spectrometry ............................................................................. 72

Conclusions ............................................................................................ 76

CHAPTER 4: ............................................................................................. 77

APPLICATION OF PLATFORM TECHNOLOGY TO MONITOR COMPONENTS OF INTEREST .......................................................................... 77

Chapter overview ................................................................................... 78

Introduction ............................................................................................ 79

Molecule of Interest 1 (285 amu) ....................................................... 80



Experimental .......................................................................................... 84

Case studies ........................................................................................ 84

VII


Molecules of interest .......................................................................... 85

Case 1: G1527-027N, G1527-027O and G1527-027G. ..................... 92

Case 2: G1527-027L and G1527-027M........................................... 100

Case 3: G1528-028C, G1527-028D and G1527-028H .................... 104

Conclusions .......................................................................................... 110

CHAPTER 5: ........................................................................................... 111

IDENTIFICATION AND SELECTIVE DETECTION OF SYNTHETIC CANNABINOIDS ............................................................................................... 111

Chapter overview ................................................................................. 112

Introduction .......................................................................................... 113

Solid Phase Extraction ..................................................................... 114

Experimental ........................................................................................ 115

Chemicals, samples and standards ................................................... 115

High Performance Liquid Chromatography (HPLC) ....................... 116

Chemiluminescence detection .......................................................... 117

Electrospray mass spectrometry ....................................................... 117

Electrochemiluminescence (ECL) and Cyclic Voltammetry (CV) .. 118

Solid Phase Extraction (SPE) ........................................................... 118

Results and discussion ......................................................................... 120

Synthetic cannabinoid extraction ..................................................... 120

Identification .................................................................................... 122

Chemiluminescence detection .......................................................... 125

Electrochemiluminescence (ECL) and Cyclic Voltammetry (CV) .. 134

Solid Phase Extraction ..................................................................... 139

Conclusions .......................................................................................... 144

CHAPTER 6: ........................................................................................... 145

TARGETED NMR SPECTROSCOPY FOR SELECTIVE DETERMINATION OF SYNTHETIC CANNABINOIDS ............................... 145

VIII

Chapter overview ................................................................................. 146

Introduction .......................................................................................... 147

Experimental ........................................................................................ 150

Samples ............................................................................................ 150

Instrumentation and Experiments .................................................... 150

Results and discussion ......................................................................... 151

Conclusions .......................................................................................... 160

Future Work ............................................................................................ 161

REFERENCES ........................................................................................ 163

IX

ABSTRACT

Complex samples are commonly problematic in chemistry and pose many

challenges in both forensic science and natural product extractions. As

developments in analytical instrumentation affords increases in sensitivity,

complex samples become more of an issue as previously undetectable levels of

chemical components can now be identified. There is a demand therefore driving a

need for improved protocols to deal with the comprehensive data generated from

this superior technology. This allows the analysts to observe previously

undetectable analytes, while limiting the impact of the complexity of data analysis.

As such, investigations into new methodologies are warranted for the isolation and

identification of drugs in complex samples in both forensic and pharmaceutical

applications.

Synthetic cannabinoids are a relatively new class of illicit substance and due

to the large numbers of emerging compounds, these require attention to aid law

enforcement agencies. Consequently, in this thesis these compounds have been

isolated and identified from within complex herbal substrates using a range of

optimised novel sample preparations and analytical instruments. Likewise,

methods for mapping the Sun Pharmaceutical Industries Australia opiate

processing streams were developed, and peaks of interest between stages of the

process were identified to meet a genuine industrial problem for this multi-national

company. Changes in the process and the subsequent impact of the molecules

present was then investigated with the aid of mass spectrometry. This was carried

out to inform the company of origins of impurities and major differences occurring

in the extraction line so that better response times to process adjustments can be

made.

Two-dimensional high performance liquid chromatography (2D-HPLC)

was utilised for the separation of opiate extractions from opium poppy (Papaver

Somniferum). Multidimensional chromatographic separations require the selection

of two orthogonal columns; this is a labour intensive and time consuming process,

which in many cases is an entirely trial-and-error approach. This thesis introduces

a novel blind optimisation method for column selection of a black box of

constituent components. A data processing pipeline, created in the open source

X

application OpenMS®, was developed to map the components of equal mass within

the mixture across a library of HPLC columns; 2D-HPLC separation space

utilisation was compared by measuring the fractional surface coverage, fcoverage. It

was found that for a test mixture from an opium poppy (Papaver somniferum)

extract, the combination of diphenyl and C18 stationary phases provided a

predicted fcoverage of 0.44 and was matched with an actual usage of 0.48. OpenMS®,

in conjunction with algorithms designed in house, have allowed for a significantly

quicker selection of two orthogonal columns, which have been optimised for a 2D-

HPLC separation of crude extractions of plant material.

Principal components analysis (PCA) loading plots were used for the first

time to elucidate key differences between two-dimensional high performance liquid

chromatography (2D-HPLC) fingerprint data from samples collected from stages

along the Papaver somniferum industrial process chemistry workflow. Data

reduction was completed using a 2D-HPLC peak picking approach as a precursor

to chemometric analysis. Comparisons of the final stages of product extraction as

an example, PCA analysis of characteristic 2D-HPLC fingerprints accounted for

84.9% of variation between the two sample sets measured in triplicate, with 64.7%

explained by PC1. Loadings plots of PC1 on each sample set identified where the

significant changes were occurring and normalised bubble plots provided an

indication of the relative importance of each of these changes. These findings

highlight 2D-HPLC with appropriate chemometric analysis as a useful tool for the

exploration of bioactive molecules within biomass.

Two-dimensional HPLC and PCA was utilised for the first time here to

develop a map of the morphine and thebaine/oripavine processing streams. In total

68 samples were analysed, including the main processing line, waste streams and

crude impurity standards. The major changes between crucial steps in the

processing were investigated with the biggest changes identified by mass.

Impurities of interest were also investigated to find if they are a consequence of the

extraction process or originate in the natural plant product. These findings will

assist the company in quality control and understanding of the significant impurity

changes throughout the process.

XI

Measures to control the expanding use of synthetic cannabinoids demand

new analytical methodology to identify and determine compounds within this

rapidly evolving class. Seven synthetic cannabinoids (AM-2201, UR-144, XLR-

11, A796,260, 5F-AKB-48, PB-22 and 5F-PB-22) were identified to be present in

eleven herbal products sold in Victoria, Australia, prior to their ban in 2014, using

a combination of GC-MS, HPLC, ESI-MS, and NMR. An additional cannabinoid

in a twelfth product was not detected but not identified. In aid of this work nine

synthetic cannabinoids standards (AM-2201, AB-001, 5F-AB-001, JWH-018,

JWH-073, 5F-AKB-48, 5F-AB-PINACA, 5F-PB-22 and PB-22) were synthesized

or purchased. Furthermore, chemiluminescence detection of synthetic cannabinoids

was investigated using three commonly used reagents: acidic potassium

permanganate, colloidal manganese(IV), and tris(2,2′‐bipyridine)ruthenium(III).

Using the permanganate reagent, no chemiluminescence signal was obtained, but

the manganese(IV) and tris(2,2′‐bipyridine)ruthenium(III) reagents gave

analytically useful responses with all synthetic cannabinoids under investigation

except 5F-AKB-48 and 5F-AB-PINACA. Calibration curves for PB-22, 5F-PB-22,

AM-2201 and 5F-AKB-48, prepared using HPLC with UV absorbance and/or

chemiluminescence detection, were used to determine the total cannabinoid content

extracted with methanol (1 mL) from six of the herbal products (10 mg), which

ranged from 0.082 to 0.77 mg. Limits of detection were determined for all standards

with both UV absorbance and/or chemiluminescence detection, which ranged from

1 × 10-7 M to 9 × 10-4 M. A solid phase extraction study was conducted to further

isolate the synthetic cannabinoid compounds. Six cartridges were trialled and all

were found to clean up the sample, however a C18 silica cartridge proved superior.

Solid-state 13C and 19F NMR spectroscopy has been used for the first time

to detect forensically relevant synthetic cannabinoids on the surface of herbal

substrates. The 13C spectra showed narrow peak widths suggesting that the

synthetic additives are deposited in an ordered, potentially crystalline, phase.

Importantly, the use of solid-state NMR, particularly 19F, offers a relatively fast,

non-destructive, selective detection protocol for synthetic cannabinoids on complex

samples and has potential for other applications such as the direct detection of

pesticides on plants.

XII

ACKNOWLEDGEMENTS

I would like to acknowledge a number of people for all of the guidance and

support they have provided throughout my candidature.

Firstly, to my primary supervisor Dr. Xavier Conlan I would not have done

a PhD if you hadn't offered me this great opportunity and interesting project. I can't

thank you enough for all your support, encouragement and help throughout the

years. We have had some great laughs and hopefully there will be many more

conferences to come!

To my associate supervisor, Alfred Deakin Professor Neil Barnett I am

grateful for the words of wisdom, encouragement and support. Thank you for your

part in convincing me to do a PhD. I will miss the Lakehouse lunches.

To my other associate supervisor Dr. Paul Stevenson, thank you for your

support and help with driving the data analysis portion of my thesis. The data would

not have made sense without your help and it is greatly appreciated. The last few

months without you being down the hall were definitely tough!

To Professor Paul Francis, thank you for all your help with

chemiluminescence and ECL. It has been a great learning experience and I really

appreciate your significant input into the development of that work.

To Sun Pharmaceuticals (formerly GSK) thank you for the opportunity to

participate in such an interesting project. To my external co-supervisor Dr. Stuart

Purcell your wealth of knowledge has been very helpful and you time much

appreciated. Thank you also to Fiona Fry and Tim Bowser for your guidance and

support throughout my candidature.

Thank you to Dr. Zoe Smith for all your assistance over the years. All your

help with the weird and unique instrument problems and chemiluminescence was

much appreciated. Your words of wisdom, guidance and support when needed have

been incredibly valuable.

Thank you to Dr. Jim Pearson for your part in the work presented in

Chapters 5 and 6. I appreciate the time you have taken and input you have had into

XIII

the project. To Dr. Luke O'Dell, Haijin Zhu and Dr. Fred Pfeffer thank you for all

you help with the solid state NMR. The opportunity to use this technique and the

outcomes achieved have been an enjoyable experience.

To Dr. Luke Andrighetto you have been a great desk buddy, the laughs

definitely made the years easier. To Luke and the soon to be Drs Ashton Theakstone

and Natalie Gasz, lunch was always a highlight of the day. Thanks for your

friendship and listening to me vent when nothing was working. We have had some

great times and adventures that I'll never forget!

To the Deakin analytical group past and present, thank you for being fun

and easy to work with and helping shape the memories from lab days to

conferences, it has been a great ride!

A heartfelt thank you to my family especially Mum, Dad, Chrystal, Nan and

Pop for always believing in me and encouraging me to follow my dreams. I could

not have made it this far without your unconditional love and support.

And last but definitely not least my partner in crime, Kris. Thank you for

always being there to support me and listen to me vent. Your love, hugs and ability

to make me forget about bad days have been priceless and I could not have done it

without you!

XIV

PUBLICATIONS

Listed below are references to publications that have resulted from the work presented in this thesis:

Niki K. Burns*, Luke M. Andrighetto*, Xavier A. Conlan, Stuart D. Purcell, Neil W. Barnett, Jacquie Denning, Paul S. Francis, and Paul G. Stevenson, Blind column selection protocol for two-dimensional high performance liquid chromatography, Talanta, 154, 85-91 (2016) DOI: 10.1016/j.talanta.2016.03.056, (*Co-first authors)

This paper is a result of the work presented in Chapter two of this thesis.

Paul G. Stevenson*, Niki K. Burns*, Stuart D. Purcell, Paul S. Francis, Neil W. Barnett, Fiona Fry, Xavier A. Conlan, Application of 2D-HPLC coupled with principal component analysis to study an industrial opiate processing stream, Talanta, 166, 119-125 (2017) DOI: 10.1016/j.talanta.2017.01.044, (*Co-first authors)

This paper is a result of the work presented in Chapter three of this thesis.

Niki K. Burns, Trent D. Ashton, James R. Pearson, I. Leona. Fox, Frederick M. Pfeffer, Paul S. Francis, Zoe M. Smith, Neil W. Barnett, Lifen Chen, Jonathan M. White and Xavier A. Conlan, Extraction, identification and detection of synthetic cannabinoids found pre-ban in herbal products in Victoria, Australia, Forensic Chemistry, Submitted: 28/02/2017

This paper is a result of work presented in Chapter five of this thesis.

Niki K. Burns, Haijin Zhu, Luke A. O’Dell, James R. Pearson, Fred M. Pfeffer and Xavier A. Conlan, Identification of synthetic cannabinoids on herbal substrates by solid-state NMR, The Analyst, Submitted: 2017

This paper is a result of the work presented in Chapter six of this thesis.

Other publications not presented in this thesis include:

Luke M. Andrighetto*, Niki K. Burns*, Paul G. Stevenson, James R. Pearson, Luke C. Henderson, Christopher J. Bowen and Xavier A. Conlan, In-silico optimisation of two-dimensional HPLC for the determination of Australian methamphetamine seizure samples, Forensic Science International. 266, 511-516 (2016) DOI: 10.1016/j.forsciint.2016.07.016, (*Co-first authors)

XV

ABBREVIATIONS

2D-HPLC Two-Dimensional High Performance Liquid

Chromatography

µm Micrometre

ACN Acetonitrile

amu Atomic mass unit

API Active Pharmaceutical Ingredient

Au Absorbance units

CE Capillary Electrophoresis

CHS Cannabinoid Hyperemesis Syndrome

CP Cross Polarisation

CPS Concentrate of Poppy Straw

CV Cyclic Voltammetry

DAD Diode Array Detector

DMF Dimethylformamide

DMSO Dimethylsulfoxide

DNP Dynamic Nuclear Polarisation

ECL Electrochemiluminescence

ESI-MS Electrospray Ionisation-Mass Spectrometry

FC Ferrocene

FIA Flow Injection Analysis

GC Gas Chromatography

GC/MS Gas Chromatography Mass Spectrometry

GSK GlaxoSmithKline

XVI

HPLC High Performance Liquid Chromatography

LC Liquid Chromatography

LC/MS Liquid Chromatography Mass Spectrometry

LOD Limit of Detection

M Molar/ moles per litre

MAS Magic Angle Spinning

Mn(IV) Manganese(IV)

MS Mass Spectrometry

MSD Mass Selective Detector

MW Molecular Weight

m/z Mass to charge ratio

NMR Nuclear Magnetic Resonance spectrometry

PCA Principle Components Analysis

PMT Photo Multiplier Tube

ppm Parts per million

psi Pounds per square inch

QCA Quality critical attributes

QCP Quality critical parameters

RT Retention Time

[Ru(bipy)3]3+ Tris(2,2-bipyridine)ruthenium(III))

SDS Safety Data Sheets

SPE Solid Phase Extraction

SPIA Sun Pharmaceutical Industries Australia

THC Tetrahydrocannabinol

1

CHAPTER 1:

INTRODUCTION

2

Complex samples in context

Complex samples have long been an issue in analytical chemistry and are

increasingly common due to improvements in instrument sensitivity [1]. Highly

sensitive methods mean that increasingly low amounts of analyte can be determined

leading to identification of previously undetermined contaminates. The isolation of

drugs from complex samples is a priority in both forensic and pharmaceutical

disciplines, and can be particularly difficult when samples are obtained from natural

plant sources [2, 3]. Natural product extractions for purposes such as

pharmaceuticals and food may still contain numerous components, requiring

multiple steps to purify the target analytes [3]. With the improvement in impurity

profiling, demands are placed on developing protocols to remove these compounds.

Also known for sample complexity, testing in the forensic discipline ranges from

seized crude drugs [4] to biological samples [5] where the target compound is only

a small portion of the sample. Seized drugs often contain excipients, precursors and

bulking agents in addition to the main drug which demands the analytical

techniques be refined to allow efficient analysis of these and other complex

samples.

This thesis examines two types of drug sample matrices in detail: opiates

from poppy straw obtained from Sun Pharmaceuticals and synthetic cannabinoids

sprayed onto herbal substrates, in conjunction with Victoria Police Forensic

Services Department. Sun Pharmaceuticals, formerly known as GlaxoSmithKline

are a pharmaceutical company that employ a complicated multistep extraction

process to isolate opiates of commercial interest. Synthetic cannabinoids have

relatively recently been banned in Victoria [6], prompting Victoria police to

investigate these complex samples in order to develop robust separation and

identification protocols.

Opiates

The opium poppy, thought to have originated in Asia Minor, is one of the

oldest plants recorded for medical use [7] with evidence suggesting that it could

have been used as a pain reliever or as a drug of abuse for thousands of years. The

opium poppy has been referenced in Sumerian clay tablets dating back to 3000BC

3

and in Greek mythology in 850BC [7]. Archaeological evidence has also

discovered their presence from the Minoan civilisation, dating back to the 14th

century BC [8].

Alkaloids can be extracted from many different poppy species including

Papaver Bracteatum, Papaver Rhoeas and Papaver Setigerum, however, Papaver

Somniferum is the species cultivated for opium production as it produces the most

viable amount of opiate products [9]. The opium poppy plant produces natural

opiate alkaloids, such as morphine and thebaine and in 1806 Sertürner published

his paper on isolating the first alkaloid (morphine) from the opium poppy [10].

After morphine was isolated, other opiates (noscapine, papaverine and thebaine)

were identified with codeine being isolated in 1833 [10]. Morphine is the highest

concentrated opiate in the species Papaver somniferum making this an industrially

significant plant. Metabolically engineered strains of the plant have been developed

[11] allowing a crop of poppy plants to selectively produce more of specific opiates.

This has been achieved with selection [12], cross pollination [13], mutation

breeding [14, 15] and more recently metabolic engineering [16]. Production of

thebaine and oripavine are among the opiates that have expanded from poppy

hybrids [14].

The industrial scale extraction of these alkaloids from the plant is

commonly done one of two ways. The traditional way involves slicing the capsule

of the opium poppy and collecting the gum (latex) that oozes out to seal the wound

and is typically collected 12 - 24 hours later. The latex is then air dried over

approximately 10 days before the crude opium is stirred every half an hour over a

1-3 week period in the sun [7], before following a similar method to the following

technique for isolation of morphine. This method is still used in northern Indian

regions however it is outdated and inefficient and as such is not used in modern

processing [7]. The most common extraction method is the use of poppy straw,

where 83% of the morphine and 91% of thebaine manufactured worldwide utilise

this approach [17]. The poppy straw contains the upper stalk of the plant including

the crushed capsule that has been dried and made into straw pellets. These straw

pellets are subjected to a water and organic solvent extraction process to isolate the

pharmaceutically beneficial opiate compounds in bulk. This method is utilised by

Sun Pharmaceutical Industries Australia for their morphine and thebaine crops,

4

with other target opiates being extracted and further processed for purification. The

four common alkaloids morphine, codeine, oripavine and thebaine are of interest

in this thesis and are illustrated in Figure 1.1.

a) b)

c) d)

Figure 1.1: Structures of the four major opiates a) Morphine, b) Codeine, c) Oripavine and d) Thebaine.

Morphine, the main opiate found in the opium poppy (Papaver

Somniferum), is used as a pain reliever or analgesic for a variety of different types

of pain and is commonly used in a hospital setting to bring effective relief from

severe pain [18]. Common side effects of morphine include sedation, euphoria,

constipation and a lowered blood pressure [7] however the euphoric and addictive

effect of this drug has also led to the abuse of this medicine and consequently it is

now a controlled substance along with other opiates [19]. Morphine is commonly

5

harvested as a precursor to the illicit drug heroin (diacetylmorphine), a compound

that is also legally manufactured in the UK for medical purposes to treat severe pain

in particular for terminally ill patients, during caesarean sections and other major

surgeries [20].

Codeine is a commonly prescribed analgesic and is found in cold and flu

medication or strong pain relief tablets and is suited to relieve mild to moderately

strong pain [7]. Oripavine is extracted from an opium poppy that has been

genetically modified to produce the opiate in a larger quantity and is then used to

produce other pharmaceutical drugs such as buprenorphine and naltrexone [21].

Oripavine is unable to be used as an analgesic despite having an effect comparable

to morphine, due to its toxicity [22]. The difference in toxicity is noted in safety

data sheets (SDS), where the LD50 of morphine is 250 mg/kg (rat) compared to

oripavine with an LD50 of 26.1 mg/kg (intraperitoneal, rat). Likewise, thebaine is

unable to be utilised due to its high toxicity (LD50 of 114 mg/kg, rat, oral) that

causes strychnine-like convulsions [23, 24]. Hence it is extracted from the opium

poppy as a precursor to other pharmaceutically important drugs such as oxycodone,

oxymorphone, naltrexone, nalbuphine, buprenorphine and etorphine [25].

Opium poppies are grown on a large scale in countries including Australia,

India, and Turkey for the pharmaceutical industry. The multinational company Sun

Pharmaceutical Industries are a medicinal based company responsible for the

production of morphine and other opiates used in pharmaceuticals. Heritage

companies to Sun Pharmaceutical Industries Australia (SPIA) have been operating

in Australia since 1886. In 1964, Glaxo began growing opium poppies in Tasmania

for the extraction of morphine and other related opioid medications [26]. Since then

the company has grown significantly, contributing to the global supply of opiate

pharmaceuticals and in 1995 the company merged to form Glaxo Wellcome and

more recently GlaxoSmithKline (GSK) (2000-2015). In 2012 GSK invested

AUD$54 million to research and development [26] with trials in Victoria to

determine the viability of poppy production on mainland Australia. In 2015 Sun

Pharmaceuticals purchased the opiate section from GSK and since then has

continued with the cultivation and processing of important opiate alkaloids. The

core opiates isolated by Sun Pharmaceuticals are morphine, codeine, oripavine and

thebaine (Figure 1.1) for use in various medicinal products or as a precursor to

6

many important semi-synthetic opiates. As these bulk opiates are being utilised in

consumable products it is important to monitor the process for impurities. A

mapping technique for the extraction process will also help to inform the company

of what is occurring at critical points in the stream thus facilitating quality control.

In general mapping techniques allow visualisation of changes by using the data

collected to identify differences between data sets or in this case, each step of

manufacturing processes.

Synthetic cannabinoids

Synthetic cannabinoids, originally based on naturally occurring

cannabinoids, were developed as therapeutic agents and possible alternatives to

THC [27, 28]. However, these failed to diminish the side effects already commonly

observed from the natural cannabis plant. The first instances of these molecules

were published in academic journals after development and investigation by

pharmaceutical industries and researchers in the 1960’s [29, 30]. John W Huffman,

the inventor of several commonly seized synthetic cannabinoids is also the name

sake of the compounds starting with JWH, such as JWH-018 [31, 32]. Other

compounds are either named after their inventor, their base structure or the class in

which they belong. Examples of this include CP-47,497 which whilst invented by

Pzifer, is named after its cyclohexylphenol base structure, as are other synthetic

cannabinoids described with the CP prefix [29].

Since the early 2000s in Australia [27, 33] a large number of products

marketed as "legal highs" were readily available in herbal shops and adult stores.

However, in 2008 these herbal substrates were tested and found to contain

previously reported synthetic cannabinoid compounds as the active ingredient [29,

34]. These products consisted of herbal substrates that had been laced with one or

more synthetic cannabinoid compounds and were then smoked in a similar manner

to cannabis. These chemicals are commonly white powders or oils that are

dissolved into acetone or a similar organic solvent and sprayed onto the herbal

substrates [35, 36]. These substrates are commonly made from the leaves, bark,

flowers and roots of plants such as damiana (Turnera Diffusa), marshmallow leaf

(Althaea officinalis) and mullein leaf (Verbascum Thapsus) [37, 38].

7

To date hundreds of different synthetic cannabinoid structures have been

identified making the detection and identification of these compounds a difficult

task, and law enforcement agencies are in need of improved protocols [27, 39]. A

common way to avoid the legislation against the compounds was to make analogues

or change the structures in such a way that they were no longer illegal. As is the

case with many of the compounds, an example being UR-144 and XLR-11 where

the latter has only the addition of a fluorine on the end of the alkyl chain (Figure

1.2). This is a common analogue along with changes to the aromatic rings or the

addition of extra carbons on the alkyl chain. This has made the detection of these

difficult as not all structures have the same properties, prompting a push to find

alternative/new analytical methods that can selectively detect these compounds.

Figure 1.2: Structures of UR-144 and its fluoro pentyl analogue XLR-11.

To further complicate the chemical fingerprint of these products, fragrances

and flavourings such as menthol, strawberry, mango and pineapple are added to

some brands to give a higher appeal to the purchaser. These products, while

marketed as legal highs, were sold under the radar as potpourri and incense and

their devastating impact on society is illustrated in Figure 1.3 [40-42]. Most

products contain labels stating the herbs were ‘not for human consumption’ and

‘for aromatherapy use only’ with some blends including warnings of the potential

to get a high, as is illustrated in Figure 1.4 with the brand Cloud 9.

8

Figure 1.3: A snapshot of news headlines relating to the emergence of synthetic cannabinoids and the effects caused by these drugs.

9

Figure 1.4: Warning label on the brand Cloud 9 stating “WARNING: This product is intended to be used as incense, for Aromatherapy use only. The deliberate smoking of

this product may result in you getting high and experiencing hallucinations or delusions.”

Synthetic cannabinoids have a similar effect in the body as

tetrahydrocannabinol (THC) as these compounds interact with the same chemical

receptors in the brain, Cannabinoid 1 (CB1) and Cannabinoid 2 (CB2) [28, 43-45].

A series of emergency department case studies have found users suffer from

anxiety, tachycardia, tachypnoea, hypertension, hallucinations or psychosis and

seizures [46]. Hopkins and Gilchrist [47] reported a link between these compounds

and cannabinoid hyperemesis syndrome (CHS). A common problem found in

heavy marijuana smokers, CHS is believed to be caused by an over stimulation of

the CB 1 receptor [47]. As these products gained popularity, more cases of adverse

side effects were reported with numerous deaths and over doses recorded as

illustrated in Figure 1.3 by a snapshot of Australian news headlines [48]. Fatal

intoxication associated with these synthetic drugs has occurred due to organ failure,

where in particular the kidneys and heart are irreparably damaged [49]. A

newspaper article in the Geelong Advertiser in December 2013 reported on three

people in the Geelong Hospital intensive care unit suffering with organ failure after

using synthetic cannabinoids and numerous others that had reported to emergency

in the preceding weeks [41].

As synthetic drugs such as cannabinoids and cathinones are a new craze,

statistics are yet to reflect the full extent that these products are having on society.

In 2013 the National Drug Strategy Household Survey (NDSHS) [50] included new

psychoactive drugs, including synthetic cannabinoids, into the survey for the first

time. This survey found that 1.3% of Australians aged 14 and over had used

10

synthetic cannabis in the past 12 months, with the most common age group being

the 14-19 year old range. The Australian data from the Global Drug Survey found

that in 2014 synthetic cannabis was the 20th most used drug with 4.1% of

respondents having used the drug in the last 12 months [51]. Even more alarming

is the findings of the 2016 Global Drug Survey [51] where it was found that for the

fourth year in a row synthetic cannabinoids have topped the list of drugs ‘leaving

you most likely to need emergency medical treatment’. The report found that 1 in

30 people who had used in the last year required emergency medical treatment.

This is a significant number for a drug that has only been around for 13 years and

highlights the importance of controlling such dangerous substances.

In May 2014, a ban was put in place in Victoria, Australia to prevent the

purchase, sale and possession of 13 classes of synthetic cannabinoids that included

any analogues of these compounds that fall into one of the specified classes under

an amendment to the Drugs, Poisons and Controlled Substances Act 1981 [6]. This

new legislation took into account the previous prohibition of 8 synthetic

cannabinoids in 2011 [52], and introduced the analogue approach which targeted

small changes to the molecules that had previously led to the evasion of the initial

ban. As these compounds are still an issue in Victoria further action is in progress

with the Drugs, Poisons and Controlled Substances Miscellaneous Amendment Bill

2017 that introduces a blanket ban of all ‘psychoactive substances’ as was detailed

in a press release in March 2017 [53], however the bill is still progressing through

parliament. Due to these recent bans methodology for the detection of these

compounds needs to be developed.

Separation and Detection methods

Separation and detection methods for complex samples such as natural plant

products and synthetic mixtures typically involve techniques including

chromatography and spectroscopy. Synthetic cannabinoids and opiates have been

detected with methods such as Mass Spectrometry (MS) [54], Chemiluminescence

[55] and Nuclear Magnetic Resonance (NMR) spectroscopy [3] post separation

with analytical systems such as High Performance Liquid Chromatography (HPLC)

11

[17, 37], Gas Chromatography (GC) [56] and Capillary Electrophoresis (CE) [57].

A summary of each of these important analytical instruments, including detail about

HPLC, is outlined below with a targeted summary given in the introduction section

of relevant results chapters.

High Performance Liquid Chromatography (HPLC)

High Performance liquid chromatography (HPLC) is a major tool in

analytical chemistry and contemporary science for the separation of a vast array of

compounds [58]. The applications of this already popular technique has steadily

increased in recent times as a review by Gika et al. in 2015 highlighted, by the

increase in the number of papers published utilising HPLC (Figure 1.5) [58].

Reverse phase is the most common HPLC system where a polar mobile phase and

a non-polar stationary phase are utilised to separate chemical components [59]. This

chromatography technique separates analytes based on their hydrophobicity due in

the case of the C18 stationary phase for example to the methylene selectivity of the

reverse phase column, where polar compounds are eluted first followed by non-

polar compounds [60].

Figure 1.5: An illustration by Gika et al. describing the number of papers published utilising high performance liquid chromatography [58] (where liquid chromatography and one of the four traces in the graph were searched in the Scopus database in 2014).

12

Reverse-phase HPLC has been used previously for the separation of

synthetic cannabinoids with the majority of published studies utilising solvent

gradients of either acetonitrile or methanol in water, with modifiers such as formic

acid [54]. Other solvent systems employed for the determination of synthetic

cannabinoids include a gradient of 1:1 methanol: acetonitrile in water [38], and an

isocratic aqueous sodium acetate buffer [61].

Reverse phase HPLC is also the method of choice to analyse opiates

including morphine, thebaine and buprenorphine [62]. Acevska et al. developed

and validated a reverse phase HPLC method for the resolved separation of six

targeted opiates [17]. Methods developed for Sun Pharmaceutical Industries

Australia (formerly GSK) utilise reverse phase HPLC for the analysis of the

alkaloid processing streams [63].

Two-Dimensional High Performance Liquid Chromatography (2D-HPLC)

In some cases the separation power of a one dimensional system is not

enough to resolve all components of a sample [64]. In this case two-dimensional

High Performance Liquid Chromatography (2D-HPLC) has the ability to achieve a

greater separation power by increasing the peak capacity [65]. Two-dimensional

High Performance Liquid Chromatography is a technique that utilises two

orthogonal (different retention mechanism) columns to achieve a greater separation

potential [66]. This technique can be performed in either offline or online methods.

Offline 2D HPLC requires the first dimension to be run and the fractions collected

into vials; subsequently the samples in these vials are then manually reinjected into

the second dimension [67]. The online method can be done comprehensively or via

heart cutting, where both utilise a switching valve and sample loops to store

fractions before transferring these fractions into the second dimension as illustrated

in Figure 1.6 (a diagram of the system used by Stevenson et al. at Deakin

University) [68]. In comprehensive 2D-HPLC the whole first dimension is

transferred into the second dimension whereas in heart cutting only selected peaks

are forwarded to the second dimension [64]. Online 2D-HPLC is a quicker and

more efficient method than offline, however it requires the second dimension to

have a fast flow rate as, in order to avoid peak overlap, the second separation must

be complete before the first dimension re-fills the sample loop [67].

13

Figure 1.6: Two-position, eight-port switching valve configuration utilised in 2D-HPLC separations as described by Stevenson et al. [68]

Other applied separation techniques

Other separation techniques that have been applied to the substrates

analysed here include gas chromatography and capillary electrophoresis, however

they have not been used in this thesis due to the limitations outlined in this section.

Gas chromatography (GC) is a separation technique commonly used for the

analysis of a wide variety of volatile compounds. The use of GC is widespread in

chemistry with applications in forensics [5, 69, 70], pharmaceuticals [71, 72] and

quality assurance in chemical industries [73], where it is often coupled to mass

spectrometry (GC-MS). It is essential for GC that the samples are volatile and that

the compounds to be analysed do not degrade with high temperature. This is an

issue with some synthetic cannabinoids such as XLR-11 which contain a

cyclopropyl ring as the ring opens to form a propenyl rearrangement product of the

original compound when heat is applied [74].

Capillary electrophoresis (CE) is an analytical technique utilised for

applications such as pharmaceuticals [57, 75] and forensics [76] where charged

analytes are separated based on their differing electrophoretic mobility’s. Hindson

et al. has previously used this technique for the determination of opiate alkaloids in

14

process liquors [77]. However a limitation of this technique is the lack of sensitivity

or stability often required for forensic applications and as such it is not well

established in these laboratories [76].

Chemiluminescence

Chemiluminescence is the production of light from a chemical reaction

without the generation of heat [78]. In an analytical chemistry setting the light is

commonly detected with a photodetector as over a finite range the intensity of the

light is proportional to the concentration of reactants involved [79]. It is important

to note that the amount of light produced is not necessarily about how well the

reagent reacts with an analyte but is due to the production and stability of the

excited state. This is a system that is easily coupled to HPLC [80, 81] or Flow

Injection Analysis (FIA) [82, 83] and may be used in conjunction with a UV-

absorbance detector [63, 84].

There are a range of different chemiluminescence reagents that can be

utilised for detection and they are typically prioritised based on the chemical

structure of the target analytes, as each reagent is generally known to elicit light in

the presence of particular functional groups. [79]. The most common

chemiluminescence reagents utilised at Deakin University are tris(2,2‐

bipyridine)ruthenium(III) ([Ru(bipy)3]3+), colloidal manganese(IV), acidic

potassium permanganate and luminol of which the preceding three have been

utilised in this thesis (Figure 1.7). The use of chemiluminescence is ideally suited

where selectivity is required. For example the opiates morphine and codeine are

similar in structure, however they react differently with the aforementioned

reagents. As described by Francis et al., codeine produces an intense emission with

[Ru(bipy)3]3+ where a relative intensity of 100 was obtained, but affords minimal

light with acidic potassium permanganate with a relative intensity of only 2

observed [55]. Morphine however is the opposite with the relative intensity with

acidic potassium permanganate observed at 100 as opposed to less than 0.01 with

the [Ru(bipy)3]3+ reagent [55]. This is proposed to be due to the lack of a phenol in

codeine that morphine contains, an attribute that elicits light with acidic potassium

permanganate but quenches [Ru(bipy)3]3+ [79]. However, Lenehan et al. overcame

15

this selectivity by utilising a mix of these two reagents to detect both morphine and

codeine in the same system [85].

a) b) c)

Figure 1.7: Structure of a) potassium permanganate, b) [Ru(bipy)3]3+ [55] and c) manganese dioxide (Mn(IV)).

A review on chemiluminescence by Adcock et al. highlights the sensitivity

of the technique with detection limits as low as 1 × 10-11 M reported. Table 1.1

highlights the use of the following three reagents: [Ru(bipy)3]3+, acidic potassium

permanganate and colloidal magnesium(IV) for the detection of a range of dugs.

Table 1.1: Detection limits of a range of opiate based molecules gained

with three chemiluminescence reagents.

Analyte Reagent Instrumentation LOD

Codeine Manganese(IV) FIA 1.0 × 10-8 M [86]

Thebaine Manganese(IV) FIA 5.0 × 10-9 M [86]

Norfloxacin Manganese(IV) FIA 3.0 × 10-8 M [87]

Cannabidiol Acidic potassium

Permanganate

HPLC 1.0 × 10-6 M [88]

Oripavine Acidic potassium

Permanganate

HPLC 1.0 × 10-9 M [89]

Morphine Acidic potassium

Permanganate

HPLC 7.5 × 10-10 M [89]

MDMA [Ru(bipy)3]3+ HPLC 4.8 × 10-7 M [90]

Codeine [Ru(bipy)3]3+ HPLC 5.0 × 10-10 M [63]

Thebaine [Ru(bipy)3]3+ HPLC 1.0 × 10-9 M [63]

16

[Ru(bipy)3]3+ is particularly useful for the detection of alkaloids,

biomolecules, pharmaceuticals and controlled drugs containing secondary or

tertiary amines [90-93]. The [Ru(bipy)3]2+ is first oxidised to [Ru(bipy)3]3+ and then

reduced by an analyte to produce light and [Ru(bipy)3]2+ 79 . Recently, a highly

stable form of this reagent was developed by McDermott et al. [91], where the

complex is prepared as a perchlorate salt which allows the reagent to remain in the

3+ state for a longer period of time. This reagent has been found to react strongly

with tertiary amines [78].

Acidic potassium permanganate was first reportedly used in 1917 to study

the oxidation of pyrogallol [94] and was also the first reagent used to determine

morphine in 1986 [95]. The acidic potassium permanganate also known as

manganese(VII) is proposed to form an excited state manganese(II) species via a

multi-step reduction with certain analytes, in particular those that contain phenols

[96]. This reagent has a wide range of analytical applications and has been used

extensively in pharmaceutical analysis, in particular opiates such as morphine, and

for illicit drugs, biomolecules, pollutants, antioxidants and pesticides [94, 96].

The colloidal manganese(IV) reagent was first reported in 2001 and was

found to produce light with many organic and inorganic compounds when used in

conjunction with an enhancer such as formaldehyde or ethanol [97]. This reagent

has a similar mechanism to acidic potassium permanganate where manganese(IV)

is reduced to manganese(II). This reagent exhibits a broader selectivity than that of

acidic potassium permanganate and reacts with amino acids, alkaloids, indoles,

plant-based extracts and analytes of forensic interest [86, 97-101]. This broader

selectivity is demonstrated with opiate alkaloids where it was found to react with

both morphine and codeine as opposed to just one as is noted with the previously

discussed reagents.

Electrochemiluminescence (ECL)

Electrochemiluminescence (ECL) is a technique that involves applying a

potential to an electrode in the presence of a reagent and co-reactant to induce an

electron-transfer reaction to form light emitting excited states[102]. It was first

reported for use with Ru(bipy)22+in 1972, where the sample was in acetonitrile with

the electrolyte tetrabutylammonium tetrafluoroborate [103]. It is commonly used

17

in biological and clinical testing [104], food and water assessment [105] and drug

detection [106]. Microfluidic analytical devices have also used ECL as a detection

method [104]. This technique is commonly used due to its high sensitivity and

selectivity that it affords an anlalyst [102, 104].

Mass Spectrometry (MS)

Mass spectrometry (MS) is a technique used for the identification of

compounds by determining the molecular mass and the mass of fragment ions

[107]. There are three main processes within the mass spectrometer; 1) the sample

is first ionised and then 2) passed through an electric field, ionic field, magnetic

sector or drift region where the ions are separated before 3) detection [108]. Mass

Spectrometry is a technique highly utilised in analytical chemistry for the detection

and identification of compounds of interest. It is heavily utilised in the

pharmaceutical industry for the determination of impurities and degradation

products and is often coupled with GC (gas chromatography), LC (liquid

chromatography) or CE (capillary electrophoresis) [71]. It is also used extensively

in forensics for biological samples and the identification of drugs or other

compounds.

Mass spectrometry is used in the forensic analysis of synthetic cannabinoids

and is the main identification technique utilised in conjunction with gas

chromatography [54]. This technique has been instrumental in identifying new

compounds where it is coupled to either LC or GC [109-112] and in biological

testing where it has been used to test for synthetic cannabinoids in substrates such

as blood [113-115], hair [116, 117] and urine [118, 119].

Extensive literature is available on the use of mass spectrometry for opiates

in biological samples [120-122] and in environmental applications to test their

presence in waste water [123-125]. MS has been instrumental in mapping the

opium poppy biosynthesis pathways [126] and for identifying opiate alkaloids after

isolation [127]. Stranska et al. utilised HPLC/MS/MS for the determination of

alkaloids from empty Papaver somniferum capsules [128]. This technique is most

commonly used in conjunction with liquid chromatography, where LC/MS is

commonly used on sites that extract bulk opiates as a quality control measure.

18

Nuclear Magnetic Resonance Spectroscopy (NMR)

Nuclear Magnetic Resonance (NMR) Spectroscopy is a technique often

employed for structural determination of a compound. This method has been widely

employed in both synthetic and analytical chemistry, for structural elucidation or

confirmation of compounds [108]. NMR is most commonly performed in a solution

state using a deuterated solvent, however compounds can also be analysed in solid

state [129]. Solid state NMR is an emerging technique that is often used for the

analysis of bulk samples and to investigate the surface chemistry of catalysts,

zeolites and metal organic frameworks [130-132].

Synthetic cannabinoids have evolved greatly since the discovery of the first

compounds on marketed "legal high" products. In conjunction with mass

spectrometry, these emerging compounds are commonly identified with NMR.

There are a number of papers reporting the use of 13C, 19F and 1H solution state

NMR for the structural elucidation of these new compounds [110, 133-135]. Fowler

et al. used 1H NMR and proton correlation spectroscopy to quantitatively identify

synthetic cannabinoids on herbal blends [136]. Opiate alkaloid compounds have

also previously been determined using techniques that include NMR [137, 138].

Furuya et al. utilised this technique to identify a new alkaloid present in the callous

tissue of Papaver Somniferum [127].

Data Analysis

The data obtained from LC/MS and HPLC can often be considerable and

difficult to handle when dealing with a large number of samples. Due to this, data

handling software is often employed to minimise the time it takes to interpret the

information. There are numerous open source software, such as OpenMS® and

Open chrome, available in addition to the programs supplied by the manufacturers

of the instrumentation, namely Mass hunter (Agilent), Profiling solutions

(Shimadzu) and Xcaliber (Thermo Fisher). Mathematica and other programming

software such as Matlab, allow for better interpretation of complex mass

spectrometry data and 2D- HPLC. With algorithms designed in house these

19

programs have the potential to determine important data from a large dataset and

compose heat maps for 2D separations.

Principal components analysis (PCA) is a common statistical technique to

determine differences between sample sets and identify the variables. This

technique was first reported in the literature by Hotelling in 1933 and has since

been utilised for a vast array of applications. It has previously been exploited for

analysis of pharmaceuticals in rats [139], water quality testing [140], food testing

[141, 142] and wine testing [143, 144].

Project aims

This project aims to develop improved analytical separation and detection

capacity for compounds in complex samples. The basis for this work is to

interrogate complex samples associated with forensic drug detection and

manufacture of licit drugs in order to enable these industries to best deal with

priority chemical challenges.

An efficient method for the selection of orthogonal columns will be

explored using a combination of multidimensional chromatography, mass

spectrometry and principal components analysis. The aim is to provide a foundation

for the development of a separation designed for complex opiate processing stream

samples. An investigation of all stages of the manufacturing process starting at the

final product and working back to the starting material, raw poppy straw (Papaver

somniferum) for the opiate processing streams will be aimed at mapping the

components in samples from key processing steps. The aim of this is to determine

if compounds of interest had originated from the raw product or as a consequence

of the manufacturing process. The work presented here aims to address the mapping

of the complex opiate manufacturing process so that process owners and industrial

chemists can respond more efficiently to changes identified in the process stream.

In the forensic science arena this work aims to exploit novel analytical

chemistry based on chromatography, chemiluminescence detection and solid state

nuclear magnetic resonance spectroscopy to selectively determine the presence of

synthetic cannabinoids on herbal substrates. The project aims to elucidate the

20

chemical composition of synthetic cannabinoids obtained pre-ban in Victoria

Australia in order to understand the current state of play with synthetic

cannabinoids in this region. The aim of this component is to understand the

structure function relationship between a variety of chemiluminescence reagents

and the synthetic cannabinoids in order to work towards at-scene detection

chemistry. This aim is to be achieved by comparison of the real world synthetic

cannabinoids samples and standards purchased or synthesized at Deakin

University. The fact that synthetic cannabinoids are sprayed onto a herbal substrate

enables the potential exploitation of solid state nuclear magnetic resonance

spectroscopy to probe these samples. For the first time solid state NMR will be

utilised to interrogate synthetic cannabinoids and assess it as a methodology that

may be useful for policing services.

Due to the broad nature of the two components of this thesis namely the

forensic and industrial chemistry more detailed introductory material is given at the

start of each experimental chapter. This has been performed to inform the reader of

the key areas of literature relevant to each of these bespoke areas of scientific merit.

21

CHAPTER 2:

BLIND COLUMN SELECTION PROTOCOL FOR TWO‐DIMENSIONAL HIGH PERFORMANCE LIQUID

CHROMATOGRAPHY

22

Chapter Overview

This chapter explores the use of an open source program, OpenMS,

for handling large amounts of mass spectrometry data, to aid the comparison of

chromatography columns in order to select a pair that will provide the maximum

use of 2D-HPLC separation space. A greater separation space is represented by a

large fcoverage indicating that the chromatographic system allows for better resolution

of compounds in complex samples. The separation was developed using a crude

morphine poppy extract to represent a real world complex sample matrix. Solvent

mismatch was investigated in the development of a suitable 2D-HPLC method.

Once developed, this method was applied to a range of process stream samples

(morphine, thebaine and oripavine poppy extracts) with the fcoverage for each

determined in order to fully assess the suitability of the protocol.

23

Introduction

High performance liquid chromatography (HPLC) is a technique commonly

used for separating compounds in complex substrates [145] such as biological

[146], food [147] and environmental [148] samples. The complexity of a sample

often exceeds the separation potential of traditional HPLC, thus two-dimensional

HPLC (2D-HPLC) is often performed through separating compounds with the aid

of more than one chromatography column [149]. The ability to separate complex

mixtures is greatly increased by using 2D-HPLC as the theoretical peak capacity is

the product of each individual dimension [149]. However, to achieve the best

performance, it is vital that the separation chemistries offered by the two stationary

phases selected are as different (orthogonal) as possible. Consequently, the process

of achieving this is often onerous, requiring a trial-and-error approach or

laboriously injecting a series of representative standards [66, 150].

Many disciplines have utilised 2D-HPLC including pharmaceutical [151,

152], forensic [153], environmental [150, 154] and food sciences [68]. As noted

above, the separation potential of a 2D-HPLC system is determined by the peak

capacity that is used to describe the chromatographic resolving power [155].

Resolution or resolving power of a chromatographic system is defined as the ability

for peaks to be both Gaussian and base line separated. The peak capacity is

proportional to the fractional coverage defined by Watson et al. [156] according to

Equation 1:

∑ (Equation 1)

where fcoverage is the proportional surface coverage that ranges from 0 to 1, ∑bins is

the number of bins (the segmented components of the analogue data signal) that

contain peaks, and P2 is the total number of available bins. The proportional surface

coverage is directly related to the two-dimensional peak capacity, Equation 2,

which illustrates that the full advantage of a 2D-HPLC separation will only be

realised by maximising the fcoverage term [155]:

24

,, ,

(Equation 2)

where, nc,2D is the practical 2D-HPLC peak capacity, nc,i is the peak capacity in

separation dimension i, fcoverage is the proportional amount of separation space

utilisation, and is the under-sampling correction factor, which is directly

proportional to the modulation frequency per peak from the first dimension to the

second, as described by Li and co-workers [157]. This is achieved through a given

number of bins used to divide the separation space, Σbins, that is equal to the

number of peaks; the area of all normalised bins containing peaks is then totalled

giving P2 [158]. Using this method, the fcoverage is calculated as a number between

0 and 1 with the value of 1 indicating full use of the separation space. Gilar et al.

[158] illustrated the use of separation space as shown in Figure 2.1, where it is

important to achieve a separation that is not in a diagonal line and utilises the space

available in the square otherwise low orthogonality is achieved.

25

Figure 2.1: Representation of separation space utility by Gilar et al. [158] where (A) is a non-orthogonal separation representing 0% orthogonality, (B) is a hypothetical, full use

of separation space and (C) is an ideal system representing 100% orthogonality.

26

The separation space is maximised using orthogonal columns; however,

finding two columns that fit this requirement can be a lengthy and difficult task

[66]. Previously, analysing the relative retention profile of each HPLC column

against each other was the only way to identify the best separation. Murahashi and

co-workers [150] selected columns based on differing elution orders by injecting

five standards into eight columns. This was further explored by Bassanese et al.

[66], but it was found that matching peaks by their area with HPLC proved

problematic between different columns when using the sample itself as the

optimisation matrix. Moreover, it was not practical when performing single

injections of a comprehensive series of representative standards. Other methods

include selecting column stationary phases on the knowledge that they are unlikely

to produce similar separations based on the interactions of the compounds being

analysed [68]. These techniques were found to be incredibly time consuming and

difficult to achieve when the components of the sample being analysed are

unknown [66].

For an unknown mixture, one way to match peaks from different separations

is to compare their molecular weight. Mass spectrometry is a technique used for the

identification of compounds by determining the molecular mass and is particularly

important when a greater understanding of the analytes is required [159]. Mass

spectrometry is superior to UV-absorbance detection when analysing complex

sample extractions as it provides an information rich snapshot of the elution such

as a mass associated with each compound in the mixture. This technique also allows

a fingerprint of the sample where the fragmentation pattern can discriminate

between compounds of the same mass [160] thus providing greater information on

the components in complex mixtures than UV-absorbance detection affords. Mass

spectrometry is often coupled with LC (liquid chromatography), GC (gas

chromatography) or CE (capillary electrophoresis) to enhance knowledge of the

concomitant species of a sample [161]. This is particularly important for unknown

samples where as much information as possible is essential for further analysis.

This technique has been utilised in this selectivity study to provide as much

information about the peaks as possible, which can then be used to compare the

columns via the masses of the peaks in the sample.

27

One of the greatest challenges in dealing with mass spectrometry is

handling the enormous amount of data generated. This problem is further

exaggerated by the creation of two-dimensional separations. Consequently,

OpenMS® (version 1.11.1), a program used to manage and analyse data obtained

by mass spectrometry, will be explored to interrogate the complex set of

chromatographic information. In comparison to other programs that are available

for mass spectrometry data analysis, OpenMS® offers a more flexible platform as

the user is able to combine components of the software to suit their individual

requirements [162]. Using this program, filters can be applied to the mass spectra

to extract important information, such as particular peaks of interest. There are a

large number of pipelines that can be followed to obtain the desired outcome for

the analysis, including a simple pipeline that eliminates noise from the spectra or

increasingly more complex pipelines that incorporate a number of filters and peak

picking tools. Visualisation at stages through the pipeline with the TOPPView

feature is also an important element, allowing the user to open a single spectrum or

a peak map in either two or three dimensions [163].

A major difficulty in the optimisation of 2D-HPLC operating conditions is

the selection of columns with divergent retention mechanisms [66]. The aim of this

work is to introduce a new approach to remove the guesswork in column selection;

the methods developed in this study are not limited to alkaloids but represent a

reliable column selection protocol that can be applied to any complex sample where

the constituent components are unknown. This chapter presents an approach to

optimise column selection in 2D-HPLC with a wide range of applications, and is

illustrated here by the separation of crude opiate extracts. Sixteen columns were

used to separate the components within the concentrate of (morphine) poppy straw

(CPS) using LC/MS. OpenMS® was used to evaluate data from each column and

determine which two columns together will provide the best use of separation

space. The selected columns were then used to develop an appropriate 2D-HPLC

method that was applied to additional opiate samples. This method took into

account solvent mismatch from the first to second dimension and optimised both to

allow for the requirement of a fast flow rate in the latter.

28

Experimental

Chemicals and samples

Water was obtained from a Milli-Q filtration system (Merck Millipore,

Victoria, Australia). HPLC grade acetonitrile (ACN) was purchased from Ajax

Finechem Pty. Ltd. (Taren Point, NSW, Australia). Trifluoroacetic acid was

purchased from Sigma-Aldrich and formic acid was purchased from Hopkin and

Williams (Essex, UK). Crude morphine and other opiate samples were supplied by

Sun Pharmaceutical Industries Australia Pty (Port Fairy, Victoria, Australia) that

were extracted from Papaver somniferum following proprietary protocols. All solid

samples were prepared at a concentration of 10 mg mL-1 in water prior to LC/MS

analysis and liquid samples were diluted by a factor of 10 with 10% acetic acid

(VWR International, Fontenay-sous-Bois, France).

Standards were made up into a 10 mL acetonitrile stock with 10 mg each of

thiourea, caffeine, phenol, anthracene and 10 µL of dimethyl phthalate and n-propyl

benzene (purchased from Sigma-Aldrich Pty Ltd, Castle Hill, NSW, Australia).

The standard stock was then added (50 µL) to 6 different compositions of

acetonitrile (50 µL, 200 µL, 400 µL, 500 µL and 600 µL) making up 1 mL with

water. Viscosity was adjusted to address solvent mismatch by adding an appropriate

amount of either isopropyl alcohol or ethanol (Chem-Supply Pty Ltd, Gillman SA,

Australia) according to viscosity calculations (Table 2.3).

Liquid Chromatography Mass Spectrometry (LC/MS)

Mass spectrometry analysis was performed with an Agilent Technologies

6210 LC/MSD TOF system (Agilent Technologies, Mulgrave, Victoria, Australia).

It was equipped with an Agilent technologies 1200 series solvent degasser, binary

pump and an Agilent technologies 1100 series autosampler (ALS). Positive ion

polarity was used in conjunction with nitrogen drying gas (5 L min-1) and nitrogen

nebuliser gas (30 psi). The morphine sample was injected (5 µL) into a linear

gradient over 40 minutes with an initial mobile phase of 5% ACN (95% water) that

increased to 100% ACN. All analyses were run at a flow rate of 0.5 mL min-1.

Chromatographic data were obtained and processed with Agilent ChemStation

software. The sixteen columns that comprised the stationary phase library for this

study are listed in Table 2.1.

29

Table 2.1. Details of the HPLC columns utilised.

BRAND TYPE SIZE DIMENSIONS PART NUMBER

Column 1 Agilent Poroshell 120 EC-C8 2.7 µm 4.6 ×100 mm 695975-906

Column 2 Agilent Poroshell 120 EC-CN 2.7 µm 4.6 × 100 mm 695975-905

Column 3 Phenomenex Kinetex PFP 100 Å 2.6 µm 4.6 × 100 mm 00D-4477-EO

Column 4 Phenomenex Kinetex Phenyl-Hexyl 100 Å 2.6 µm 4.6 × 100 mm 00D-4495-EO

Column 5 Phenomenex Synergi Fusion - RP 80 Å 4 µm 4.6 × 150 mm 00F-4424-EO

Column 6 Phenomenex Synergi Hydro - RP 80 Å 4 µm 4.6 × 150 mm 00F-4375-EO

Column 7 Phenomenex Luna C5 100 Å 5 µm 4.6 × 50 mm 00B-4043-EO

Column 8 Phenomenex Luna HILIC 200 Å 5 µm 4.6 × 150 mm 00F-4450-EO

Column 9 Cosmosil Cosmosil πNAP 2.5 µm 4.6 × 100 mm 08084-51

Column 10 Cosmosil Cosmosil 5PBB-R 5 µm 4.6 × 150 mm 05697-21

Column 11 Cosmosil Cosmosil 5NPE 5 µm 4.6 × 150 mm 37904-01

Column 12 Agilent Poroshell 120 EC-C18 2.7 µm 4.6 × 100 mm 695975-902

Column 13 Agilent Poroshell HPH-C18 2.7 µm 4.6 × 100 mm 695975-702

Column 14 Agilent Poroshell 120,SB-C18 2.7 µm 4.6 × 100 mm 685975-902

Column 15 Agilent Poroshell 120 Bonus RP 2.7 µm 4.6 × 100 mm 695968-901

Column 16 Agilent Pursuit XRs 3 Diphenyl 3 µm 4.6 × 100 mm A6021100X046

30

Two-Dimensional High Performance Liquid Chromatography

An Agilent Technologies (Mulgrave, Vic., Australia) 1290 2D

chromatograph was used for all separations, including: an auto sampler; solvent

degasser; capillary pump; binary pump; quaternary pump; column thermostat; 2

position 8 port switching valve; and two diode array detectors (one in each

dimension, both of which monitored absorbance at 210 nm and 254 nm). The 2D-

HPLC separations were performed using 100 µL injections and a controlled

temperature of 25°C. All mobile phases contained 0.1 % formic acid. The first

dimension was completed at a flow rate of 0.1 mL min-1 with an initial mobile phase

of 15% ACN (85% water) that increased to 40% ACN over 40 minutes, increasing

to 60% ACN over a further 20 minutes and was then isocratic (60% ACN) for a

further 30 minutes. A counter gradient was used in the first dimension, after the

column, with an initial mobile phase of 9% ACN (91% water) to 4% ACN over the

first 40 minutes, decreasing to 0% ACN over the next 20 minutes and was then

isocratic at 100% water for a further 30 minutes at a flow rate of 0.5 mL min-1. The

second dimension was completed at 2 mL min-1 with an initial mobile phase of 10%

ACN (90% water) that increased to 15% ACN over 2 minutes, increasing to 60%

ACN over a further minute. Thus, the comprehensive two-dimensional separation

had an overall completion time of 90 minutes. Two-dimensional HPLC was

completed on-line with a modulation time of 4 minutes, whereby a fraction volume

of 75 µL was transferred to the second dimension via a sample loop (500 µL) and

switching valve that was controlled with the HPLC control software. It should be

noted that ACN was used throughout as due to solvent mismatch it is of significance

to two dimensional chromatography

Data analysis

The chromatographic data obtained from the LC/MS was analysed with

OpenMS® 1.11.1 [162, 163]. The OpenMS® pipeline included a file converter, file

filter, resampler, noise filter (Savitzky Golay), peak picker wavelet, feature finder

(metabo) (metabolite features), collect, feature linker (unlabelled QT) and a text

exporter. Output files were present at the feature finder, feature linker and the text

exporter. All parameters within these tools were default except in the file filter

where m/z values were specified to be within the range of 50-700 and intensity

31

values were set to a range 1,000-30,000,000 counts. The output files from

OpenMS® consisted of a list of retention times for features that were common

amongst the investigated HPLC columns. This data was processed with Wolfram

Mathematica 10 (distributed by SHI UK, Milton Keynes, England) to match these

retention times and predict the relative retention behaviour using algorithms written

in-house to inform the analyst of possible orthogonal columns. Two-dimensional

contour plots were prepared with Wolfram Mathematica 10.

32

Results and discussion

Identifying the columns that provide the greatest separation space utilisation

for 2D-HPLC can be a labour intensive process, particularly when comparing

column combinations individually [150]. Previous researchers have attempted to

shorten the process of identifying orthogonal columns however issues such as

differing peak elution times on the variety of columns tested and unknown sample

matrices [66] has prompted the use of LC/MS to manage the peaks using molecular

masses which overcomes these issues and allows for an accurate comparison of the

columns. When analysing a complex sample matrix, where the components are

completely unknown, using standards to aid in column selection is not possible [66]

as they may not accurately represent the unknown compounds from a structural

perspective.

OpenMS® was used here to identify the columns that achieved a high

degree of fcoverage in a 2D-HPLC separation of an opium poppy extract. Although

this was not its intended purpose, it allowed a large variety of columns (256

combinations) to be compared in a significantly reduced time relative to the

traditional trial-and-error process. Additional benefits were obtained with a

reduction in method development costs, volume of solvent needed and amount of

sample required. Method development expenditure is a significant factor in

industry, where a more efficient outcome reduces the time and wage of employees

which is a major cost to the company.

OpenMS® was able to help process the data from each column based on

the masses obtained from the actual samples. The information pipeline that was

used by OpenMS® in this work is illustrated in Figure 2.2, which outlines the data

processing pipeline utilised to identify the most orthogonal columns.

33

Figure: 2.2: The final OpenMS® pipeline applied to map peaks of common mass between HPLC columns.

The initial part of the pipeline (highlighted in red) was used to analyse

samples for each of the different columns individually; this included a file

converter, resampler, noise filter, peak picker, and feature finder. In order for

OpenMS® to be able to read the mass spectrometry data files, they were converted

from the file format .mzdata to .mzml. The customised pipeline then employed a

file filter where parameters such as intensity, retention times and m/z values were

optimised to focus on the areas of interest [162]. These parameters were optimised

through the TOPPView feature of OpenMS® as shown in Figure 2.3. For this

pipeline, a set of rules were developed to selectively extract the mass values (50 to

700) from each of the input files. The LC/MS map was then transformed into a re-

sampled map before a noise filter was employed. The noise filter ‘Savitzky Golay’

was utilised for this data to smooth the baseline and eliminate peaks caused by noise

[163]. The peaks in each of the samples were then acquired from the mass spectra

using the ‘peak picker wavelet’, and all features found using the ‘feature finder

metabo’ [163]. In the second stage of the pipeline (shown in green) each of the

individual files were collected together. The ‘feature linker unlabelled QT’ then

grouped the corresponding features from multiple maps to produce an output file

[163]. From this pipeline, results were then statistically processed using

Mathematica and algorithms designed in house.

Input files File converter File Filter Resampler Noise Filter SGolay

Peak Picker Wavelet

Feature Finder Metabo

Feature Linker

Unlabelled QT

CollectText Exporter

Output Files Output Files Output Files

34

Figure 2.3: Thebaine sample a) before optimised parameters, b) optimised parameters (m/z values: 50-700 and intensity 1,000 – 1,000,000)

35

Table 2.2 lists the expected fcoverage for each column combination after using

OpenMS® to extract the retention times of features from LC/MS chromatographic

data and Mathematica to predict the fractional surface coverage when combined in

a hypothetical 2D-HPLC analysis. From the sixteen columns tested with CPS

morphine, the two columns that displayed the greatest predicted fcoverage were

identified as the Agilent Pursuit XRs 3 Diphenyl and the Agilent Poroshell 120 EC-

C18, with an expected fcoverage of 0.48 (Figure 2.4A). The retention mechanisms of

these stationary phases are different and should therefore operate on distinct aspects

of the analytes’ chemical structure. For example, diphenyl columns comprise of

two phenyl groups that are connected via a silica group and are suited to separate

aromatic compounds due to their strong dipole-dipole hydrogen bonding and π-π

interactions [164]. Surfaces that utilise these π-π interactions have been previously

used for the separation of opiate compounds and natural product extractions at large

[165]. To contrast the diphenyl separation mechanism, the C18 column separates

on the basis of analyte size and hydrophobicity via a chain of 18 carbons [166]. The

C18 surface is the workhorse of modern chromatography and has myriad

applications leading to its almost universal use in modern HPLC separations [167].

36

Table 2.2. Predicted fcoverage of each column combination. Refer to Table 2.1 for column identification.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1 0.44 0.12 0.32 0.24 0.16 0.16 0.20 0.20 0.28 0.24 0.12 0.20 016 0.32 0.24

2 0.44 0.16 0.20 0.24 0.20 0.28 0.28 0.16 0.32 0.24 0.28 0.16 0.24 0.40 0.24

3 0.12 0.16 0.12 0.20 0.20 0.40 0.36 0.16 0.36 0.40 0.32 0.16 0.36 0.16 0.20

4 0.32 0.20 0.12 0.24 0.12 0.16 0.20 0.16 0.20 0.36 0.12 0.36 0.16 0.40 0.16

5 0.24 0.24 0.20 0.24 0.16 0.16 0.24 0.28 0.12 0.16 0.20 0.24 0.20 0.36 0.24

6 0.16 0.20 0.20 0.12 0.16 0.12 0.24 0.20 0.20 0.40 0.16 0.20 0.12 0.24 0.20

7 0.16 0.28 0.40 0.16 0.16 0.12 0.16 0.20 0.28 0.20 0.16 0.16 0.16 0.16 0.24

8 0.20 0.28 0.36 0.20 0.24 0.24 0.16 0.24 0.16 0.32 0.28 0.20 0.24 0.36 0.24

9 0.20 0.16 0.16 0.16 0.28 0.20 0.20 0.24 0.40 0.28 0.16 0.16 0.20 0.20 0.28

10 0.28 0.32 0.36 0.20 0.12 0.20 0.28 0.16 0.40 0.36 0.40 0.24 0.24 0.28 0.32

11 0.24 0.24 0.40 0.36 0.16 0.40 0.20 0.32 0.28 0.36 0.20 0.24 0.24 0.28 0.32

12 0.12 0.28 0.32 0.12 0.20 0.16 0.16 0.28 0.16 0.40 0.20 0.20 0.16 0.16 0.48

13 0.20 0.16 0.16 0.36 0.24 0.20 0.16 0.20 0.16 0.24 0.24 0.20 0.20 0.36 0.20

14 0.16 0.24 0.36 0.16 0.20 0.12 0.16 0.24 0.20 0.24 0.24 0.16 0.20 0.16 0.24

15 0.32 0.40 0.16 0.40 0.36 0.24 0.16 0.36 0.20 0.28 0.28 0.16 0.36 0.16 0.20

16 0.24 0.24 0.20 0.16 0.24 0.20 0.24 0.24 0.28 0.32 0.32 0.48 0.20 0.24 0.20

37

The two plots in Figure 2.4 demonstrate the bins used for the predicted

separation. A value of orthogonality can tell how much of the fcoverage the separation

is utilising, however it is not indicative of where the space is being used. If a

predicted plot is too correlated (diagonal trend) it indicates that the two stationary

phases are operating in a very similar manner which in turn leads to an inefficient

separation of compounds in both dimensions [158]. This could lead to a separation

only in the first dimension and thus no benefit is achieved by using a second

column. This is illustrated in Figure 2.4A and B, with the latter demonstrating a

correlated separation where the bins are all located along the diagonal and uses

minimal bins with an fcoverage of only 0.20. Figure 2.4A however has a much larger

fcoverage of 0.48 and is less correlated, thus should have a better stationary phase

separation. The difference here shows the significance behind this method as it

allows an elimination of two columns that are not orthogonal.

38

(A)

(B)

Figure 2.4. Scaled retention times of peaks predicted by OpenMS® and the associated occupied bins that have been filled in the fcoverage calculation for (A) Agilent Pursuit XRs 3

Diphenyl and Agilent Poroshell EC-C18 and (B) Cosmosil πNAP and Phenomenex Synergi Fusion.

C18 (normalised retention time)

πNAP (normalised retention time)

Syn

ergi

Fus

ion

(n

orm

alis

ed r

eten

tion

tim

e)

Dip

heny

l (n

orm

alis

ed r

eten

tion

tim

e)

39

Comprehensive online 2D-HPLC requires that the second dimension be

significantly shorter than the first, requiring optimisation of the columns selected

to obtain the fastest possible separation [149]. Due to the requirement of a quick

second dimension, the methods that were utilised to produce the LC/MS maps were

unsuitable to use for the 2D-HPLC analysis. Another consideration is solvent

mismatch from the first to the second dimension which can potentially cause

viscous fingering and a solvent front that is visible on chromatograms as band

broadening, where the peaks are shorter and wider [168] and can increase retention

times [169]. This is illustrated in Figure 2.5 where a chromatogram produced by

Mayfield et al. [170] highlights the effects of viscous fingering.

Figure 2.5: Representative chromatogram by Mayfield et al. [170] highlighting the effects of viscous fingering on a HPLC separation.

A counter gradient is another method that can address differing

compositions thus eliminating the solvent front formed by mismatched starting

compositions and as such it was investigated here during method development. A

counter gradient utilises a third pump and enters the system after the first dimension

column but before transferring to the second dimension ensuring that the fraction

matches the starting mobile phase composition in the second dimension. This

40

method developed by Stevenson et al. [171] was shown to significantly improve

the peak shape of a separation and as such was trialled here in both online and

offline with a determination that a counter gradient significantly improves the

separation by eliminating the large solvent front, which can be an issue in particular

when viscous fingering is formed [172].

Figure 2.6: Illustration of viscous fingering as determined by Catchepoole et al. in the paper “Visualising the onset of viscous fingering in chromatography columns” [168].

Viscous fingering is a mismatch in solvent viscosity that results in fingers

of the heavy solvent leaching into the less viscous solvent instead of having a linear

approach. This is illustrated in Figure 2.6, where Catchpoole et al. added dye to the

more viscous solution and injected into a clear column [168]. In an attempt to solve

this problem in an alternate way to a counter gradient, isopropanol was added to a

sample of six standards in a series of differing injection compositions. This was

based on the paper by Stevenson et al. [171] where this technique was used to

predict a counter gradient. Both ethanol and isopropanol were tested to match the

viscosity of the solvent being injected into the second dimension. These two

solvents were added in the amount calculated using kinetic viscosity (Table 2.3),

where kinetic viscosity is:

41

Equation 3

The kinetic viscosity of acetonitrile was calculated to be 0.439 mm2 s-1,

water as 0.8926 mm2 s-1, isopropanol as 2.494 mm2 s-1 and ethanol as 1.232 mm2 s-

1. These values were used to calculate the required amount of isopropanol or ethanol

to match the viscosity of the starting composition.

Table 2.3. The viscosity calculations for isopropanol and ethanol.

Acetonitrile

(%v/v)

Water (%v/v) Isopropanol

(mL)

Ethanol (mL)

0.05 0.95 0 0

0.20 0.80 0.1275 0.3423

0.30 0.70 0.2125 0.5704

0.40 0.60 0.2975 0.7986

0.50 0.50 0.3825 1.0268

0.60 0.40 0.4675 1.2549

A starting percentage of 5% acetonitrile was utilised as it is a very common

initial mobile phase composition and the values were then increased at regular

intervals until 60% acetonitrile. It was observed that the peak shape of the standards

was not significantly improved by the addition of a viscosity altering additive as

shown in Figure 2.7.

42

Figure 2.7: Separation of standards, a) sample just contains standards, b) standards containing isopropyl alcohol and c) standards containing ethanol at 0.4 % Acetonitrile,

0.6 % water and 0.2975 mL-1 isopropanol or 0.7986 mL-1 ethanol

This led to the use of a counter gradient instead as this eliminated the solvent

front issue from the first to second dimension. The counter gradient was calculated

based on the Stevenson et al. [171] paper and the need for this is illustrated in Figure

2.8B and 2.9B where a solvent front is visible in the 2D-HPLC bins and contour

plots.

To reduce back-pressure concerns when operating at an elevated flowrate

in the second dimension, the 100 mm Poroshell C18 column tested with mass

spectrometry was substituted with a 50 mm Poroshell C18. With this column

length, all components of the sample were eluted within 3 minutes. Despite the

adjustments to the second dimension, the fcoverage achieved when these two columns

were utilised was 0.43 (Figure 2.8A and 2.9A), which is only slightly less than

predicted (0.48).

0

100

200

300

400

500

600

700

800

900

1000

0 2 4 6 8 10

Absorbance (mAu)

Time (mins)

No additive

Isopropanoladditive

Ethanol additive

43

(A)

(B)

Figure 2.8. Scaled retention times of peaks identified from peak picking algorithm and occupied bins of the 2D-HPLC separation of opiate alkaloids for (A) Agilent Pursuit XRs

3 Diphenyl and Agilent Poroshell EC-C18 and (B) Cosmosil πNAP and Phenomenex Synergi Fusion.

Diphenyl(Normalised retention time)

πNAP (normalised retention time)

44

A column combination with a low fcoverage was also investigated to validate

LC/MS mapping across columns and determine the reliability of this method. The

Cosmosil πNAP and Phenomenex Synergi Fusion-RP columns were expected to

operate via similar retention mechanisms, with the πNAP having preferential

retention to aromatic compounds through π-π interactions and the Fusion-RP being

a polar embedded C18; however, after mapping the masses between these columns,

a fcoverage of 0.20 was expected (Figure 2.8B). After performing the 2D-HPLC

analysis of the opiate extraction with the πNAP column in the first dimension, the

analysis was determined to have an fcoverage of 0.34 (Figure 2.9B). The 2D-HPLC

plots for the combination of diphenyl × EC-C18 and πNAP × Synergi Fusion are

represented in Figure 2.9A and 2.9B, respectively. In the OpenMS® prediction the

components were correlated, which was also observed in the real sample; however,

the actual analysis did give a significantly higher fcoverage than predicted. This was

due to the solvent mismatch from the first dimension to the second dimension

artificially decreasing the retention time of some of the peaks [171].

45

(A)

(B)

Figure 2.9. (A) 2D-HPLC plot using an Agilent pursuit XRs 3 Diphenyl in the first dimension and a Phenomenex Poroshell 120 EC-C18 in the second dimension. (B)

Illustration of a 2D-HPLC separation with a small fcoverage: the first dimension (πNAP) was completed at a flow rate of 0.05 mL min-1 with an initial mobile phase of 15%

aqueous ACN that increased to 55% aqueous ACN over 200 minutes and was isocratic (55% aqueous ACN) for a further 10 minutes. The second dimension (Synergi fusion) was completed at 2 mL min-1 with an initial mobile phase of 20% aqueous ACN that increased

to 60% aqueous ACN over 3 minutes. Two-dimensional HPLC was completed on-line with a modulation time of 5 minutes and completion time of 210 minutes.

46

The separations were optimised with a sample from the later stages (data

above) of the extraction process in order to use a less complex opiate alkaloid

sample, so as to not overload the computing processors (4 Gig 8 core which is

typical of a standard computer as of 2017). As the previous approach showed

successful multidimensional separation of the opiates, the universal capabilities of

the methodology were tested with other more complex opiate samples (data below)

from the Sun Pharmaceutical extraction process. The opiate processing stream and

impurity standards consisted of 68 samples that have all been analysed with this

2D-HPLC method. The following two-dimensional plots (Figure 2.10 and 2.11)

confirm that the protocol developed can be applied to a variety of opiate samples

and afford a separation of the components in a complex sample. An identical

method was used when scaling up the complexity of the separation and the degree

of space utilisation was maintained.

47

a)

b)

Figure 2.10: fcoverage plots for the samples (a) G1527-030G (0.47) and (b) G1527-027C (0.47).

0.0 0.2 0.4 0.6 0.8 1.00.0

0.2

0.4

0.6

0.8

1.0

1st Dimension

2nd

Dim

ensi

on

0.0 0.2 0.4 0.6 0.8 1.00.0

0.2

0.4

0.6

0.8

1.0

1st Dimension

2nd

Dim

ensi

on

48

a)

b)

Figure 2.11. Two example 2D-HPLC chromatograms (G1527-030O and G1527-027C) representing complex samples from an opiate alkaloid extraction process line. (Please

note Purple represents minima in this example set at zero and Red maxima)

49

Table 2.4: fcoverage calculated from each stage of the processing stream

Sample Average fcoverage Sample Average fcoverage

G1527-027A 0.39 G1527-028A 0.38

G1527-027B 0.42 G1527-028B 0.33

G1527-027C 0.45 G1527-028C 0.32

G1527-027D 0.39 G1527-028D 0.34

G1527-027E 0.37 G1527-028E 0.34

G1527-027F 0.36 G1527-028F 0.40

G1527-027G 0.37 G1527-028G 0.38

G1527-027H 0.36 G1527-028H 0.44

G1527-027I 0.37 G1527-028I 0.40

G1527-027J 0.43 G1527-028J 0.38

G1527-027K 0.37 G1527-028K 0.41

G1527-027L 0.35 G1527-028L 0.43

G1527-027M 0.36 G1527-030A 0.37

G1527-027N 0.40 G1527-030B 0.40

G1527-027O 0.37 G1527-030C 0.38

G1527-027P 0.39 G1527-030D 0.39

G1527-027Q 0.39 G1527-030E 0.36

G1527-027R 0.37 G1527-030F 0.36

G1527-032A 0.41 G1527-030G 0.36

G1527-032B 0.29 G1527-030H 0.29

G1527-032C 0.32 G1527-030I 0.39

G1527-032D 0.24 G1527-030J 0.32

G1527-034A 0.29 G1527-030K 0.33

G1527-034B 0.31 G1527-030L 0.30

G1527-034C 0.41 G1527-030M 0.36

G1527-034D 0.36 G1527-030N 0.42

G1527-034E 0.31 G1527-030O 0.46

G1527-034F 0.41 G1558-131A 0.41

G1527-034G 0.39 G1558-131B 0.31

G1527-034H 0.34 G1558-131C 0.40

G1527-034I 0.32 G1558-131D 0.41

G1527-034J 0.32 G1558-131E 0.37

G1527-034K 0.39 G1558-131F 0.37

G1527-034L 0.39 G1527-034M 0.36

50

Some of these samples contain a large number of compounds whilst others

are not as complex which is noticeable from the fcoverage’s determined from each

stage of the processing stream as shown in Table 2.4. Using previously described

peak picking algorithms and the bins method, the measured average fcoverage of the

complicated extracts vary between 0.24 and 0.46 depending on the complexity of

the sample. The lowest average fcoverage achieved was 0.24, however this sample is

an impurity standard papaveraldine, where there are only a few other components

present beside the major peak. The largest average fcoverage achieved was 0.46 for

the sample G1527-030O which consists of the mother liquor after the target

compound oripavine has been precipitated out of the processing stream. The

complex sample G1527-027C is from the start of the morphine processing stream

and had the next greatest average fcoverage at 0.45. The fcoverage plots for G157-030O

and G1527-027C are illustrated in Figure 2.10a and 2.10b respectively, where it is

evident that a large amount of the available separation space is utilised and the

peaks are not focused along the diagonal line. The 2D-HPLC plots for these two

samples are also shown in Figure 2.11 where the complexity of these samples can

be observed. These plots highlight the need for a two-dimensional separation as all

the compounds would not be resolved using a single column and the methodology

developed here delivers the capacity to resolve the complex samples.

51

Conclusions

Due to developing technology, the vast quantity of available HPLC columns

has given considerable choice when selecting columns for HPLC analysis [66].

When selecting columns for LC×LC, it is essential that their retention mechanisms

are as different as possible. However, current methods to identify columns that fit

this criteria are time consuming and not ideal for blind samples. Utilisation of

LC/MS to compare columns has eliminated this problem as the unknown

compounds are compared via their molecular mass. This allows columns to be

blindly selected for unknown sample components without comparing each column

pair individually. Even when the sample is known, injecting multiple standards for

each column is still incredibly time consuming, whereas with the above method all

standards can be combined, requiring each column to only be analysed once. This

method not only allows for more columns to be compared in a shorter amount of

time, but also spares the amount of sample and mobile phase required and the man

power and wages for the time spent on the selection thus reducing the cost

excluding of course the purchase price of a mass spectrometer.

The use of the program OpenMS® enables successful selection of

orthogonal columns for LC×LC. It was determined here that a first dimension

diphenyl stationary phase and a C18 column in the second dimension was optimal

for the separation of complex opiate extraction samples. An fcoverage of 0.48 was

predicted by OpenMS® and algorithms designed in-house, with an actual

separation fcoverage of 0.43 obtained after solvent mismatch was addressed with a

counter gradient in the second dimension. The developed method was then applied

to 68 opiate processing samples where average fcoverage values ranged between 0.24

and 0.46, depending on complexity of the sample. For experiments requiring repeat

analysis, the methods outlined here could allow for a comprehensive study to find

the columns with the greatest orthogonality.

52

CHAPTER 3:

OPIATE PROCESS STREAM ANALYSIS

53

Chapter overview

The industrial process involved in extracting opiates from the opium poppy

is quite complex and involves many steps. Of interest to Sun Pharmaceuticals is

what components are present in each stage of the process, and which areas afford

the most amount of chemical change in the manufacturing line. This chapter

focuses on the analysis of the processing stream and the development of the

platform technology to chemically map the manufacturing process. Principal

Components Analysis (PCA) is employed after 2D-HPLC analysis of 65 samples

from morphine, thebaine and oripavine streams. Mass spectrometry is then used to

identify the masses that the PCA process identifies as the largest areas of change.

The platform technology developed here enables the analysis of each stage of the

manufacturing process or changes between different batches to be identified.

54

Introduction

An important aspect of manufacturing pharmaceuticals is the elimination of

impurities from the final product. Impurities can come in a variety of forms; as a

starting material (including solvents and reagents), as a product of the

manufacturing process or as a degradation product [71]. Impurities in

pharmaceutical products are an issue that is closely monitored with strict

regulations followed, as impurities could alter the effectiveness and safety of the

product [173].

Sun Pharmaceuticals extract opiates for use in medications, making it

essential that the impurity levels are low. If impurities are high the batch fails,

costing money and time, which has a negative impact on the business. Buyers of

the opiates want to know the product they are getting is pure, leading to enquiries

about unknown trace impurities. Of great interest to the manufacturers of

pharmaceuticals is where the impurities come from and if there is a way to prevent

them from forming or a way to eliminate them early in the process to ensure they

don’t end up in the final product [174]. With improvements in the detection limits

afforded by modern analytical chemistry instruments manufacturers can now

determine impurities in ever diminishing concentrations. Due to having a better

understanding of the composition of the products manufactures need to develop

strategies to mitigate the impact of the impurities in the material produced.

The use of a combination of 2D-HPLC, PCA and LC/MS has the potential

to allow mapping of the opiate processing streams and discern what is changing at

each stage. Principal components analysis (PCA) affords a deeper understanding of

what is occurring in samples. This technique has previously been utilised in the

chemical manufacturing of wine to inform vintners of the relationship between

wine quality and the grapes physiochemical index [175]. It has also been used for

the quality control of a medication in Belgium, where counterfeit pharmaceuticals

are an issue. PCA was used to fingerprint medicine based on impurities and other

components from the manufacturing process to compare legal and counterfeit

profiles [176]. This highlights that this technique is very useful in chemical

manufacturing and quality control.

55

This is also important for different batches of poppy straw as growing

opium poppies is subject to natural growing environments and each crop is likely

to be affected in different ways. This can be caused by different geographical

locations or even different cropping techniques employed on the farm. These

factors can lead to varying levels of extracted compounds or the potential for new

or differing impurities. This was noted in a study by Hopfer et al. where wine from

different vineyards was analysed and found to differ in the elemental compositions

[177]. The winery these wines were processed in also had an effect on the elemental

composition [177]. In addition to this, new poppy hybrids are trialled and

occasionally these new breeds could have a huge impact on the amount of opiates

and or new impurities. In cases such as this, the platform technology developed

here has the potential to make the analysis of any fundamental changes easier and

allow monitoring of batches and crops before processing.

56

Experimental

Chemicals and samples

Water was obtained from a Milli-Q filtration system (Merck Millipore,

Victoria, Australia). HPLC grade acetonitrile (ACN) was purchased from Ajax

Finechem Pty. Ltd. (Taren Point, NSW, Australia) and formic acid added at 0.1%

to all mobile phases, was purchased from Hopkin & Williams Ltd, Essex, UK.

Opiate processing stream samples were supplied by Sun Pharmaceutical Industries

Australia Pty (Port Fairy, Victoria, Australia) that were extracted from Papaver

somniferum following proprietary protocols. All solid samples were prepared at a

concentration of 10 mg mL-1 in water prior to 2D-HPLC analysis and liquid

samples were diluted by a factor of 10 with 10% acetic acid (VWR International,

Fontenay-sous-Bois, France). All solid samples were prepared at a concentration

of 1 mg mL-1 in water prior to LC/MS analysis and liquid samples were diluted by

a factor of 100 with 10% aqueous acetic acid. Where these liquid processing

samples fit into the process is illustrated in Figures 3.2 and 3.3. Other samples not

in these diagrams include impurity standards and powdered samples. Because of

the sensitive proprietary nature of the work, codes were used as-is to de-identify

the samples to make publication of the data possible while not divulging proprietary

information.

Two-Dimensional High Performance Liquid Chromatography

This was performed on the same instrument with the same method

developed and outlined in Chapter 2.

Principal Components Analysis (PCA)

The 2D-HPLC separations were exported from the chromatography data

system ChemStation as .AIA files before the peak information was extracted by

first smoothing the raw data. Algorithms built in-house in the Wolfram

Mathematica environment were used for all data analysis. The first and second

derivatives of the curve were then scrutinised to find the start, end and maxima

times for resolved, overlapping and shouldering peaks. After this the peaks were

mathematically described and their area found by applying a combination of model

57

Gaussian curves to the data as illustrated in Figure 3.1 where Pap et al. [178] fitted

model Gaussian peaks to a GC separation. To ensure the same compound is not

spread over several first dimension fractions the peak profiles are compared in

adjacent chromatograms.

Figure 3.1: A representation of the fitting of model Gaussian curves to data by Pap et al. [178] for a GC separation of n-penthane and n-hexane.

An algorithm designed in-house [65] using Wolfram Mathematica 10

(distributed by SHI UK, Milton Keynes, England), processed all chromatograms in

batch using the below peak picking parameters:

Least squares tuning parameter (λ): 1 × 106

Initial standard deviation for Gaussian fitting: 0.03 (min)

Savitsky-Golay window size: 3; polynomial order: 2

Minimum peak response threshold: 0.2

First derivative minimum response threshold: 5 × 10-4

Second derivative minimum response threshold: 1 × 10-5

90% overlap requirement to group peaks in adjacent

chromatograms.

58

Data pre-treatment was then performed in Mathematica 10 on the extracted

2D-HPLC fingerprints with an in-house developed algorithm. In short this included

peaks being reported as absolute multidimensional elution times after their

retention times were analysed before peak retention times were binned at 6 second

increments and the peak areas summed to define variables. Normalisation was

conducted on these variables for chosen pairs of sample sets and the mean

subtracted before a batch wise calculation of the first 2 principal components of the

sample set pairs. Reconstruction of the multidimensional retention times for the

loadings led to the development of the scores and loadings plots from principal

components analysis using a covariance matrix. This two dimensional data was

transposed to the bubble plot formation and all loadings were represented by the

colours which can be used to either identify the positive and negative loadings or

relate to the sample of interest. Throughout the text below this is described in each

of the figure legends respectively

Liquid Chromatography/ Mass Spectrometry (LC/MS)

A Shimadzu LCMS-8040 Triple Quadrupole Liquid Chromatograph Mass

Spectrometer (LC-MS/MS) was used for the analysis. The data in chapter 3 and

chapter 4 were generated in MS/MS all other data was generated in single

dimension MS. Positive ion polarity was used in conjunction with nitrogen drying

gas (15 L min-1) and nitrogen nebuliser gas (3 L min-1). All processing stream

samples (Chapter 3) were analysed with an injection volume of 1 µL and a flow

rate of 0.5 mL min-1 for both methods. Method A (scaled to match first dimension

LC×LC) was completed with an Agilent pursuit XRs 3 diphenyl column at an initial

mobile phase of 15% ACN (85% water) that increased to 40% ACN over 8 minutes,

increasing to 60% ACN over a further 4 minutes and was then isocratic (60% ACN)

for a further 6 minutes. Method B (scaled to match second dimension LC×LC) was

completed with a Phenomenex Poroshell C18 column at an initial mobile phase of

10% ACN (90% water) that increased to 15% ACN over 8 minutes, increasing to

60% ACN over a further 4 minutes and was then returned back to 10% ACN for a

further 4 minutes. Chromatographic data were obtained and processed with

59

Shimadzu profiling solutions software with total ion count thresholds set to either

1,000,000, 500,000 or 1,000 counts.

60


Figure 3.2 is an illustration highlighting where in the system the 18 samples

obtained from Sun Pharmaceutical Industries morphine processing stream

originated. Samples included here comprise of those from the main opiate line, the

final product and the solvent waste streams. Likewise, Figure 3.3 illustrates the

oripavine and thebaine manufacturing streams, where selected waste stream

samples and the main processing line for both opiates have been provided for

analysis.

61

Figure 3.2: Diagram of the samples from the morphine processing stream.

G1527-027A

G1527-027D

G1527-027F

G1527-027E

G1527-027B

G1527-027H Fresh Solvent

G1527-027K

G1527-027JG1527-027I

G1527-027M

G1527-027L

G1527-027R

G1527-027Q

G1527-027P

G1527-027N

G1527-027O

G1527-027G CPS

Crystallise & precipitate

First aqueous (Aq.) extract

Aq. - tank 2 pH adjustment

Aq. – Post Filtration pH adjustment

Aq. Added complexing agent with impurity removal

Aq. extract tank 1

PSE (partway through extraction)

Aq. input

Aq. waste

Solvent waste

Fresh aq.

Rich solvent

Rich aq. composite

Rich aq. (CE) fresh

Solvent

AEX aq.

62

Figure 3.3: Diagram of the samples from the thebaine and oripavine processing stream.

G1527-028A

G1527-028L

G1527-028E

G1527-028B

G1527-028D

G1527-030H

G1527-030F

G1527-028HG1527-030A

G1527-028F

G1527-028C

G1527-030D

G1527-028G

G1527-028J

G1527-030B

G1527-030I

G1527-030G

G1527-030C

G1527-030L

G1527-030E

G1527-030J

G1527-030M

Oripavine

Precipitate

G1527-030K

G1527-030O

Process continues –No samples

First Aq./ water soluble solvent extract

Heated & water soluble solvent removed

Aq. Filtered (Drum)

Rich solvent

Steam stripped of solventSpent Aq. freshFresh solvent

Rich solvent

Bulk solvent (rich) Thebaine, No oripavine

Rich solvent Caustic

Once extracted solvent (PST)

Rich aq. Oripavine (CEO)

Waste aq.

Waste aq.

Twice extracted solvent (PST 2)

Waste solvent

Acid

Acid wash

Acid

Caustic

Caustic

Acid wash

Aq. Rich extract 2

Aq. Rich extract 1

63

Principal Components Analysis (PCA)

Raw chromatographic data requires peak alignment before chemometric

analysis can be performed due to minor variations in retention times. To align the

2D-HPLC peaks from this data correlated optimised warping [179] was trialled,

however it proved to be inefficient on a standard computer where it was unable to

align 6 chromatograms with approximately 216,000 data points. In order to achieve

a more rapid data analysis that is able to be performed by a standard computer, the

data set was reduced by extracting the multidimensional retention times [65] and

cumulative peak areas [180] to create characteristic sample fingerprints.

Keeping the restrictions used to determine peak parameters constant, batch

PCA was performed on comparisons of 89 sample pairs, each with triplicate 2D-

HPLC data. In this chapter two samples (G1527-028H and G1527-030O) have been

selected as representative of all sample sets to show the developmental process of

the techniques employed. A more in depth analysis of other samples of interest is

included in the case studies in Chapter 4 where the significance of the chemistry at

each stage of the processing stream is discussed. This chapter deals with the

development of the analytical protocol. The samples selected are of interest due to

their location in the oripavine processing stream, where G1527-028H comes

directly before the target compound is removed and G1527-030O is the spent

stream after the removal of oripavine.

In Figure 3.4, the 2D-HPLC data (sample G1527-030O) has been illustrated

in two distinctive ways to highlight the value of the information that can be gained

from these plots. The most common (Figure 3.4a) way of illustrating the data is via

a contour plot, where the purple represents the background, the red signifies a large

intensity peak and the white dots indicate the peak maxima. A disadvantage of this

plotting technique is the ability to appropriately visualise the third dimension to the

data, the peak intensity. This occurs because compounds that are less-absorbing

may be lost when scaling the data or large peaks lose their detail when trying to

visualise these smaller peaks. An alternative way of visualising this data is a bubble

plot developed specifically for this project where a characteristic fingerprint was

produced by picking the key chromatographic features. The bubble plot allows the

peak height to be visualised by the diameter of the circle, where the centre is the

64

peak maxima at the corresponding first and second dimension retention times (x-

and y-axis, respectively). The diameter of the bubble in Figure 3.4b is proportional

to the relative peak area and they have been normalised between 0 and 1. It is also

important that the size of the bubble is only indicative of the height of the peak and

does not convey the resolution of the separation based on overlapping bubbles.

65

A)

B)

Figure 3.4: Illustrative 2D-HPLC chromatograms of sample G1527-030O (replicate 1). A) Contour plot with peak intensities from 1 to 75 amu is represented and is scaled from purple (low) to red (high), B) peak maxima are located at the centre of the bubbles and

the peak intensities are represented by the bubble diameter, which have been normalised.

66

The sample or variable matrix is required to be rectangular for the principal

components calculation. This required the use of the binning technique, which is

where the separation space is broken up into square ‘bins’ [155]. The information

can then be characterised by the amount of bins the data points occupy. This was

illustrated in Chapter 2, where for example Figure 2.10 shows grey squares that

represent the bins occupied by data points and the white area are those that are not

occupied. Using this method, the retention data was divided evenly in 0.1 minute

intervals for the 90 minute run time to overcome the differing number of peaks

detected in each chromatogram. By creating the rectangular sparse matrix, the

analytes with slightly differing retention times were aligned thus eliminating the

need to align the peak retention data prior to chemometric analysis. Prior to binning

the data, it was normalised and the mean from each bin’s cumulative peak area was

subtracted. Principal components analysis was completed with randomised

triplicate 2D-HPLC analysis of each sample in batch using a covariance matrix and

scores plots of the first two principal components were created along with their

corresponding loadings plots.

Figure 3.5 displays the PC1 × PC2 scores plots for the samples G1527-028H

and G1527-030O, where 67.4% of the sample variance is accounted for in PC1 and

89.4% by the 2 components. A clear distinction can be noted between the two

samples with the G1527-028H replicates grouped in the first quadrant of PC1 and

PC2 replicates of G1527-030O in the second quadrant. One replicate of G1527-

030O is not grouped tightly with the other two, however this can be explained by

this sample being of a solvent base, where some solvent remained in the sample on

the first replicate thus affecting this result. Despite this, it can be observed that there

was a large distinction present in PC1 between these two samples with a large

majority of variance covered by the first two principle components.

67

Figure 3.5: PC1 × PC2 scores plot of G1527-028H and G1527-030O samples.

Indications of what variables are responsible for the bulk variance between

samples were observed by the loadings of the principal components. For this

analysis the loadings of the first two principal components were converted into a

multi-dimensional representation of the data, which is illustrated by the bubble

plots in Figure 3.6. The centre of the bubble corresponds to the coordinates of the

multidimensional retention time while the absolute normalised loading on each

principal components equals the radius of the bubble. A larger impact on the change

that peak had on the overall discrimination between groups on the scores plot is

signified with a larger diameter bubble. To simplify, the more significant the

change in peak area, the larger the corresponding bubble). The colour of the bubbles

is also significant as they indicate which sample is responsible for an increasing or

decreasing change in average cumulative area, where blue indicates a larger

average area of the peak in G1527-028H and red indicates the peak in G1527-030O

had a larger average area.

68

A)

B)

Figure 3.6: PCA loadings of A) PC1 and B) PC2 for the samples G1527-028H and G1527-030O. A blue bubble indicates a larger average area in the 028H samples, and a

red bubble indicates the 030O samples were more concentrated.

69

The comparison of these two samples indicated from the loadings of PC1

and PC2 that the most variation between samples occurs in the lower left hand

section of the multidimensional separation. The high concentration compounds

would most likely be highly polar in the aqueous streams, but as the pH is adjusted,

polar compounds may become neutral and crystallise out. As such they would

demonstrate the most significant change in relative area. In each of the loadings

plots, the bubbles were normalised separately so the PC1 and PC2 changes cannot

be directly compared, for example the largest bubble in the second principal

component does not necessarily mean that there was a greater impact than a smaller

first principal component and vice versa. The biggest change in PC2 was caused by

a peak eluting at 72 minutes in the first dimension and 3.17 minutes in the second

dimension. The data for this peak (72, 3.17 minutes) of the triplicate

chromatograms of G1527-030O are presented in Table 3.1, in addition to the

loadings of this peak when compared to G1527-028H. The loadings values of both

principal components were similar, but this smaller value was more significant

compared to the other loadings in PC2 compared to PC1. Due to this peak having

a greater influence on PC2 in the scores plot it was likely that the relatively large

value of the standard deviation (σ = 2.289) of these areas was causing most of the

spread on this axis in Figure 3.6.

Table 3.1. Replicate fingerprint data for multidimensional chromatograms

of G1527-030O.

Replicate tR,1 tR,2 Area Loading PC1

(Rep 1-3)

Loading PC2

(Rep 1-3)

1 72 3.160 7.812 -0.0841 -0.06394

2 72 3.164 2.781

3 72 3.165 3.151

To find if the molecules were being formed by the manufacturing

process or whether impurities present at the end of each stage were from the opium

poppy itself, these loading plots from the 2D-HPLC fingerprints were interrogated.

Representative chromatograms of before (G1527-028H) and after (G1527-030O)

the key compound oripavine is extracted are illustrated in the contour plots in

70

Figure 3.7, where the PC1 loadings are overlayed. It was postulated that if

differences between these samples were from the poppy biomass extraction alone,

the resultant changes in these chromatograms would be peaks that are missing from

G1527-030O that remain in G1527-028H, which would be represented by yellow

bubbles only. However, the biggest changes between these two chromatograms

were the presence of new compounds throughout the G1527-030O chromatogram

as highlighted by green bubbles in Figure 3.7. The G1527-030O sample was taken

from a reaction vessel that had been under storage for several days between cycles

and thus kept at atmospheric temperature and pressure. This would suggest that

new compounds seen in this sample are a result of degradation or oxidation

reactions taking place in the vessel and thus confirm that changes are taking place

during the manufacturing process. To date, Sun Pharmaceuticals have done very

little work in this space and the work presented in this thesis has enabled the

company to further refine its priorities for sample interrogation.

71

A)

B)

Figure 3.7: Overlay of PC2 loadings on 2D-HPLC chromatograms of A) G1527-028H and B)G1527- 030O (replicate 1). A yellow bubble indicates a larger average area in the G1527-028H samples, and a green bubble indicates the G1527-030O samples were more

concentrated.

72

An indication that this protocol is working well can be determined by the

larger yellow disk observed in Figure 3.7 from the G1527-028H sample, this

identifies the removal of oripavine before the next sample (G1527-030O) the key

step at this point of the process. This reiterates some features that were expected

however other aspects such as the formation of new compounds in the latter sample

are an important finding in the manufacturing process. This is significant for quality

control and to understand what is occurring at each step of the process. If new

compounds are formed at stages such as this and then become problem impurities

in the final product, it is possible that it can be addressed by not allowing the batch

to be exposed for as long. To enable the identification of the compounds causing

major changes in between processing steps, LC/MS can be utilised to pinpoint the

responsible masses.

Mass Spectrometry

Mass spectrometry was utilised to develop an in depth understanding of the

peaks that were responsible for the most amount of change between process

samples. It was anticipated that between G1527-028H and G1527-030O there

would be a significant amount of change to the amount of oripavine present due to

the step in between these two samples being the removal of this compound. To fully

validate this, a standard of oripavine was also analysed by 2D-HPLC and LC/MS

as shown in Figure 3.8. Due to the significantly greater sensitivity of the mass

spectrometer in comparison to the UV-vis detector, the oripavine was too

concentrated in all three samples. Thus, when the mass spectrometry data is

observed the change in oripavine was not noted as a big change as in all three

samples the detector recorded a maximum total ion count for this compound. This

however is significant as it highlights that although the bulk of oripavine is

precipitated out at this step, there is still a large portion carried over to the mother

liquors that continue on in the process.

73

Figure 3.8: 2D-HPLC contour plot of a supplied oripavine sample (G158-131A).

Whilst oripavine is expected to change, other unknown components of the

sample also differ, requiring the use of a combination of 2D-HPLC, PCA and mass

spectrometry to identify those compounds. The areas responsible for the greatest

change are presented in Table 3.2, in conjunction with the first and second

dimension retention times. These retention times can then be matched to the scaled

LC/MS data to locate what compounds correspond to these changes and the signal

intensities can then be compared to identify the mass of the responsible molecule.

74

Table 3.2. The PCA loadings (Figure 3.5A) scores for the comparison of

G1527-028H and G1527-030O.

First dimension RT Second Dimension

RT

PC1 loadings

scores

24 0.85 0.243382

24 1.05 0.196234

28 0.65 -0.126360

36 0.65 -0.147570

32 1.05 -0.458970

32 0.85 -0.492530

28 0.55 -1.000000

The positive PC1 loadings scores indicate the compound has a greater

average area in G1527-028H than in G1527-030O and the negative values indicate

a greater presence in the latter, as visualised in Figure 3.5. From these values and

coordinates the LC/MS data was consulted to identify the masses responsible for

the changes (Table 3.3).

75

Table 3.3: Masses corresponding to first dimension retention times

identified by PCA (PC1).

Ion m/z Ion RT

(mins)

Adj. RT

(mins)

G1527-028H

(Ion counts)

G1527-030O

(Ion counts)

314.19 4.80 24.02 1322110 0

617.30 4.94 24.68 4101544 3714015

298.17 5.02 25.10 20000000 20000000

299.19 5.02 25.10 20000000 20000000

314.19 5.68 28.38 0 1121306

329.20 5.71 28.57 1303501 1699879

318.15 5.72 28.59 0 1462715

328.18 5.73 28.65 7626469 8626498

370.18 5.79 28.94 2270866 2851843

330.20 5.81 29.06 20000000 20000000

344.22 6.45 32.23 7411381 14269454

328.18 6.54 32.70 17465896 20000000

329.20 6.55 32.73 3215862 5032537

344.23 6.60 33.02 20000000 20000000

345.24 6.62 33.12 20000000 20000000

312.19 7.22 36.09 18173297 20000000

316.19 7.22 36.12 0 1353942

314.23 7.23 36.13 3935946 7778106

313.21 7.23 36.17 3725287 6779475

342.21 7.41 37.06 3828494 4116965

In this chapter, the methodology has been shown to be effective at

identifying key features of the data at hand. To realise the full potential of this

methodology, application to the entire processing streams of morphine, thebaine

and oripavine has been performed in Chapter 4 to establish changes in a chemical

system that encompasses numerous steps and chemical environments.

76

Conclusions

A platform technology has been developed here for the mapping of opiate

extraction streams that will allow for a greater understanding of changes along the

process. A multidimensional separation addressed the complexity of the process

samples and achieved the necessary resolving capacity not afforded by

conventional HPLC. These two-dimensional HPLC fingerprints were used in

conjunction with PCA loading plots to highlight key differences between pairs

of sample sets taken from the Sun Pharmaceuticals morphine and thebaine

extraction process. It was found that relative changes between two sets of samples

could be identified by the analyst using bubble plots of loadings overlayed on a

contour representation of the data and relative peak area could be labelled as either

increasing or decreasing. The major changes could then be pinpointed using scaled

LC/MS, to find the mass of the peaks of interest.

77

CHAPTER 4:

APPLICATION OF PLATFORM TECHNOLOGY TO MONITOR COMPONENTS OF INTEREST

78

Chapter overview

This chapter applies the methods previously described in Chapters 2 and 3

to a series of case studies based on monitoring key compounds of interest within

the opiate processing streams. Segments of the processing stream where the key

differences between the samples were observed are investigated to highlight the

impact that each stage of the process has on the chemistry of the sample matrix.

This chapter highlights the utility of this type of methodology to solve tangible

process chemistry problems and builds capacity for the process owners to make

timely informed decisions.

79

Introduction

The opiate processing stream is complex, with many steps required in order

to extract and purify opiates on a large scale. These steps can be as simple as being

held in a holding tank or as complex as large column separations. In-process control

tests at each stage of drug extraction are important to the final quality of the

intermediate or API. As such, it is important to know what is occurring along the

process in case of an abnormality or to track trace impurities to ensure the quality

of the drugs [181]. Manufacturers of pharmaceuticals must adhere to the strict

guidelines set out in the ICH-Q3A of 2006 [182] and comply to the relevant

pharmacopeia monographs. The ICH-Q3A specifically targets the regulation of

impurities in new drug substances to ensure all known and unknown drug

impurities are reported correctly and meet safe standards [182, 183]. Currently a

threshold for impurities and degradation products is in place for active

pharmaceutical ingredients (API) with a maximum of 0.1% unknown compounds

allowed [71]. However, as sensitivity of detection methods improve it has a flow

on effect to the customer of these products as they begin to question these trace

impurities, thus applying pressure to the manufacturer to know the identity of each

of these unknown compounds.

Quality critical attributes (QCA) and quality critical parameters (QCP) are

both important complimentary concepts. The quality critical attributes are

characteristics that could potentially impact the safety and effectiveness of the

product [184] and as such quality critical parameters are put into place by

companies to ensure the attributes are met. To ensure effective quality critical

parameters are in place an understanding of what may affect the impurities is

required and identification of compounds (e.g. structure) would help with this

insight.

Of further interest is where these compounds originate and what is lost and

gained at each step of the process, to determine if impurities of interest are a product

of the manufacturing stream, introduced with raw materials or solvents, or if they

come from the key starting materials, i.e. opium poppy, itself. This also has the

potential to be complex as the variety and growing conditions of the crops have the

potential to influence the type and quantity of impurities, and as such highlights the

80

importance of having robust analytical monitoring that can detect and measure

these changes. Currently there are many molecules of interest to Sun

Pharmaceutical Industries Australia that to date have only been identified by their

molecular weight. While the list is extensive, for this thesis the molecules with a

molecular weight of 285 amu, 297 amu and 622 amu have been identified as

significant impurities in their processing streams by Sun Pharmaceuticals. It is of

great interest to know where within the process stream these impurities originate

for Sun Pharmaceuticals to develop informed and subsequently targeted

elimination or mitigation protocols.

Molecule of Interest 1 (285 amu)

The molecule with a mass of 285 amu is of great interest to Sun

Pharmaceuticals as it has been observed as an impurity for a long time in the

morphine processing stream and is present in the final product. Interestingly

morphine also has a molecular mass of 285 amu and it has been proposed that it is

likely to be an isomer of morphine. Isomers are typically difficult to separate and

this may explain why it is such a persistent impurity in the morphine line.


The molecule with a mass of 622 amu is also a key component, more

structural elucidation including NMR spectroscopy has been performed on this

molecule and the compound has been identified by Sun Pharmaceuticals as O-

Methylsomniferene (O-M-Somniferene), an asymmetric dimer of thebaine (Figure

4.1). As such this compound may possess similar chemical attributes to thebaine

causing it to be persistent within the processing stream. This compound has

historically been a problem in the thebaine processing stream and as such it is of

great interest where this compound emanates from in order for the process owners

to limit its production or improve on its removal.

81

Figure 4.1: Structure of O-M-Somniferine.


A further impurity of both the thebaine and oripavine processing stream

(due to the same starting point/product) is the molecule with a molecular mass of

297 amu, previously determined to be nor-thebaine (Figure 4.2). Interestingly, this

impurity has the same molecular mass as oripavine and due to its potential as an

isomer is of great interest to the company. Further, nor-thebaine is used as a

precursor to pharmaceutical intermediates such as those used for buprenorphine

[185]. Understanding its formation either in the plant or process may provide a

method for simple and cheap production over existing synthetic routes.

Figure 4.2: Structure of Nor-thebaine.

82

The use of in silico optimised 2D-HPLC has been shown to be an effective

way of resolving isomers and recent work, supplementary to this thesis, has shown

the application of this process to the resolution of ephedrine and pseudoephedrine

in methamphetamine seizure samples [186]. In short, this work was developed for

the interrogation of methamphetamine samples including model, real world seizure,

and laboratory synthesised samples. The protocol used Drylab® software to rapidly

identify the optimum separation conditions from a library of chromatography

columns. The optimum separation space was provided by the Phenomonex Kinetex

PFP column (first dimension) and an Agilent Poroshell 120 EC-C18 column

(second dimension). To facilitate a rapid 2D-HPLC analysis the particle packed

C18 column was replaced with a Phenomenex Onyx Monolithic C18 without

sacrificing separation performance. The Drylab® optimised and experimental

separations matched very closely, highlighting the robust nature of HPLC

simulations.

A fully comprehensive two-dimensional separation was completed with a

total analysis time of 115 minutes, the separation is illustrated in Figure 4.3. The

retention times predicted by DryLab® are shown by the white circles, which have

been overlaid on the actual two-dimensional separation (Figure 4.3).

Figure 4.3: Overlaid DryLab® and actual two-dimensional separation of the model

methamphetamine seizure sample.

83

After training the software with data from separations occurring over 2

different gradient durations the simulation was able to extrapolate 2 variables (i.e.

S and kw) for each peak that are required to predict the retention time with varying

gradient durations. DryLab® is able to simulate the retention profile of a complex

separation matrix with a single step (linear) gradient, or by adding multiple steps;

a 4 step gradient is presented here. When this process is done to optimise each

retention dimension individually a 2D-HPLC retention map can be extrapolated

and the final multidimensional simulation is predicted.

This is one of the early points that in-silico software has been used to predict

a two-dimensional separation of methamphetamine samples and indeed

forensically relevant samples at large. This is of great significance for the

development of 2D-HPLC protocols, a process that historically takes considerable

time. The total method development time for the separation presented in Figure 4.3

was less than two hours with in-silico aided optimisation. This is significantly less

than the time required for typical two-dimensional separation optimisation that

normally requires many individual sample injections [153]. The results of 2D-

HPLC separation prediction was impressive as all components of interest were

closely matched (Figure 4.3). This is not a trivial outcome due to the complex

nature of the separation mechanisms involved allowing the user to have high

confidence in the in-silico prediction. The power of 2D-HPLC observed in this

work is directly applicable to the challenge of identifying key components in

complex process chemistry samples. The work in this chapter takes this technology

a step further by incorporating the first application of PCA to two-dimensional

chromatographic data.

84

Experimental

Case studies

The techniques previously described in Chapters 2 and 3, have been applied

to samples from select sections of the process stream. Initially, 2D-HPLC was

performed on each of the samples and compared as mentioned in Chapter 2 and

subsequently analysed using Principle Components Analysis (PCA) as described

in Chapter 3. The samples were then analysed with mass spectrometry in a method

scaled to match the first and second dimension of the 2D-HPLC, as previously

described in Chapter 3. The areas of change were identified from the PCA plots

and all mass spectral peaks within the corresponding retention times were

compared to determine which compound was the cause of the greatest changes in

between the respective steps of the process stream. Specific points of interest were

examined using these methods and reported as individual cases defined by the list

below.

Case 1: G1527-027N, G1527-027O and G1527-027G.

Case 2: G1527-027L and G1527-027M

Case 3: G1528-028C, G1527-028D and G1527-028H

85


Molecules of interest

The opiate processing stream contains many molecules of interest to both

the company and its customers. Three masses of interest have been focused on in

this chapter, where they have been mapped throughout all 81 opiate processing

stream samples in the main line, waste lines and crude impurity standards. The

presence of the molecules at 285 amu, 297 amu and 622 amu are illustrated through

the processing stream in Figures 4.4 and 4.5.

86

Figure 4.4: Morphine processing streams with masses 622 (present when red) and 285 (present when blue).

G1527-027A

G1527-027D

G1527-027F

G1527-027E

G1527-027B

G1527-027H Fresh Solvent

G1527-027K

G1527-027JG1527-027I

G1527-027M

G1527-027L

G1527-027R

G1527-027Q

G1527-027P

G1527-027N

G1527-027O

G1527-027G CPS

Crystallise & precipitate

First aqueous (Aq.) extract

Aq. - tank 2 pH adjustment

Aq. – Post Filtration pH adjustment

Aq. Added complexing agent with impurity removal

Aq. extract tank 1

PSE (partway through extraction)

Aq. input

Aq. waste

Solvent waste

Fresh aq.

Rich solvent

Rich aq. composite

Rich aq. (CE) fresh

Solvent

AEX aq.

87

.

Figure 4.5: Thebaine and oripavine processing streams with masses 622 (preset when red) and 297 (present when yellow).

G1527-028A

G1527-028L

G1527-028E

G1527-028B

G1527-028D

G1527-030H

G1527-030F

G1527-028HG1527-030A

G1527-028F

G1527-028C

G1527-030D

G1527-028G

G1527-028J

G1527-030B

G1527-030I

G1527-030G

G1527-030C

G1527-030L

G1527-030E

G1527-030J

G1527-030M

Oripavine

Precipitate

G1527-030K

G1527-030O

Process continues –No samples

First Aq./ water soluble solvent extract

Heated & water soluble solvent removed

Aq. Filtered (Drum)

Rich solvent

Steam stripped of solventSpent Aq. freshFresh solvent

Rich solvent

Bulk solvent (rich) Thebaine, No oripavine

Rich solvent Caustic

Once extracted solvent (PST)

Rich aq. Oripavine (CEO)

Waste aq.

Waste aq.

Twice extracted solvent (PST 2)

Waste solvent

Acid

Acid wash

Acid

Caustic

Caustic

Acid wash

Aq. Rich extract 2

Aq. Rich extract 1

88

Molecule of interest 1 (285 amu)

Morphine has a molecular mass of 285, however there are isobaric

components present throughout the processing stream that have the same mass and

are considered impurities. One such compound of this mass is of great interest as it

is present in the final morphine product and to date the structure is unknown. The

company are interested in whether the peak was an impurity extracted from the

opium poppy or if it is a product of the processing stream, to determine if it is

something that could be controlled or eliminated. From the LC/MS data of all the

process samples, the impurity was found in the first extraction (G1527-027A) and

continued through the main stream to the CPS morphine (Figure 4.4). This indicates

that the steps taken along the process have not been effective in removing this

impurity. The powdered samples, CPS-M (G1558-131D) and technical morphine

(G1558-131B) were both tested where the latter is a purified form at 94% compared

to 79%. The impurity with a molecular weight of 285 amu was only present in the

CPS morphine (G1558-131D). In the thebaine processing line it was found to only

be present in two samples tested, G1527-030N and G1527-030O which are both at

points along the stream where concentration occurs leading to the possibility that

the impurity is minute in the rest of the stream but becomes noticeable via mass

spectrometry when the stream is concentrated up.

The fragmentation pattern of compounds with the same mass can be used

to elucidate the chemical structure. The two compounds at m/z 286, as shown in

Figure 4.6 were identified by mass spectrometry and have different fragmentation

patterns and chromatographic retention times that vary by approximately two

minutes. The first peak is consistent with the fragmentation pattern of morphine,

however the second is significantly different indicating that it is a structural isomer.

89

a)

b)

Figure 4.6: Mass spectral fragmentation pattern of a) morphine and b) the unknown impurity of the same mass.

50.0 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0 275.0 m/z0.0

2.5

5.0

7.5

Inten. (x100,000)165.1

181.1152.1

128.1

115.055.0 171.0201.091.0 209.1 286.177.1

50.0 75.0 100.0 125.0 150.0 175.0 200.0 225.0 250.0 275.0 m/z0.0

0.5

1.0

1.5

2.0

2.5

Inten. (x100,000)185.1157.0

58.1

129.1193.1

229.2153.1211.4 267.2122.8

90

To further characterise this compound and the following two, NMR

spectroscopy needs to be performed however due to the complex nature of these

samples the target compounds required separation from the other components in

the sample to enable structural elucidation. This was attempted by utilising a one-

dimensional HPLC separation with the previously discussed diphenyl column on

an analytical scale. Using an Agilent fraction collector after the detector, target

peaks were collected on a time based scale, the mobile phase evaporated and then

reconstituted with deuterated methanol before NMR analysis. A portion of the

fractions was taken for direct injection mass spectrometry to confirm the collected

peaks were the target compound. Whilst this was confirmed the lack of sensitivity

of the NMR meant the peaks could not be characterised, even after in excess of 100

fractions were collected and combined. Semi-preparative HPLC was also trialled

to increase the amount of sample that could be utilised however due to poor

chromatography, this idea was also abandoned. Due to the difficulties with

collecting enough of the target compound an alternate way to concentrate the

impurity is required however this is now beyond the scope achievable here.


The second mass of interest is 622 amu which corresponds to the compound

O-M-Somniferine (Figure 4.1). This mass was investigated as it is a known

impurity of the thebaine processing stream, however the results found with LC/MS

also identified a mass corresponding to the m/z 623 molecular ion in the morphine

samples. This is a mass not currently looked for in the morphine processing line,

however it is postulated to be an asymmetric dimer of thebaine. The MS/MS data

showed thebaine in the fragmentation pattern and as such this supported the

proposed compound assignment. As illustrated in Figure 4.4 this compound is in

the first sample extract from the poppy straw pellets and continues through the main

processing stream until the caustic wash where it is expelled into the recycled

solvent waste. This suggests that it originates in the raw product and is efficiently

extracted under the conditions afforded by the process. Finding this impurity in the

recycled solvent waste may also have implications on the process as the impurity

is being reintroduced into the stream towards the start of the process, potentially

leading to an increased concentration throughout the batches. This highlights the

importance of tracking impurities as this is potentially a sampling point that can be

91

monitored to ensure the concentration of the impurity is not increasing significantly

and possibly causing issues in the extraction process.

This compound is also observed in the thebaine stream, only emerging at

G1527-030A, the rich solvent after thebaine and oripavine are separated (Figure

4.5). This compound stays in the stream despite caustic impurity washes and is

removed at the point of the acid washes. During the acid washes the compound is

absent from the main stream but appears in each of the sequential downstream acid

washes. It is postulated here that this occurs due to the main stream being

significantly more dilute than the acid wash sample, where the compound is

gradually washed into these lines.

At the first acid wash stage (G1527-O30G) that uses organic acid an isomer

of m/z 623 is observed denoted by two peaks of the same mass with different

retention times. This highlights the importance of the concentration steps as the

second isomer is only observable when concentrated downstream. As such for any

thorough investigations into these compounds that Sun Pharmaceuticals may need

to do for a client, concentration of the sample streams prior to the acid wash would

be required. Alternatively, this isomer could be forming at this stage due to a

chemical reaction induced by a lowering of the pH.

At the third acid wash stage using mineral acid (G1527-O30K) the original

isomer of m/z 623 was no longer observed. This indicates that the efficiency of the

organic acid wash was sufficient to remove the primary m/z 623 isomer. Further,

the mineral acid washes completely remove the second m/z 623 isomer from the

process stream. This work highlights the issues associated with compounds of the

same mass that are observed only at certain points in the stream and this technology

allows process owners to carefully monitor these compounds.


The third mass of significant interest is 297 amu, which corresponds to the

impurity nor-thebaine (Figure 4.2). This compound is a known impurity in the

thebaine stream and shares the same mass as oripavine and a further unknown

compound present primarily in the morphine stream. Due to its presence in the

combined thebaine and oripavine stream, this impurity is of particular interest. The

impurity was not detected in the morphine stream, however its presence was traced

92

through the thebaine and oripavine stream as illustrated in Figure 4.5 denoted by

yellow highlights.

This impurity elutes at 6.9 minutes and as such elutes after oripavine. From

this the impurity is first present at sample point G1527-028L, after the opiates have

been extracted from the straw with water miscible solvent and subsequently heated

to remove the solvent. The impurity is then present in the main processing stream

until the acid washes. During the organic acid treatment the impurity is present in

both the washes and the process stream. At the mineral acid stage the remainder of

the compound is extracted out. In the sample G1527-030G, this compound is found

to be the most concentrated, exceeding the detection capabilities of the LC/MS.

This sample is the first acid wash to form aqueous rich extract and as such is more

concentrated. This compound is also found in the rich aqueous oripavine and the

resulting mother liquor. This shows the capacity to gain information on quality

from the concentration stages of the processing streams.

Case 1: G1527-027N, G1527-027O and G1527-027G.

Case 1 is from the morphine processing stream and focuses on the samples

G1527-027N, G1527-027O and G1527-027G. G1527-027N (see Figure 4.4) is

fresh rich aqueous caustic extract. G1527-027O directly follows G1527-027N and

is the rich aqueous composite. This is then subjected to a crystallisation and

precipitation step, where CPS morphine is recovered and G1527-027G is the

mother liquor left from this process (Figure 4.4).

The PC1 × PC2 scores plots for G1527-027N and G1527-027O were first

consulted to determine the sample variance. Figure 4.7 illustrates that 63.5 % of the

sample variance is accounted for in PC1 and 92.3 % by the first two components.

93

Figure 4.7: PC1 × PC2 scores plot of G1527-027N and G1527-027O samples.

The loadings plot for G1527-027N and G1527-027O were compared and

the changes investigated. It was found that there were significant changes between

these two samples with eight major differences. This is illustrated in Figure 4.8 and

the retention times of these changes are shown in Table 4.1.

Figure 4.8: PC1 loadings plot for G1527-027N against G1527-027O from 2D HPLC analysis.

94

Table 4.1: The top eight PC1 loading scores for G1527-027N and G1527-027O

First Dimension RT

Second Dimension RT

PC1 Loadings score

24 0.5 1.000 28 0.8 0.479 28 1.0 0.466 32 3.3 0.172 24 0.6 0.172 72 3.1 0.165 32 0.6 0.144 80 3.3 0.105

These retention times were then compared to the profiling solutions LC/MS

data, where an ion count threshold of 1,000,000 was utilised to focus on the

important large changes. It is important to note that the focus here was placed on

the larger changes in order to test the protocol and that for broad application

individualised assessment of the mass spectral thresholds will need to be made. As

the technology developed here gets greater use an informed peak count threshold

will be required however the threshold described here was shown to be sufficient.

Care will need to be taken for application of this to other sample types beyond

opiates which may possess lower mass spectral ionisation efficiencies. The changes

are highlighted in Table 4.2

Table 4.2: Pseudoolecular ion mass to charge ratios, total ion counts and retention

times for the peaks with significant changes observed in the loadings plots for

G1527-027N and G1527-027O respectively (1,000,000 threshold).

Ion m/z Ion RT (mins)

Adj. RT (mins)

G1527-027O (Ion counts)

G1527-027N (Ion counts)

344.23 5.69 28.44 4129200 2329157

329.20 5.71 28.57 1171678 0

328.18 5.73 28.65 5010879 5270546

194.10 6.41 32.04 1507660 0

328.18 6.54 32.70 19225078 14066945

329.20 6.55 32.73 4228279 2858543

314.22 6.58 32.88 8123023 5099602

338.36 16.06 80.31 1408922 1634005

95

The changes at a retention time of 28.44 and 28.57 minutes (Figure 4.8)

correspond to the molecular ions m/z 344 and m/z 329 respectively. The mass to

charge ratio is consistent with the compound laudanine (Figure 4.9) which is readily

known in the opiate steam as it is a biosynthetic intermediate [187] and as such this

was compared to its impurity standard (G1527-034M), an in house fractionated

sample of laudanine (Figure 4.9). These standards are routinely separated and

concentrated within the Sun Pharmaceuticals R and D laboratory at Port Fairy. The

bespoke nature of these materials makes it difficult to find commercial standards

and as such it is important to note that these standards typically vary in their

impurity levels and care needs to be taken when using them. Interestingly the

laudanine standard was found not to match the retention time of 28.44 minutes

appearing approximately 5 minutes earlier. This implies that the molecule is likely

to be similar to laudanine including its interaction with the diphenyl stationary

phase.

Figure 4.9: Structure of Laudanine.

As mentioned previously, an ion count threshold of 1,000,000 was set, in

this study to limit the data size to a reasonable and workable amount, however if

necessary the threshold may need to be adjusted to enable a thorough investigation

of the changes of interest to other sample sets. To visualise the implications of a

significant adjustment to the ion count threshold a study was performed on the same

data set as presented in Figure 4.8 with a new threshold of 500,000 ion counts. The

molecular masses of the compounds responsible for the major changes are

presented in Table 4.3.

96

Table 4.3: Molecular ion mass to charge ratios, total ion counts and

retention times with the peaks with the most observable change in PC1 loadings for

G1527-027N and G1527-027O (500,000 threshold)

Ion m/z Ion RT (mins)

Adj. RT (mins)

G1527-027O (Ion counts)

G1527-027N (Ion counts)

344.22 5.68 28.40 4129200 2329157 345.24 5.69 28.46 998658 0 329.20 5.72 28.60 1171678 1139468 328.18 5.73 28.65 5010879 5270546 194.10 6.41 32.04 1507660 1044668 328.18 6.54 32.70 19225078 14066945 329.20 6.54 32.72 4228279 2858543 360.20 6.56 32.82 862555 772804 314.22 6.57 32.86 8123023 5099602 330.21 6.57 32.86 746301 0 315.21 6.59 32.93 1962637 996321 171.14 14.40 72.00 543210 0 228.21 14.43 72.17 693820 0 540.44 14.50 72.47 790429 680697 334.21 14.50 72.48 0 582340 304.22 14.54 72.71 856586 1004728 378.26 14.57 72.86 753073 785251 228.20 14.58 72.92 568811 0 436.29 16.09 80.44 694525 625891

As can be observed from Table 4.3, the change in threshold allowed changes

at 72 minutes to be identified. This highlights the importance of adjusting the

threshold to allow a better understanding of changes if no masses are noted at the

retention times identified by PCA analysis.

The next sampling point of interest to case study 1 is G1527-027G, which

directly follows G1527-027O (Figure 4.4). Figure 4.10 describes the loadings plot

for the samples G1527-027G and G1527-027O. As can be noted from the plot a

large amount of change can be observed between 16 and 17 minutes in the first

dimension. This is also observed in the PCA loadings scores in Table 4.4, where

the seven significant changes in the samples are shown with the biggest change

observed at a retention time of 16 minutes.

97

Figure 4.10: PC1 loadings plot for G1527-027G against G1527-027O from 2D HPLC analysis.

Table 4.4: The seven PC1 loading scores for G1527-027G and G1527-027O

First Dimension RT

Second dimension RT

PC1 loadings score

16 0.4 1.000000 16 0.3 0.779080 36 2.7 0.166485 16 0.5 0.102417 32 0.5 0.081450 56 0.3 0.076003 24 1.1 0.070652

The mass spectral peaks were compared to the known retention time

differences noting that this was performed at an ion count threshold of 1,000,000.

The mass spectral peaks that were identified at these retention times are shown in

Table 4.5, where the values highlighted in yellow are those that produced the most

significant change at that respective retention time.

98

Table 4.5: Molecular ion mass to charge ratios, total ion counts and

retention times with the peaks with the most observable change in PC1 loadings for

G1527-027G and G1527-027O. The highlighted rows indicate the values

associated with the most notable changes corresponding to table 4.4.

Ion m/z Ion RT Adjusted RT

Ion Count G1527-027G

Ion Count G1527-027O

324.11 3.29 16.45 0 6651041 286.16 3.31 16.55 20000000 20000000 308.18 3.31 16.56 0 20000000 325.13 3.33 16.63 0 1421907 669.25 3.33 16.63 3807454 0 268.14 3.33 16.64 2387128 19796688 287.17 3.33 16.64 20000000 19350048 593.27 3.33 16.64 0 2963348 288.20 3.33 16.66 8296774 12787382 192.12 3.34 16.70 20000000 0 193.16 3.34 16.72 8179070 0 316.18 3.36 16.78 1174037 0 302.18 3.36 16.79 19904082 0 603.30 3.36 16.79 11272815 0 604.20 3.36 16.79 4500447 0 240.16 3.36 16.80 7938151 4955320 318.16 3.39 16.94 7439495 0 208.13 3.39 16.96 3024467 0 314.19 4.80 24.02 1429778 0 490.25 4.87 24.34 1725044 0 194.10 6.41 32.04 6184907 1507660 344.22 6.45 32.23 5324913 0 328.18 6.54 32.70 20000000 19225078 329.20 6.55 32.73 4971785 4228279 360.18 6.55 32.77 7421490 0 378.18 6.56 32.81 2050925 0 314.22 6.58 32.88 20000000 8123023 316.19 7.22 36.12 1276425 0 314.23 7.23 36.13 2151253 0 329.23 7.23 36.17 9909379 2319867 328.21 7.24 36.18 20000000 12676532 274.29 11.31 56.53 19983254 0 275.31 11.32 56.60 3660635 0 346.31 11.33 56.65 1199210 0 230.28 11.38 56.88 7978807 0

From Table 4.5, it is clear that the three changes at 16 minutes in the first

dimension are due to the molecular ions m/z 308, m/z 192 and m/z 302. The m/z

302.18 is characteristic of 10-Hydroxymorphine, where it is present in the G1527-

99

027G sample point but is absent at the G1527-027O sample point. As the G1527-

027G sample point directly follows G1527-027O the lack of this mass in the data

is possibly due to oxidation or other reactions of compounds within the sample over

time. 10-Hydroxymorphine is a degradation product of morphine [188] leading to

the possibility that it further degrades under the right conditions to form a different

impurity. As the morphine and other unknown compounds are removed through

crystallisation, new compounds now constitute a significantly higher percentage of

the remaining liquors and due to this the large change in the presence of 10-

Hydroxymorphine can be more easily observed. This is consistent with unpublished

data from Sun Pharmaceuticals where under high pH an increase in three as yet

unknown impurities has been observed [174]. This again highlights how this type

of comprehensive sample analysis can identify key changes and can enable the

company to prioritise points of interest for detailed investigation.

This is supported by the second dimension separation where m/z 302.18 is

at an adjusted retention time of 0.288 minutes, which is consistent with the PCA

loadings plot identifying the second biggest change at 16 minutes in the first

dimension and 0.3 minutes in the second dimension. Also consistent is the intensity

change, whilst in the first dimension this mass was absent in G1527-027O, the

second dimension had a relatively small intensity (4697069 counts) indicating that

the mass is present at a much lower concentration than in G1527-027G (19633025

counts) again a direct observation of the successful elimination of a range of

impurities at this point of the morphine processing stream.

Importantly several other key components have been identified via this

process which may be targets for further study by Sun Pharmaceuticals when

defining future issues for customers. The change at 24 minutes is from the

molecular ions m/z 490, the change at 32 minutes m/z 314, the change at 36 minutes

m/z 328 and the change at 56 minutes from m/z 274. After discussion with Sun

Pharmaceutical scientists the molecular ion m/z 314 has been postulated to be

codeine methyl ether which is commonly observed in these extracts. Codeine

methyl ether has previously been reported for use to make thebaine with high yields

(83 %) in comparison to other potential starting products making it an important

alkaloid with prospective commercial viability [25].

100

Case 2: G1527-027L and G1527-027M

Case 2 is also from the morphine processing stream however the focus here

is on the solvent waste procedure. Sample points G1527-027L and G1527-027M

are both partially spent solvent and it is important to note they come from two

different Karr reciprocating-plate extraction columns that are used in this stream

before the next step and are expected to contain similar components. An example

of a Karr reciprocating column is shown in Figure 4.11 as described by Smith et al.

[189]. It is a solvent extraction column that utilises plates that move in a

reciprocating fashion to separate and purify components [190] and as such has a

wide variety of applications in chemical industries, including pharmaceutical and

petrochemical [189]. The difference between the two columns is that the first

column (sample point G1527-027L) has fresh caustic running through it whereas

the second column (sample point G1527-027M) utilises the partially rich caustic

from the first column. The sample G1527-027Q is next in line downstream from

these and is then followed by G1527-027R, which is recycled in the process at

sample G1527-027D (Figure 4.4).

Figure 4.11: Depiction of a Karr reciprocating plate-extraction column as described by Smith et al. [189].

101

As can be noted from the loadings plot in Figure 4.12, there is only one

major change between G1527-027L and G1527-027M and approximately five

minor changes denoted by the circle diameter. The PC1 loadings values were

investigated to find the coordinate retention times as displayed in Table 4.6. From

this plot the major chemical changes can be observed and correlated to the mass

spectral data.

Figure 4.12: PC1 loadings plot for comparison of G1527-027L and G1527-027M.

Table 4.6: PC1 loadings values for G1527-027L and G1527-027M.

First dimension

RT Second dimension RT PC1 loadings score

44 2.7 1.000000

44 1.5 0.119075

44 1.4 0.102645

32 2.9 0.064433

44 1.8 0.020958

44 1.6 0.016224

102

The main change between the two large scale processing columns was

interrogated with LC/MS to determine the mass of the compound which was

determined to correspond to m/z 171.14. When the 2D-HPLC plots were

consulted, as shown in Figure 4.13 the change between the two samples by eye is

small highlighting the importance of this type of PCA data handling. This

protocol has allowed the biggest changes between the samples to be found and

scales them from 0 to 1, which is an important aspect to be aware of as in the case

of these two samples the change is small.

103

Figure 4.13: 2D-HPLC contour plots of the samples G1527-027L and G1527-027M.

104

The ion m/z 171.14 has not previously been identified as a key component

of interest to Sun Pharmaceuticals. Here for the first time the company has the

capacity to assess compounds based on the PCA data processing which may afford

a deeper understanding of the components that play key roles in differentiation

between samples. This highlights the significance of the work developed here as it

could be applied to monitoring key low concentration components in batch to batch

processing samples to help inform process owners.

Case 3: G1528-028C, G1527-028D and G1527-028H

This case is from the thebaine and oripavine processing stream and is at the

point where they are separated from each other. Sample point G1527-028C is the

rich solvent containing both thebaine and oripavine, they are then separated with a

caustic wash before G1527-028D the thebaine only stream and G1527-028H the

purely oripavine stream. The 2D-HPLC contour plots for these samples are shown

below in Figure 4.14.

105

Figure 4.14: 2D-HPLC contour plots for G1527-028C and G1527-028D.

106

As highlighted by the contour plots (Figure 4.14) there are minimal small

changes observed between these two samples, however importantly there is one

large change which is due to the removal of oripavine from the stream. This was

also confirmed by the mass spectral data that showed the key change was due to

m/z 298 the molecular ion of oripavine which has been extracted. The second

largest observable change was m/z 344 which was confirmed with the use of the

standard to be the molecular ion of laudanine. This again highlights the complexity

of real world process chemistry samples and that isomers can be challenging to

monitor. Here we have shown that we can differentiate between key compounds of

the same mass and as such add to the capacity for Sun Pharmaceuticals to

interrogate samples more clearly.

This clear data set could be easily used as a visual inspection tool by the

process owners to quickly assess the quality of the separation of thebaine and

oripavine. In this case PCA data analysis would only be required to understand the

significance of the minor components within this sample. The components of low

concentration observed here are likely to be compounds that prefer solvent

conditions in the oripavine stream over the thebaine stream. This will be primarily

driven by the pH adjustment afforded by the caustic wash step and could be a key

point in the process where Sun Pharmaceuticals could focus on for efficiency

improvements.

The differences between sample points G1527-028C and G1527-028H

were then investigated to see what followed in the oripavine stream, where the

biggest change is expected to be the removal of thebaine. The contour plots were

first compared to identify the obvious changes in the two samples. Figure 4.15

shows the contour plot for sample point G1527-028H (rich in oripavine), and when

compared to G1527-028C (Figure 4.14A) the removal of the large peak associated

with thebaine at 40 minutes (First Dimension) and 2.4 minutes (Second Dimension)

can be observed.

107

Figure 4.15: Contour plot for G1527-028H.

The PC1 and PC2 loadings plots and scores were then consulted to

determine other changes that may be significant to the process. It was important for

the PCA that the samples were of the same concentration to eliminate a big change

being identified due to differing concentrations. Due to this G1527-028H was

diluted by a further factor of 10 to bring it into line with G1527-028C and G1527-

028D before PCA was applied. As can be noted in the PC1 loadings plot of G1527-

028C and G1527-028H (Figure 4.16), the biggest change identified is the removal

of oripavine.

108

Figure 4.16: PC1 loadings plot for the comparison of G1527-028C and G1527-028H.

Figure 4.17: PC2 loadings plot for the comparison of G1527-028C and G1527-028H.Showing the comparison between the two samples

109

The location of the major change (removal of thebaine) in Figure 4.16 is at

40 minutes in first dimension and 2.65 minutes in the second dimension. The next

biggest change is observed at 48 minutes in the first dimension and 2.75 minutes in

the second dimension and was identified as the previously mentioned molecular

ion m/z 171, highlighting the need for further investigation into this molecule that

is beyond the scope of this project.

This case highlights the importance of this protocol in processing samples

of increased complexity. It was clear from the 2D-HPLC contour plots that this data

was more complex than those previously investigated and it has been shown that

the key compounds of interest can be identified. Importantly minor constituents

have been observed to contribute to the loading plots and these compounds can now

be tabulated into data sets that could be applied for future use by Sun

Pharmaceuticals. The key use of these data sets will be to have a focus for any

future queries on these types of samples and gives Sun Pharmaceuticals some

capacity to pre-empt areas of concern/interest to clients.

110

Conclusions

Three masses of interest have been explored and traced through the

processing stream to determine if they originate from the opium poppy itself or as

a product of the extraction and isolation process. All three masses (unknown

impurity MW285, Nor-thebaine and O-M-Somniferine) were identified in the first

sample extracts and as such were deemed to have originated in the opium poppy

itself before being extracted alongside the target compounds.

The methods described in Chapter 2 and 3 have successfully been applied

to determine the differences between points of interest on the manufacturing

stream. Three points of interest from the opiate processing stream were investigated

and using a combination of 2D-HPLC, PCA and LC/MS differences could be

elucidated. The concept was proven to be effective by focusing on a major known

change, in which PCA identified this as a large area of difference on the two-

dimensional plots and was confirmed with mass spectrometry taking into account

the chromatographic variations observed between the two instruments. These

techniques provide Sun Pharmaceuticals with the means to identify large areas of

change between stages of the process or batches and to pinpoint the associated

masses. This is a tool that can be utilised as a quality control method for the

company or to pinpoint potential impurities.

111

CHAPTER 5:

IDENTIFICATION AND SELECTIVE DETECTION OF SYNTHETIC CANNABINOIDS

112

Chapter overview

This chapter investigates the isolation of synthetic cannabinoids from herbal

substrates followed by a systematic study of selective detection methods for these

compounds. Several herbal blends purchased in Victoria, Australia before the

recent ban are explored in order to detect and determine the synthetic cannabinoid

compounds present. Numerous standards were also synthesised at Deakin

University to fully test the detection methods on a variety of different compounds.

Chemiluminescence was used to detect these compounds with three different

reagents, of which two proved successful for the detection of all but two of the

thirteen synthetic cannabinoids. Limits of detection were calculated with both

reagents for all standards and an estimated concentration of the cannabinoids on the

corresponding herbal substrates was determined. A solid phase extraction

selectivity study was also conducted, with the use of 6 different sorbents in order

to improve the isolation of the cannabinoid from the chemically complex herbal

substrates.

113

Introduction

Synthetic cannabinoids have been abused since the early 2000s, and have

become of increasing concern to law enforcement agencies and health

professionals. These drugs have a vast array of side effects and as highlighted in

Chapter one (Figure 1.3), are responsible for many disastrous effects for users of

these herbal alternatives [40, 42]. There are a large number of known synthetic

cannabinoids and new analogues are constantly entering the market [191]. New

synthetic cannabinoids have been produced by varying the (usually heterocyclic)

core, bioisosteric replacement and/or changing or adding functional groups, often

with the intention of creating derivatives that were not prohibited by the law [44,

192, 193].

Currently to determine these compounds the samples are first extracted

using organic solvents (e.g., methanol, ethanol or acetonitrile) to exploit the non

polar nature of the synthetic cannabinoid, typically at a ratio of 1 mL of the solvent

to 10 mg of dried plant material [37, 194]. Generally, the sample extracts are then

either sonicated or vortexed and centrifuged before filtering with a 0.45 µm filter

[38, 112, 195]. The sample extracts are then analysed and the compounds detected

and identified with a variety of different techniques such as GC and LC mass

spectrometry [54].

In previous work, mass spectrometry was used to identify the actives

present in seven of the twelve synthetic cannabinoid samples utilised in this study

[196]. This technique for the identification of synthetic cannabinoids has been well

documented in the literature where it is either coupled to liquid chromatography or

more commonly gas chromatography [54].

Chemiluminescence is a sensitive and selective mode of detection, easily

coupled to HPLC, which has been exploited for the determination of numerous

compounds of forensic importance [55, 79]. Rasanen and co-workers have used

gas-phase chemiluminescence nitrogen detection for designer drugs including

some synthetic cannabinoids [197], but to date, liquid-phase chemiluminescence

has not been reported for the detection of these compounds.

Electrochemiluminescence (ECL) is a technique that has also not previously been

114

reported for use with synthetic cannabinoids. Herein, the complementary selectivity

derived from solvent extraction, chromatographic separation and

chemiluminescence detection for the determination of synthetic cannabinoids in

complex herbal sample matrices is explored.

Solid Phase Extraction

Due to the complexity of the synthetic cannabinoid samples, removing

unnecessary components from the samples will aid ‘at scene’ detection. A common

method of sample clean-up, solid phase extraction (SPE) is a selective and rapid

preparation method used for isolation, concentration, clean up and medium

exchange [198]. The aim of SPE is to separate the analytes of interest from the other

components in the sample before chromatography or analytical methods [198], and

has been used on the synthetic cannabinoid UR-144 for extraction from the herbal

product. Kavanagh et al. [61] utilised a C-18 SPE cartridge for this which is a

cartridge that has also been used for synthetic cannabinoids in urine [199, 200].

Whilst SPE is not often used to clean up herbal extractions it is commonly

used for biological samples where it is an important purification and concentration

step before mass spectrometry analysis [54]. SPE has also been used for large

variety of cannabinoids in differing biological samples such as urine [201, 202],

blood [203] and oral fluid [35, 204, 205]. Urine studies have included extracting

and quantifying the metabolites of the synthetic cannabinoids JWH‐018 and JWH‐

073 with a strong cation exchange SPE cartridge [206].

115

Experimental

Chemicals, samples and standards

Twelve commercially available herbal synthetic cannabinoid samples

(brands: Puff, Red Dot, Bombay Blue - pineapple flavour, Cloud 9, Code Black -

strawberry flavour, Northern Lights - Special K, Malibu, Atomic Bomb, Stoner,

Storm, Supanova and Voodoo; Figure 5.1) were obtained in 2013/2014 pre-ban

from a range of stores in Victoria, Australia. Synthetic cannabinoids were extracted

by adding 10 mg of the herbal mixture to 1 mL of the extraction solvent and

sonicated for 10 minutes. The organic solvents used in this study were acetone,

methanol, ethyl acetate (Chem-Supply, SA, Australia) and acetonitrile (Ajax Fine

Chemicals, NSW, Australia); all solvents were of analytical grade or higher.

Figure 5.1: Synthetic cannabinoid brands utilised in this study.

Seven synthetic cannabinoid standards (AM-2201, AB-001, 5F-AKB-48,

5F-AB-001, JWH-018, JWH-073 and 5F-AB-PINACA) were synthesized at

Deakin University. A further two synthetic cannabinoid standards (PB-22 and 5F-

PB-22) were obtained from the National Measurement Institute (NSW, Australia).

Stock solutions of these were initially prepared at 1 mM using HPLC grade

methanol (Merck, Vic, Australia) and diluted as required.

116

High Performance Liquid Chromatography (HPLC)

An Agilent Technologies 1260 HPLC with an auto injector, solvent degasser,

quaternary pump, column thermostat (20°C) and a DAD ( = 254 nm) and

chemiluminescence detector were used. Chemiluminescence reagents were

pumped to the custom-made detector with a Gilson Minipuls peristaltic pump at 1

mL min-1 as described by Adcock et al. (Figure 5.2) [207]. Reagent and column

eluent were merged at a T-piece shortly prior to entering the coiled tubing flow-

cell, as described by Holland et al. [208] (Figure 5.2). An Agilent Eclipse XDB‐

C18 column (4.6 mm × 15 mm, 5 μm particle diameter) was used for all separations.

The sample (20 µL) was injected onto the column with a flow rate of 1 mL min-1.

The mobile phase comprised HPLC-grade methanol (Merck) and Milli-Q water

(Merck Millipore, Victoria, Australia). In method A, the methanol was increased

from 0% to 100% over 40 minutes, and then held for 20 minutes, with 5 minute re-

equilibration time. In method B, both the methanol and deionised water contained

0.1% (v/v) formic acid; the methanol component was increased from 10% to 80%

over 10 minutes, and then to 100% over 10 minutes, with 5 minute re-equilibration

time.

Figure 5.2: Custom HPLC chemiluminescence detector setup as described by Adcock et al. [207]. Key: q = quaternary pump; v = injection valve; c = column; u = UV-visible absorbance detector or other non-destructive detector; p=peristaltic pump; d = flow-

through chemiluminescence detector.

117

Chemiluminescence detection

A 1 mM solution of acidic potassium permanganate (Chem-Supply) was

prepared in deionised water containing 1% (m/v) sodium polyphosphate (Sigma-

Aldrich), adjusted to pH 2.5 with concentrated sulfuric acid (Merck).

The stable tris(2,2′ bipyridine)ruthenium(II) perchlorate salt

([Ru(bipy)3](ClO4)2) was prepared by an aqueous metathesis reaction as described

by McDermott et al. [209]. Tris(2,2′ bipyridine)ruthenium(II) chloride

hexahydrate was purchased from Strem Chemicals (CA, USA) and the sodium

perchlorate was from Sigma-Aldrich. The [Ru(bipy)3](ClO4)2 salt (0.384 g L-1) was

dissolved in HPLC grade acetonitrile (Ajax) with 0.05 M perchloric acid (Ajax).

Lead dioxide (Fluka Analytical) was added (2 g L-1) and the reagent was filtered

in-line.

Manganese(IV) was prepared according to the method of Brown et al. [86].

Manganese dioxide was formed by reducing potassium permanganate (Chem-

Supply) with an excess of sodium formate (Sigma-Aldrich) and collecting the

precipitate under vacuum on glass filter paper (GF/A, Whatman, England). The

solid was washed three times with deionised water. The wet manganese dioxide

(approx. 1.2 g L-1) was placed in 3 M orthophosphoric acid (85% w/w, Chem-

Supply) and sonicated for 30 minutes before heating (80°C) whilst stirring for

approximately 1 hr. The solution was then cooled on ice and an iodometric titration

was performed using potassium iodide (Ajax), where the iodine released upon

reaction with manganese(IV) was titrated against sodium thiosulfate (Sigma-

Aldrich) solution, with a starch (Ajax) indicator determining the end-point. As an

enhancer 2 M Formaldehyde (37% w/w, Chem-Supply) was used, merging with

the HPLC eluent after the DAD but before entering the chemiluminescence

detector.

Electrospray mass spectrometry

Pre‐separated cannabinoid fractions were directly injected into an Agilent

Technologies 6210 LC/MSD ESI-TOF equipped with a 1200 series solvent

degasser, binary pump (methanol, 1 mL min-1) and 1100 series ALS. Positive ion

polarity was used under the following conditions: drying gas (N2), flow rate and

temperature (5 L min-1, 350°C), nebuliser gas (N2) pressure (30 psi), capillary

118

voltage 3.0 kV, vaporiser temperature 350°C, and cone voltage 60 V. MS data

acquisition and analysis was carried out using Agilent MassHunter Workstation

Acquisition for TOF/Q-TOF (version B.02.00 (B1128)) and MassHunter

Qualitative Analysis (version B.03.01).

Electrochemiluminescence (ECL) and Cyclic Voltammetry (CV)

Cyclic voltammetry (CV) was carried out with a glassy carbon working

electrode, silver wire as the reference electrode and platinum wire as the counter

electrode. Samples were prepared in dry acetonitrile with 0.1 M

Tetrabutylammonium hexafluorophosphate ([TBA][PF6], Fluka). A scan rate of 0.1

V s-1 was utilised for all samples, and each was referenced to 1 mM ferrocene.

Electrochemiluminescence spectra were collected with a photomultiplier

tube (PMT). As with the CV experiments a glassy carbon working electrode and

platinum wire counter electrode was utilised. Synthetic cannabinoid standards at 1

mM were first tested with Tetrabutylammonium hexafluorophosphate as a blank

before tris(2,2′‐bipyridine)ruthenium(II) hexafluorophosphate (Aldrich) was added

and the mixture re-tested. Herbal samples were tested as above, where the synthetic

cannabinoids (20 mg) were extracted with dry acetonitrile (2 mL).

Solid Phase Extraction (SPE)

Six solid phase extraction (SPE) cartridges, detailed in Table 5.1 were

purchased from Agilent Technologies. The SPE vacuum manifold was purchased

from Phenomenex (Lane Cove, NSW, Australia).

Table 5.1: Details of the Solid Phase Extraction cartridges utilised.

Cartridge Sorbent Sorbent Size

(mg/mL) Part Number

Bond Elut-C18 EWP Silica wide pore C18 50 mg, 1 mL 12102136

Bond Elut-C8 Silica C8 50 mg, 1 mL 12102059

Bond Elut-SAX Silica strong anion exchange 100 mg, 1 mL 12102017T

Bond Elut-SCX Silica strong cation exchange 50 mg, 1 mL 12102075

Bond Elut-Plexa PCX Polymer strong cation exchange 30 mg, 1 mL 12108301

Bond Elut-SI Natural silica 100 mg, 1 mL 12102010

119

All SPE cartridges were conditioned as described in Table 5.2, with the

exception of the natural silica cartridge that was only conditioned with the

extraction solvent. Samples were extracted at 10 mg of herb to 1 mL of solvent

(vortexed, filtered and analysed on HPLC) before being added to the conditioned

cartridge. The cartridges were then washed with 5% methanol (100% extraction

solvent for silica cartridge) before elution with 100% methanol (100% water for

silica cartridge). The first pass (sample solution), second pass (wash solution) and

the third pass (elution) were collected for HPLC analysis.

Table 5.2: Cartridge conditioning and extraction solvents.

Conditioning (1 mL each) Extraction solvent

(1 mL/10 mg)

Methanol , water Methanol

Methanol , water Acetonitrile

Acetonitrile, water Acetonitrile

Methanol , water Acetone

Acetone, water Acetone

Methanol, water Ethyl acetate

Ethyl acetate, water Ethyl acetate

120


Synthetic cannabinoid extraction

Eight different solvents (methanol, ethanol, glacial acetic acid, DMSO,

DMF, acetone, chloroform, ethyl acetate) were trialled for the extraction of

cannabinoids from the herbal samples, during earlier work [196]. In that study

methanol, acetone and ethyl acetate were shown to be the preferred solvents. In this

work acetonitrile was also tested and compared to those three identified best

extraction solvents. Chromatograms for extracts from the 'Puff" sample, for

example are shown in Figure 5.3.

For each solvent extract from this sample, a large peak was observed in the

chromatogram at 37.7 minutes (identified as synthetic cannabinoid UR-144, as

discussed in the following section). The minor components were dependent on the

solvent used for extraction. Methanol is commonly used for the targeted extraction

of small polar molecules from within plant matrices [210] and it is likely that the

minor peaks observed before 25 minutes include polyphenolic compounds from the

herbal substrate. Due to the absorbance of acetone at 254 nm a large solvent front

was observed in the chromatogram when it was used for extraction. Acetonitrile

enabled extraction of large amounts of the cannabinoid without many of the

extraneous peaks seen with most other solvents and is recommended as the ideal

solvent for a targeted extraction and reduced sample complexity. Due to the

inhomogeneous nature of the herbal substrates it is difficult to directly compare the

concentration of synthetic cannabinoid extracted by the different solvents, however

all solvents have extracted a significant amount of the target compounds and as

such a clean extraction is the focus.

121

Figure 5.3: Chromatograms (HPLC with UV absorbance detection at 254 nm; separation method A) of the herbal product ‘Puff’ following extraction with: (a)

methanol, (b) acetone, (c) ethyl acetate, and (d) acetonitrile. The synthetic cannabinoid (UR-144) peak can be observed at 37.7 minutes.

0

20

40

60

80

0 10 20 30 40 50 60

Absorbance (mAu)

(a)

0

20

40

60

80

0 10 20 30 40 50 60

Absorbance (mAu)

(b)

0

20

40

60

80

0 10 20 30 40 50 60

Absorbance (m

Au)

Time (min)

(d)

122

Identification

Previously, five synthetic cannabinoids were identified in seven brands of

herbal ‘incense’ with the use of GC/MS and LC/MS as listed in Table 5.3. These

seven brands were analysed further in this study in addition to five new herbal

synthetic cannabinoid samples. Fractions of interest from chromatographic

separations were collected manually and analysed by ESI-MS to identify the

synthetic cannabinoids. A tentative assignment was made using ESI-MS, by

comparing the molecular ion peak to a database of known compounds and utilising

monographs made available online through www.SWGdrug.org [211]. The

analysis identified a further three synthetic cannabinoid compounds (Table 5.4).

The brands ‘Atomic Bomb’, ‘Stoner’ and ‘Storm’ contained the same two

compounds with m/z 359 and m/z 377, which were provisionally assigned as

cannabinoids PB-22 and 5F-PB-22. ‘Supanova’ also contained the m/z 359

compound. The assignments were confirmed by matching the HPLC retention

times with PB-22 and 5F-PB-22 standards. The brand 'Voodoo' also only contained

one synthetic cannabinoid (m/z 378) however is only tentatively assigned 5F-AEB

as there is little information currently available for this compound.

123

Table 5.3: Synthetic cannabinoids previously identified in the extracts of herbal samples.

Structure Common name(s) IUPAC name Brands

AM-2201 1-[(5-fluoropentyl)-1H-indol-3-

yl]-(naphthalen-1-yl)methanone

Red Dot

UR-144 (1-pentylindol-3-yl)-(2,2,3,3-

tetramethylcyclopropyl)methanone

Cloud 9; Code Black;

Puff; Special K

XLR-11 (1-(5-fluoropentyl)-1H-indol-3-

yl)(2,2,3,3-


Bombay Blue; Cloud 9;

Special K

A796,260 [1-(2-morpholin-4-ylethyl]-1H-

indol-3-yl]-(2,2,3,3-


Code Black

5F-AKB-48;

5F-APINACA

N-(adamantan-1-yl)-1-(5-

fluoropentyl)-1H-indazole-3-

carboxamide

Malibu

124

Table 5.4: Identified synthetic cannabinoids in the extracts of new herbal samples.

Structure Common name(s) IUPAC name Brands

PB-22; QUPIC 1-pentyl-1H-indole-3-carboxylic

acid 8-quinolinyl ester

Atomic Bomb; Stoner;

Storm; Supanova

5F-PB-22;

5F-QUPIC

1-pentyfluoro-1H-indole-3-

carboxylic acid 8-quinolinyl

ester

Atomic Bomb; Stoner;

Storm

5F-AEB;

5F-EMB-PINACA

ethyl (1-(5-fluoropentyl)-1H-

indazole-3-carbonyl)-L-valinate

*Voodoo

* Tentatively assigned

125

Chemiluminescence detection

Chemiluminescence detection has previously been utilised for the detection

of synthetic cannabinoids in the brands Red Dot, Puff, Cloud 9, Bombay Blue,

Special K, Code Black and Malibu [196]. From this previous study it was found

that acidic potassium permanganate didn’t emit light with the synthetic

cannabinoids unlike the natural cannabinoids [88]. However, they were found to

react with stable [Ru(bipy)3]3+ for four out of the five synthetic cannabinoids.

Under identical conditions (extraction with methanol; HPLC separation method

B), chromatograms were obtained for all nine synthetic cannabinoid standards

(Table 5.5) and the twelve herbal samples, with UV absorbance detection at 254

nm, and then chemiluminescence detection using three different reagents:

permanganate, manganese(IV) and [Ru(bipy)3]3+. The below Table (5.5) includes

the standards utilised in this study, in addition to the synthetic cannabinoid brands

in Tables 5.3 and 5.4. Typical chromatograms produced using the four modes of

detection (using the ‘Atomic Bomb’ sample as an example) are shown in Figure

5.4.

126

Table 5.5: Synthetic cannabinoid standards.

Structure Common name(s) IUPAC name

5F-AB-PINACA N-(1-amino-3-methyl-1-

oxobutan-2-yl)-1-(5-


carboxamide

5F-AKB-48;

5F-APINACA

N-(adamantan-1-yl)-1-(5-


carboxamide

5F-AB001 Adamantan-1-yl[1-(5-

fluoropentyl)-1H-indol-3-

yl]methanone

AB001 1-pentyl-3-(adamant-1-

oyl)indole

AM-2201 1-[(5-fluoropentyl)-1H-indol-3-

yl]-(naphthalen-1-yl)methanone

JWH-018 Naphthalen-1-yl(1-pentyl-1H-

indol-3-yl)methanone

JWH-073 1-Butyl-3-(1-naphthoyl)indole

PB-22;

QUPIC

1-pentyl-1H-indole-3-


ester

5F-PB-22;

5F-QUPIC

1-pentyfluoro-1H-indole-3-


ester

127

Figure 5.4: Chromatograms for the same herbal blend extract ‘Atomic Bomb’ obtained using separation method B with: (a) UV absorbance detection, and (b-d)

chemiluminescence detection using (b) acidic potassium permanganate, (c) manganese(IV), or (d) [Ru(bipy)3]3+. The two major peaks were assigned as synthetic

cannabinoids 5F-PB-22 and PB-22 (Table 5.3).

0

30

60

90

120

0 5 10 15 20

Absorbance (mAu)

(a)

0

2.5

5

7.5

10

0 5 10 15 20

Intensity (a.u.)

(b)

0

15

30

45

60

0 5 10 15 20

Intensity (a.u.)

(c)

0

20

40

60

80

0 5 10 15 20

Intensity

(a.u.)

Time (min)

(d)

128

The acidic potassium permanganate reagent was examined for the

chemiluminescence detection of synthetic cannabinoids in the herbal samples

because it was previously used to detect natural cannabinoids in industrial grade

hemp [88, 212]. Permanganate reacts with a wide variety of organic compounds

[213], but a much smaller subset elicit the characteristic red emission (max = 689

nm) at analytically useful intensities from excited manganese(II) [94, 96, 214]. This

reagent is particularly effective for the chemiluminescence detection of readily

oxidisable phenols, anilines and related compounds, but many species outside these

classes have also been found (sometimes unexpectedly) to generate an intense

emission [94, 96], including several non-phenolic indoles [215, 216]. When

screening the cannabinoid standards and herbal samples with this reagent (e.g.

Figure 5.2b), however, no chemiluminescence signal was obtained but they were

clearly detected by UV absorbance (Figure 5.4a). In contrast to the alkyl phenol

backbones of the natural cannabinoids [88], the synthetic mimics investigated here

are indole and indazole derivatives (Table 5.3, 5.4 and 5.5), and a dissimilar

chemiluminescence response was not surprising. The permanganate reagent did

detect some other components of extract, which were likely to include polyphenolic

compounds from the herbal materials and may pose analytical utility for substrate

characterisation.

The chemiluminescence of manganese(IV) also emanates from an

electronically excited manganese(II) species (λmax = 730 nm) [86], but this reagent

generates light with a much wider range of compounds than permanganate [97].

Despite both reagents forming an excited manganese(II) species, the

manganese(IV) has a higher maximum intensity (730 nm) compared to acidic

potassium permanganate (689 nm) due to the addition of the sodium polyphosphate

enhancer [79]. Non-phenolic indoles, such as tryptamine, tryptophan and

indomethacin, have been reported to elicit light with manganese(IV) [86, 217].

Unlike acidic potassium permanganate, the manganese(IV) reagent produced a

chemiluminescence signal with each of the synthetic cannabinoids identified in the

herbal samples (e.g. Figure 5.4c), with the exception of 5F-AKB-48 (both as a

standard and in the ‘Malibu’ sample). The manganese(IV) reagent also produced a

chemiluminescence signal with each of the synthesized cannabinoid standards with

the exception of 5F-AB-PINACA and the aforementioned 5F-AKB-48. The

129

reasons for the stark difference in selectivity between the two manganese-based

chemiluminescence reagents are yet to be fully elucidated, but appear to arise from

a combination of the lower oxidation state of manganese(IV) than permanganate,

the heterogeneous nature of reactions with the colloidal manganese(IV) reagent,

the concentration of acid and the enhancers added to each system [86, 218-221].

The absence of chemiluminescence from the combination of 5F-AKB-48 or 5F-

AB-PINACA and manganese(IV) can be ascribed to the significantly higher

oxidation potential of these indazole-3-carboxamides compared to the other

synthetic cannabinoids under discussion [222], which are indole-3-carbonyl

(ketone or ester) compounds.

The [Ru(bipy)3]3+ reagent is generally an effective chemiluminescence

reagent for the detection of aliphatic tertiary amines [92] and although a variety of

other nitrogen containing compounds also produce light with this reagent, the

emission intensity is often substantially lower for those with fewer N-alkyl

substituent's, incorporating mesomeric or resonance stabilisation, or with electron

withdrawing groups in close vicinity to the nitrogen [223]. Nevertheless, some

aromatic nitrogen-containing heterocyclic compounds that do not contain

peripheral amines, including indoles such as tryptophan [224] and indomethacin

[225], have been reported to produce chemiluminescence upon reaction with

[Ru(bipy)3]3+ at intensities sufficient for analytical applications. An appraisal of the

herbal samples with the [Ru(bipy)3]3+ reagent (e.g. Figure 5.4d) showed detectable

responses for all of the synthetic cannabinoids except 5F-AKB-48 (present in the

‘Malibu’ sample) and 5F-AB-PINACA, similar to the results obtained with the

manganese(IV) reagent.

The standard and herbal extraction of 5F-AKB-48 and the standard 5F-AB-

PINACA both failed to give a chemiluminescence reaction with any of the reagents,

however a quenching was observed with the manganese(IV) reagent (Figure 5.5)

The other synthetic cannabinoid intensities varied with the different reagents where

neither [Ru(bipy)3]3+ nor manganese(IV) proved to be superior to the other (Table

5.6). Normalised chromatograms (Figure 5.6) highlight that the difference in

intensity is greatly dependant on the analyte, as is which reagent is more effective.

For example the standard PB-22 elicits a greater signal with [Ru(bipy)3]3+ (Figure

5.6a) but the standard AB-001 has a greater response with manganese(IV) (Figure

130

5.6b). [Ru(bipy)3]3+ however, doesn't require an enhancer like manganese (IV) and

as such would be a more applicable and user friendly reagent for an at-scene

detection test.

Figure 5.5: Chromatogram of the standard 5F-AKB-48 with UV detection (green) and Manganese(IV) (brown).

Table 5.6. Comparison of chromatographic peak areas for cannabinoid

standards with chemiluminescence detection (green tick > 115, orange tick 0-115,

red cross < 0). The 115 scale has been established to allow for clear differentiation

of the components.

Cannabinoid

Peak area (a. u.)

[Ru(bipy)3]3+ Mn(IV)

PB-22 5F-PB-22 AM-2201

5F-AKB-48 AB-001

5F-AB-001 5F-AB-PINACA

JWH-018 JWH-073

59

60

61

62

63

64

65

66

67

68

0

20

40

60

80

100

120

140

160

0 5 10 15 20

Intensity (mAu)

Absorbance (mAu)

Time (mins)

131

Figure 5.6: Normalised chromatograms of the synthetic cannabinoid standards PB-22 (Blue) and AB-001 (Red) with [Ru(bipy)3]3+ and Manganese(IV) detection respectively.

55

60

65

70

75

80

85

90

95

100

0 5 10 15 20

Itensity (mAu)

Time (mins)

55

60

65

70

75

80

85

90

95

100

0 5 10 15 20

Intensity (mAu)

Time (mins)

132

Mobile phase was also found to have an impact on the signal produced by

[Ru(bipy)3]3+. These particular analytes emit a larger signal with a methanol mobile

phase in comparison to an acetonitrile mobile phase (Figure 5.7) The manganese

reagents also require a methanol mobile phase as acetonitrile quenches the signal

[94].

Figure 5.7: [Ru(bipy)3]3+ comparison of acetonitrile and methanol mobile phase for the standard 5F-AB-001.

Solutions of the synthetic cannabinoids standards were used to generate

calibration curves using HPLC with UV absorbance (254 nm) and/or

chemiluminescence detection, and the resulting figures of merit are shown in Table

5.7. It is important to note some calibration equations are linear while others are

quadratic. The intensities for the quadratic equations fell into the non-linear

chemiluminescence calibration curve range and as such a second order polynomial

was fitted to accurately find the limits of detection. Non-linear calibration models

are often required when an extended calibration range is utilised [226, 227].

50

55

60

65

70

75

80

0 5 10 15 20

Intensity (mAU)

Time (mins)

Acetonitrile

Methanol

133

Table 5.7: Analytical figures of merit

Analyte Detection RT (mins) Calibration Equation R2 LOD (M)

5F-AB-PINACA 254nm 11.90 y = 1.27 × 106x - 11.2 0.9990 9 × 10-6

5F-AKB48 254nm 16.81 y = 3. × 106x – 84.2 0.9944 9 × 10-4

5F-AB001 254nm 16.69 y = 9.86 × 105x - 0.8 0.9987 8 × 10-7

Mn(IV) 16.67 y = 1.98 × 105x + 29.9 0.9946 8 × 10-5

Ru(bipy)23+ 16.82 y = 41,114.78x + 0.1 0.9952 1 × 10-6

AB001 254nm 18.85 y = 1.44 × 106x – 0.4 0.9995 3 × 10-7

Mn(IV) 18.91 y = 5,342.59x + 1.6 0.9857 3 × 10-5

Ru(bipy)23+ 19.04 y = 7.23 × 105x + 1.7 0.9921 9 × 10-6

AM2201 254nm 14.18 y = 9.92 × 106x – 40.5 0.9999 3 × 10-6

Mn(IV) 14.36 y = -2.03 × 108x2 + 5.35 × 105x + 42.9 0.9992 1 × 10-5

Ru(bipy)23+ 14.36 y = 1.72 × 105x + 18.5 0.9937 5 × 10-5

JWH-018 254nm 16.23 y = 1.53 × 106x -1.2 0.9991 8 × 10-7

Mn(IV) 16.43 y = 130,085.71x + 9.4 0.9979 3 × 10-5

Ru(bipy)23+ 16.37 y = -3.86 × 109x2 + 1.47 × 105x + 0.1 1.0000 4 × 10-7

JWH-073 254nm 15.25 y = 8.77 × 105x - 1.0 0.9971 1 × 10-6

Mn(IV) 15.23 y = 51,200.00x + 22.8 0.9953 5 × 10-5

Ru(bipy)23+ 15.39 y = -2.44 × 109x2 + 1.10 × 105x + 0.1 1.0000 9 × 10-7

PB-22 254nm 14.83 y = 8.25 × 106x + 0.6 0.9998 4 × 10-7

Mn(IV) 14.97 y = -9.53 × 109x2 + 1.77 × 106x + 3.4 0.9993 6 × 10-6

Ru(bipy)23+ 15.06 y = 8.15 × 106x + 2.2 0.9980 1 × 10-6

5F-PB-22 254nm 13.21 y = 1.08 × 107x + 23.8 0.9953 1 × 10-7

Mn(IV) 13.35 y = -8.57 × 109x2 + 1.87 × 106x + 5.7 0.9979 3 × 10-6

Ru(bipy)23+ 13.43 y = 7.51 × 106x + 1.9 0.9975 1 × 10-6

134

From these calibrations, the total cannabinoid content extracted with 1 mL

methanol from 10 mg of herbal product was calculated. Methanol was used as an

extraction solvent here for consistency with standards and due its use as a mobile

phase. The synthetic cannabinoid content was found to range from 0.036 mg to 0.77

mg per 10 mg of product (Table 5.8). It can be noted that there is considerable

variation in the amount of cannabinoid material on the substrate. Whilst these are

only representative of a small selection of all brands that were and are available,

this highlights the dangers associated with the use of these products.

Table 5.8: Concentration of synthetic cannabinoid on herbal substrates extractions

Brand Synthetic

cannabinoid

Concentration

(M)

Concentration

(mg/mL)

Red Dot AM-2201 6.9 × 10-4 0.250

Malibu 5F-AKB-48 2.0 × 10-3 0.770

Atomic bomb PB-22 1.3 × 10-4 0.047

5F-PB-22 9.5 × 10-5 0.036

Storm PB-22 9.9 × 10-4 0.360

5F-PB-22 3.3 × 10-4 0.130

Stoner PB-22 1.9 × 10-4 0.068

5F-PB-22 1.5 × 10-4 0.057

Supanova PB-22 2.3 × 10-4 0.082

UV adsorption at 254 nm generally offers a more sensitive detection as

illustrated in Table 5.7. However chemiluminescence detection affords superior

selectivity for these samples. The limits of detection determined for the two

chemiluminescence reagents are similar and hence either could be used for the

analytes, however [Ru(bipy)3]3+ is a more user friendly reagent as it doesn't require

a formaldehyde enhancer and extra equipment for analysis [86].

Electrochemiluminescence (ECL) and Cyclic Voltammetry (CV)

Electrochemiluminescence involves the use of an applied potential to

induce a light emitting reaction, which also has the possibility to be used in an at-

scene detection device. Due to the lack of light obtained from the synthetic

135

cannabinoid 5F-AKB-48, the electrochemical properties of the analytes in solution

were investigated using cyclic voltammetry (CV). The oxidation potential can be

measured using CV, which in turn can help to understand why a compound elicits

light with chemiluminescence. Cyclic voltammetry was carried out on seven

compounds, to determine their oxidation potentials.

Figure 5.8: Cyclic voltammagram (representative of the electron transfer in the system) for the synthetic cannabinoid standard 5F-AKB-48 with reference to ferrocene.

The synthetic cannabinoids tested included PB-22 and 5F-PB-22, both of

which produced a significant amount of light when reacted with [Ru(bipy)3]3+ ,AM-

2201 of which produced a smaller amount of light and 5F-AKB-48 which produced

no light. Also tested was a precursor (Figure 5.9A) and an analogue (Figure 5.9B)

of 5F-AKB-48, both of which also did not produce light with chemiluminescence

(Table 5.9).

‐1.00E‐05

‐5.00E‐06

0.00E+00

5.00E‐06

1.00E‐05

1.50E‐05

2.00E‐05

2.50E‐05

3.00E‐05

3.50E‐05

4.00E‐05

‐0.6 ‐0.1 0.4 0.9 1.4

Current (A)

Potential (V)

136

A) B)

Figure 5.9: Structures of additional compounds tested to understand the chemiluminescence results; (a) 5F-AKB 48 Precursor and (b) 5F-AKB 48 analogue.

Tris(2,2′‐bipyridine)ruthenium(II) itself was also tested to find the

oxidation potential as was codeine, which is known to produce a significant amount

of light with this co-reactant. Table (5.10) depicts the oxidation potentials with

reference to ferrocene.

Table 5.9: Comparison of chromatographic peaks areas for additional

cannabinoid standards with chemiluminescence detection.

Cannabinoid Peak area (a.u.)

[Ru(bipy)3]3+ Mn(IV)

5F-AKB-48 Precursor 0 0

5F-AKB-48 Analogue -24 0

Table 5.10: CV oxidation potentials with reference to ferrocene.

Sample Concentration (M) Oxidation cf FC

Ru(bpy)32+ 1.0 × 10-3 0.89

5F-AKB-48 analogue 1.0 × 10-3 1.18

5F-AKB-48 Precursor 1.0 × 10-3 1.50

5F-AKB-48 2.5 × 10-4 1.40

AM-2201 1.0 × 10-3 1.16, 1. 52

PB-22 1.0 × 10-3 1.08, 1.25

5F-PB-22 2.5 × 10-4 1.08, 1.24

Codeine 1.0 × 10-3 0.64, 1.03

137

Interestingly, the oxidation potentials of the synthetic cannabinoids are

higher than the oxidation potential of Ru(bipy)32+. It has previously been thought

that the Ru(bipy)33+ reagent only produces light with compounds with a lower

oxidation potential to itself (0.89), such as codeine. However the cannabinoids with

oxidation potentials 1.16 and below produced chemiluminescence and those above

1.18 produced no emission with tris(2,2′‐bipyridine)ruthenium(II). This is

important as it sets an oxidation potential limit to these compounds for

chemiluminescence detection with tris(2,2′‐bipyridine)ruthenium(II). Some of the

synthetic cannabinoids have two oxidation potentials indicating they are oxidised

at one potential and then further oxidised at a higher potential. It is of interest that

the compounds that have two oxidation potentials are those that produce light with

chemiluminescence.

Electrochemiluminescence (ECL) was then trialled to determine which

compounds produced light with Ru(bipy)33+ when a potential was applied.

Interestingly, cannabinoid 5F-AKB-48, as well as its precursor and analogue

(which did not exhibit emission with chemiluminescence) were able to produce

light with ECL, as can be seen in Figure 5.10. In contrast, the standards which could

be measured with chemiluminescence did not afford signals with ECL. This

indicates that with the use of both methods combined, all synthetic cannabinoid

compounds tested in this study could be selectively detected in a portable manner.

This technique was also carried out for the synthetic cannabinoid herbal

samples, where extractions were performed with dry acetonitrile to determine if

this would be a viable at-scene detection method. Six of the synthetic cannabinoid

brands (Bombay Blue, Cloud 9, Code Black, Puff, Special K and Voodoo) tested

did not have associated standards and as such whether they would react with ECL

was unknown. As suspected, five of these synthetic compounds did not produce

light when subjected to an applied potential. As the standards that produced light

with chemiluminescence did not react with ECL, it was not surprising that these

compounds also didn't react with ECL given their affinity for Ru(bipy)23+

chemiluminescence. The brand Storm also didn't produce light with ECL, however

this is consistent with the results obtained from the standards PB-22 and 5F-PB-22.

The brands Atomic Bomb and Stoner were not tested as they were known to only

contain PB-22 and 5F-PB-22 in a lesser concentration than Storm.

138

Voodoo (unknown compound) was an exception where light was produced

with both chemiluminescence and ECL. Supanova also produced light with ECL

despite being known to contain PB-22, a synthetic cannabinoid whose standard did

not emit light when subjected to an applied potential. It is proposed that this sample

contains further unidentified components, as other large peaks were visible on the

UV HPLC chromatogram. It is possible that these compounds are unknown to date

as no synthetic cannabinoids could be matched to their respective molecular ions

in mass spectrometry (m/z 216 amu and m/z 375). Concentration of the compounds

and NMR analysis would help, however in this instance insufficient quantities for

NMR analysis of the cannabinoid were able to be collected.

Figure 5.10: Synthetic cannabinoid standard 5F-AKB-48 analogue with ECL detection.

The brand Malibu contains the synthetic cannabinoid 5F-AKB-48, of

which a standard had previously been tested (Figure 5.10) and found to produce

light. The extraction from the herbs produced less light than the standard at 1 mM

due to a lower unknown concentration of extracted synthetic cannabinoid. This

however is important as ECL can still be used to detect the extracted synthetic

cannabinoid and with further optimisation this has the potential to be used in an at-

scene detection test.

‐0.002

0

0.002

0.004

0.006

0.008

0.01

0.012

0.014

0.016

‐0.5 0 0.5 1 1.5 2

139

Solid Phase Extraction

Six solid phase extraction (SPE) cartridges (C18, C8, SAX, SCX, PCX and

natural silica) were trialled as a sample clean up method to eliminate components

co-extracted from the herbs in order to develop a more targeted identification

protocol. The four organic solvents (acetonitrile, acetone, ethyl acetate and

methanol) identified as the most efficient at extracting the synthetic cannabinoid

from the herb were used as the initial extraction solvent. The solvent extractions

discussed earlier in this chapter indicated that acetonitrile and acetone co-extract

less components from the herbs but still a significant amount of cannabinoid.

Samples were extracted with one of the four organic solvents before being

added to the conditioned cartridge, where it was discovered that the cannabinoid

was breaking through the sorbent on all cartridges, eluting as the sample was added.

However other components were generally retained on the sorbent until washing

with 5% methanol (100% extraction solvent for natural silica cartridge) or elution

with 100% methanol (100% water for natural silica cartridge).

140

A)

B)

Figure 5.11: Chromatograms obtained from each stage of SPE extraction (Plexa PCX cartridge with methanol conditioning and acetonitrile initial extraction) for the brand

“Puff” using UV detection A) actual chromatogram (offset) B) zoomed in chromatogram (offset). Legend; Initial = before SPE, Sample = first pass, Wash = second pass and

Elution = third and final pass.

0

100

200

300

400

500

600

700

800

0 5 10 15 20

Absorban

ce (mAu)

Time (mins)

Initial

Sample

Wash

Elution

0

10

20

30

40

50

60

70

80

90

100

0 5 10 15 20

Absorban

ce (mAu)

Time (mins)

Initial

Sample

Wash

Elution

141

As can be seen from Figure 5.11, the characteristic chromatograms

produced from the extraction of the brand “Puff” have a large peak at

approximately 17 minutes which is the synthetic cannabinoid UR-144. The

chromatograms also show a small peak at the beginning which is likely to be a

solvent front from the extraction solvent. The peak at approximately 13 minutes is

an impurity as it is also present in blank samples. This and all other peaks are very

small in comparison to the cannabinoid peaks. Figure 5.11B shows that there are

still peaks in the sample solution however in comparison to the sample pre-solid

phase extraction it can be noted that they have been reduced.

The solution collected from the wash also contains a large amount of

cannabinoid, which is potentially from the residual sample solution held in the

sorbent as it can’t be allowed to dry out during the SPE process. It is expected that

the cannabinoid should be retained on the sorbent however due to the large

concentration of the cannabinoid, it breaks through the sorbent in the sample. The

synthetic cannabinoid is still present in the elution solution due to the sorbent

retaining a certain amount of the compound. Previous testing indicated that at a

smaller concentration the synthetic cannabinoid is also fully retained until the third

and final pass, however the other non-polar compounds are also retained, leading

to the use of these cartridges as a filtration system with a large concentration of

synthetic cannabinoid [196].

The initial extraction of each sample was analysed with HPLC before SPE

was performed as this provided the opportunity to see how much of the cannabinoid

was lost during the SPE process.

Table 5.11: Results of solid phase extraction for the brand Puff.

Cartridge Conditioning Extraction

solventNo. of Peaks

% Retained

Silica Ethyl acetate Ethyl acetate 4 57

C8 Acetonitrile, water Acetonitrile 3 54

C18 Acetonitrile, water Acetonitrile 3 58

C18 Methanol, water Acetonitrile 2 55

SAX Acetone, water Acetone 5 52

SCX Acetonitrile, water Acetonitrile 5 63

PCX Acetonitrile, water Acetonitrile 4 55

PCX Methanol, water Acetonitrile 4 62

142

As illustrated in Table 5.11, between 52% and 63% of the cannabinoid was

retained in the sample solution. This means that the solid phase extraction process

loses approximately 33%-50% of the cannabinoid, however due to the sensitivity

of the detection this is not an issue. The majority of the cannabinoid was retained

in the sample solution, however the remainder of the cannabinoid was eluted in the

wash and elution stages.

The first pass sample which contains the breakthrough of the synthetic

cannabinoid was analysed using chemiluminescence detection after optimisation of

the method. Although the UV indicated that a few small peaks remained in the

extracted sample after solid phase extraction, when analysed with Ru(bipy)33+ the

only component that produced light was the synthetic cannabinoid. Figure 5.12

illustrates this, as even with a sample where 5 peaks are present in the UV other

than the synthetic cannabinoid peak, the only signal produced by reaction with the

chemiluminescence was the cannabinoid. This illustrates that the methods outlined

above are selective for the synthetic cannabinoids.

Figure 5.12: Chromatogram of an acetonitrile extraction for the brand “Puff” after SPE with a Plexa PCX cartridge with UV (black) and chemiluminescence (red) detection.

0

10

20

30

40

50

60

70

80

90

100

0

50

100

150

200

250

300

350

400

450

0 2 4 6 8 10 12 14 16 18 20

Intensit y (m

Au)

Absorban

ce (mAu)

Time (Mins)

143

Depending on the application and the potential at scene sample matrix

including contamination, different cartridges may best apply and several should be

considered. The C18 cartridge provides the cleanest extraction however only

retains a 55% of the synthetic cannabinoid whereas the Plexa PCX cartridge

produced a larger number of extraneous peaks but retained a larger percentage of

the compound of interest. However, when using selective detection such as

chemiluminescence all tested cartridges can be used as only the synthetic

cannabinoid reacts with the reagent to produce light. For this application the Plexa

PCX cartridge was identified to be the cartridge of choice as it was able to provide

the largest cannabinoid sample.

The work presented in this chapter provides the ground work for the

construction of an at-scene detection test to elucidate if a synthetic cannabinoid is

present on an herbal substrate. The ideal extraction solvent, acetonitrile in

conjunction with a C18 sorbent has been shown to selectively isolate synthetic

cannabinoids in a concentration large enough for detection. Two detection systems

have been trialled here and with further optimisation there is potential for both to

be effective in revealing if these compounds are present. Chemiluminescence and

ECL have both been investigated for use in microfluidics devices and all synthetic

cannabinoids tested here have succeeded with at least one of the two methods.

144

Conclusions

The need for a selective detection method for synthetic cannabinoids has

been highlighted by the vast quantity of available compounds and blends. Polar

organic solvents such as methanol, acetone and acetonitrile provided the most

efficient and selective extraction of the target synthetic cannabinoids from the

herbal materials, which in the products under investigation here were identified as

seven known indole and indazole derivatives. Following solvent extraction, the

synthetic cannabinoids can be determined by chromatographic separation and UV

absorbance detection.

Chemiluminescence with the manganese(IV) or [Ru(bipy)3]3+ reagent was

shown to be a viable alternative for the post-column detection of most synthetic

cannabinoids, although the limits of detection were generally poorer than those

using conventional UV absorbance. The inherent advantages of

chemiluminescence, however, such as superior analytical selectivity and the

relative simplicity of detector components are attractive for the development of

rapid separations and/or at-scene devices for preliminary screening applications.

This preliminary exploration of the chemiluminescence detection of this controlled

class of compound shows that there is considerable scope for further development

in this area.

Solid Phase extraction following a simple solvent extraction with

acetonitrile and a C18 silica cartridge increases the selectivity the extraction and

eliminates the need for a separation technique. This allows the synthetic

cannabinoids to be selectively detected with chemiluminescence without the need

for separation. The methods developed here provide the background chemistry for

an at-scene detection device for synthetic cannabinoids on unknown herbal blends.

145

CHAPTER 6:

TARGETED NMR SPECTROSCOPY FOR SELECTIVE DETERMINATION OF SYNTHETIC CANNABINOIDS

146

Chapter overview

This chapter delves into the use of NMR spectroscopy to directly detect

synthetic cannabinoids on the chemically complex herbal blends. The main focus

of this is on the use of solid state NMR, a novel technique not previously used for

the detection of compounds on already difficult substrates. This technique affords

no damage to the sample which is of particular importance for forensically

significant samples. Experiments investigated included fluorine and both spinning

and static proton NMR. Carbon experiments were also completed with single pulse

excitation and cross polarisation to proton nuclei.

147

Introduction

Synthetic additives sprayed onto plants range from crop protectants such as

herbicides to illicit substances such as synthetic cannabinoids. The emergence of

synthetic cannabinoids has prompted investigations into techniques to detect these

man-made chemicals. The current practices for the determination of synthetic

cannabinoids on herbal substrates involve onerous sample preparation and result in

the destruction of the samples; for example, a solvent or solid phase extraction prior

to chromatographic separation (HPLC or GC) coupled with mass spectral detection

[228-230]. Ideally, multiple techniques could be combined to give an enhanced

fingerprint. While examples exist where solution-state nuclear magnetic resonance

(NMR) spectroscopy has been used for the confirmation of chromatographic peaks

of forensic interest [231, 232] this analysis is performed post-fractionation. Solid-

state NMR is an informative and versatile characterisation technique able to

provide quantitative structural information from inhomogeneous samples in a non-

destructive manner [233-235]. Nevertheless, due to the very weak nuclear

polarisation and correspondingly low inherent signal strength, the ability to directly

detect compounds on surfaces using this method is challenging. The majority of

solid-state NMR studies of surface chemistry involve high surface area materials

such as zeolites [236], metal-organic frameworks [130] or nanoparticles [237].

Although not often used for surface chemistry it is an important technique for

biomolecules [132], nanotechnology [132] and pharmaceuticals [238]. Importantly,

from a forensic perspective where samples are precious, solid-state NMR is a non-

destructive analytical technique that requires little to no sample preparation [239].

This is also particularly important for samples that can degrade when extracted with

solvent. The amount of sample required is generally between 2 mg and 100 mg

depending on the size of the magic angle spinning (MAS) rotor, and typically drug

seizure samples are well in excess of this amount [240].

The technique is not as widely used as solution state NMR due to its low

resolution, however this can be improved with techniques including magic angle

spinning (MAS) and cross polarization (CP) [236]. Magic angle spinning (MAS) is

an essential technique that allows better resolution of peaks as opposed to static

analysis. This involves the rotor being spun at a magic angle with respect to the

148

magnetic field, which narrows the normally wide lines thus improving spectral

resolution [239]. Cross polarisation (CP) is commonly used in solid state NMR to

improve the signal to noise ratio and obtain a more rapid analysis [241]. Cross

polarisation is the transfer of more abundant nuclei such as 1H and 19F to polarise

rare nuclei, including 13C and 15N [242]. Signal enhancement technologies such as

dynamic nuclear polarisation (DNP) have boosted the sensitivity of solid-state

NMR and are opening up a new field of NMR surface characterisation [243].

Previous work utilising solid-state NMR for surface analysis found

sensitivity to be an issue in the detection of surface-immobilised peptides on gold

nanoparticles [244]. In this publication Ashbrook et al. highlight the basic

approaches which can inform on how solids are influenced by anisotropic

(orientation-dependant) interactions including dipolar couplings and chemical shift

anisotrophy which is the driver for the broadening of peaks and subsequent loss in

ability to utlitise the solid state NMR spectra. While magic angle spinning and

decoupling and cross polarisation techniques are used to improve sensitivity and

resolution, solid state NMR has not been used in the capacity highlighted in this

thesis where molecules of interest are interrogated on a solid substrate.

Significant research was also required in the surface analysis of metal

organic frameworks to improve sensitivity where techniques such as the use of

microwaves to induce DNP were required [130]. The techniques outlined here are

standard solid-state NMR methods and equipment, not requiring modification to

the probe as was required in the investigation on surface ethoxy species of acidic

zeolite Y [131]. Early work of solid state NMR has been performed in order to

highlight key features of a surface modification of substrates including silicon [245]

where the reactivity of hydroxyl functional groups on silica with

trimethylchlorosilane (TMCS) was investigated. Interestingly the authors noted

that the technique was able to differentiate between germinal and vicinal silanols

enabling a reactivity mechanism to be postulated. A spectroscopic analysis was

performed however FTIR-PAS and XPS had the capacity to distinguish the features

afforded by this early solid state NMR. This concept was fully realised by Pallister

et al. [246] where quantitative surface coverage analysis was performed on thin

films. This study was able to exploit both 29Si and 13C NMR to generate data on

group V gallium thin films which are wide band gap materials of great interest to

149

optoelectronics and semiconductor manufacturers. In 2000 Ladizhansky and co-

workers [247] interrogated the water binding efficiency on the surface of CdS

nanoparticles that had precipitated from an aqueous solution. In this instance 1H

NMR (MAS) was used to great effect highlighting that three different bindings sites

of the water lead to three different lines in the NMR spectra. Other surfaces have

been studied including nanocrystalline materials such as hydroxyapatite where

Jager et al. [248] were able to fully characterise the material. While solid state NMR

is a fully established technique for the structural characterisation of material on the

atomic scale limited work has been performed on molecules on complex substrates

such as the herbs presented here. Complex backgrounds as demonstrated by these

herbal blends have not been reported before and as such the discovery of the

potential applications outlined here are particularly important.

150

Experimental

Samples

The samples used were brands of herbal based synthetic cannabinoid

products purchased pre ban from various adult and herb stores across Victoria,

Australia. Brands studied were Red Dot, Puff, Bombay Blue, Cloud 9, Special K,

Code Black, Malibu, Atomic Bomb, Storm, Stoner, Supanova and Voodoo.

Damiana (turnera diffusa) was purchased from the happy herb shop, Geelong,

Victoria, Australia. Approximately 100mg of each sample was ground to a fine

powder using mortar and pestle which was then packed into a 4 mm o.d. MAS

NMR rotor for analysis.

Instrumentation and Experiments

A Bruker Avance III 300 MHz Wide Bore was used for all analyses in

conjunction with a 4.0 mm MAS HX probe (Preston, Victoria, Australia). All

experiments were run at 10 kHz MAS spinning rate except for a static 1H

experiment that was carried out on all samples with 4 scans. A 1H MAS experiment

was also carried out with 4 scans. 19F was only carried out on 9 selected samples

with a minimum scan time of 2 hours. 13C CP MAS was carried out at 7.05 T on all

samples with a minimum of 2000 scans and a recycle delay of 5 seconds. 13C single

pulse excitation (SPE) was carried out on 3 samples (Damiana, Bombay Blue and

Cloud 9) at a T1 relaxation time of 10 seconds and on two samples (Damiana and

Bombay Blue) with a T1 relaxation time of 60 seconds.

151


Given that the cannabinoid is sprayed onto the substrate using a volatile

solvent (typically acetone) it was reasoned that the cannabinoid might be ordered (

potentially crystalline) on the surface and as such discernible from the herbal

substrate. Initial 13C MAS and 1H-13C cross-polarisation (CP) MAS NMR

experiments carried out on the brand Bombay Blue and Damiana confirmed that

the synthetic cannabinoids could be identified using this technique as it was evident

that the peak widths of the synthetic molecule were systematically narrower than

that of the herbal substrate, indicative of an ordered solid phase (Figure 6.1). This

feature facilitated ready identification of the signals arising from the synthetic

molecules. Damiana (turnera diffusa) was used as a reference for all NMR

experiments as this herb is most commonly used as a substrate for these commercial

products. This allowed the background spectrum generated from the substrate to be

subtracted, revealing the detailed synthetic cannabinoid spectrum.

For example, subtracting the 13C spectrum of the substrate herb from that of

the Bombay Blue sample clearly revealed the 13C solid-state NMR spectrum for the

cannabinoid XLR-11 (Table 5.3) and its distinct structural features as shown in

Figure 6.1. The distinct structural features of XLR-11 correspond to the following;

the peaks between 0 and 50 ppm correspond to the aliphatic hydrocarbons, the

peaks between 100 and 150 ppm correspond to the carbons of the aromatic rings,

and the peak at 195 ppm is attributed to the carbonyl group (C=O ketone). The peak

at approximately 80 ppm is consistent with a carbon bound to fluorine.

152

Figure 6.1: Normalised 13C CP MAS solid-state NMR spectra of (top to bottom) damiana herbs, synthetic cannabinoid brand Bombay Blue (containing Synthetic cannabinoid

XLR-11) and Bombay Blue with Damiana subtracted. Spectra were obtained with 16,384 scans.

13C CPMAS experiments were also carried out on the remaining synthetic

cannabinoid products and the spectra allow different cannabinoids to be clearly

differentiated as shown in Figure 6.2 and 6.3. The two synthetic cannabinoids can

be clearly differentiated in Figure 6.2 where the Malibu brand is observed to contain

the synthetic cannabinoid 5F-AKB-48 while the Code Black product contains UR-

144. The correlation of the NMR peak positions (chemical shifts) to the molecular

structure allows certain signals to be used as markers of cannabinoid structure class.

For example, the brand Code Black has a ketone resonance at a chemical shift of

195 ppm as do all samples containing UR-144, XLR-11 and AM-2201, but is absent

in samples that do not contain this feature. In contrast, the brand Malibu has an

amide in its place, resulting in an amide peak at a chemical shift of 160 ppm instead.

Samples that contain PB-22 and 5F-PB-22 such as the brand Storm have an ester

peak at approximately 170 ppm. This differentiation in chemical shift due to

structural differences is illustrated in Figure 6. 3.

200 150 100 50 0

Chemical shift (ppm)

153

Figure 6.2: 13C CPMAS solid-state NMR spectra (normalised and with the Damiana herb signals subtracted) from synthetic cannabinoid brands Malibu (5F-AKB-48) (top)

and Code Black (UR-144) (bottom).

Figure 6.3: 13C CP MAS solid state NMR spectra normalised with the Damiana herb subtracted. Synthetic cannabinoid brands from top to bottom are Malibu, Code Black, Bombay Blue, Puff, Red Dot, Voodoo, Storm, Cloud 9, Stoner, Special K, Atomic Bomb

and Supanova.

200 150 100 50 0


250 200 150 100 50 0 -50

Chemical shift (ppm )

154

As seen in Figure 6.1 in the CP MAS spectra, the signals from the substrate

and the synthetic cannabinoid are comparable in size. Although it indicates the

relative amount of herb to the active molecule, CP MAS experiments are not

generally quantitative. Direct 13C solid-state MAS NMR experiments (i.e. without

cross polarisation to the 1H nuclei) were therefore also carried out, with different

recycle delays. As illustrated in Figure 6.4 it was found that the 13C T1 relaxation

times for the herbal substrate were significantly longer than for the synthetic

coatings. A short recycle delay of 2 seconds therefore effectively suppressed the

signal from the substrate, while a 60 s recycle delay showed both components. This

approach, while somewhat more time consuming due to the weak 13C signal and

long recycle delays required, could potentially allow accurate quantification of the

components.

The concentration of the synthetic cannabinoid also has an impact as even

with 1H-13C cross polarisation, some of the cannabinoid signals known to be present

in low concentrations in the samples from previous analysis, were too weak to

resolve in a reasonable timeframe. As shown in Figure 6.3 the brands Special K,

Atomic Bomb and Supanova lack the peaks corresponding to the synthetic

cannabinoid compound known to be present with only a spectra of the herbs

obtained. These samples are known to contain two cannabinoids but in a lower

concentration than the other samples as was determined in Chapter 5.

155

Figure 6.4: Synthetic cannabinoid brand Bombay Blue with 13C SPE MAS solid state NMR. Red spectra is with T1 relaxation time of 2 seconds and Black spectra is with T1

relaxation time of 60 seconds.

In an effort to circumvent legislation and complicate detection, a large

number of synthetic cannabinoids were modified to include fluorine substituents

often as a fluoropentyl chain [44]. With a gyromagnetic ratio almost four times

larger than 13C and a 100% natural abundance, 19F is both a selective and sensitive

means of compound characterisation. As such, 19F MAS NMR was performed on

brands that were suspected to contain a fluorine (e.g. XLR-11, AM-2201, 5F-PB-

22 and 5F-AKB-48). From these samples a clear signal at a chemical shift of

−218ppm was apparent, characteristic of a primary alkyl fluoride. Synthetic

cannabinoids identified in the brands Puff and Supanova lack fluorine substituents

and no 19F NMR signal could be detected as is supported by the lack of a peak at

−218ppm.

250 200 150 100 50 0 -50


156

Figure 6.5: 19F solid state NMR spectra normalised of synthetic cannabinoid brands (from top to bottom) Malibu (5F-AKB-48), Cloud 9 (UR-144 and XLR-11), Bombay Blue

(XLR-11), Red Dot (AM-2201), Voodoo, Atomic Bomb (5F-PB-22 and PB-22), Code Black (UR-144), Puff (UR-144), Supanova (PB-22) and Damiana (plain herb).

As can be seen in Figure 6.5, a further peak at δ50ppm is present in all

samples including the plain herbal blend, Damiana. This peak was further

investigated to determine if a fluorine was present in the herbal blend by running

an experiment on the rotor and cap without a sample. It was found that this fluorine

peak was also present in the empty rotor spectra and hence is a background signal

possibly from the rotor cap (Figure 6.6). The rotor is known to be made from

zirconium dioxide (ZrO2), but the caps for these rotors can be made from a variety

of different compounds. However, due to the difference in chemical shift it doesn’t

interfere with the fluorine peak from the synthetic cannabinoids and is not an issue

when analysing these samples.

200 100 0 -100 -200 -300 -400

Chemical Shift (ppm)

157

Figure 6.6: Spectra from the empty rotor, plain herb and a fluorine containing synthetic cannabinoid herbal coating.

1H MAS and static 1H NMR (Figure 6.7) were both uninformative on

whether synthetic cannabinoids were present on the herbal blends. It is generally

accepted that 1H solid state NMR is low resolution and provides little information

about the protons. However, information can still be gained from this analysis in

the form of the mobility of the sample. The mobility of the sample relates to how

the structure can reorient depending on the domain. For example with polyethylene

it is in a highly ordered or crystalline domain the polymer chains can only reorient

very slowly [249]. However if they are in non-crystalline domain the polymer

chains are more mobile and as such this can give indication of crystallinity of the

compounds being detected [249]. As the structures and substrates of all he synthetic

cannabinoids are similar the spectra does not significantly vary (Figure 6.7).

158

Figure 6.7: 1H spin solid state NMR normalised spectra of synthetic cannabinoid brands (from top to bottom) Malibu, Supanova, Atomic Bomb, Special K, Stoner, Cloud 9, Storm,

Voodoo, Red Dot, Puff, Bombay Blue and Code Black.

This thesis highlights that sensitive determination of the synthetic

cannabinoid is not an issue. In comparison to GC/MS, the technique currently used

in forensic determination of these compounds, there are some differences and as

such it would be beneficial to use these methods in conjunction with each other. As

previously eluded to solid state NMR requires very little sample preparation in

comparison GC/MS which requires extraction of the compounds into a volatile

solvent. The sample preparation for GC/MS involves grinding the sample into a

powder before adding either methanol or ethanol and ultrasonicating for 10

minutes, centrifuging for a further 5 minutes and then filtering out the herbal blends

[112, 199]. This takes significantly longer than the sample preparation required for

solid state NMR where the sample is only ground to a powder before analysis.

Literature reports the analysis time for GC/MS ranging between 15 minutes and 60

minutes [111, 135, 199]. Whilst this is an improvement on the 3 hour cross

polarisation analysis time it is important to note that a comprehensive study on

improving analysis time was not performed here and could go some way to

enhancing the techniques highlighted in this chapter.

80 60 40 20 0 -20 -40 -60 -80


159

The use of 13C and 19F solid-state NMR experiments have proven to be a

powerful, non-destructive approach for the detection of synthetic cannabinoids as

a component of herbal blends. Using both 13C CP MAS and 19F MAS experiments

key structural features can be readily differentiated. This has implications and is

extremely important for forensic analysis of these as the sample isn’t required to be

destroyed in order to determine if an alterant has been applied. Sample preparation

is straightforward and these experiments are standard and do not require high

magnetic fields or fast spinning rates. This approach will only become more

powerful as sensitivity enhancement techniques such as the recently identified DNP

become more reliable. Beyond the obvious implications for forensic analysis, this

approach has the potential to be used in other applications where herbs or plants

have had compounds added such as herbicides on foreign plant or vegetable

imports.

160

Conclusions

Forensically relevant synthetic cannabinoid compounds on the surface of

herbal substrates have been identified using 13C and 19F solid-state NMR methods.

Further to this, structural differences in compounds can also be determined. Solid-

state NMR is not typically amenable to the study of molecules on surfaces, however

the synthetic cannabinoids studied here provide both high sensitivity and

resolution, indicative of an ordered solid phase. The use of this technique has the

capacity to deliver a selective detection protocol for synthetic cannabinoids on

complex forensic samples in a non-destructive manner. This technique also has the

potential to be used in other applications where a compound has been added to

herbs or plant material.

161

Future Work

This thesis explores chemistry in both pharmaceutical and forensic science

areas and as such the application for future research is limitless. The work presented

here has the potential to be expanded and some ideas have been suggested below.

The mapping techniques delivered in chapters 2, 3 and 4 illustrate the broad

applications to the processing streams undertaken by Sun Pharmaceuticals,

however this technique could be used in other chemical manufacturing companies

to monitor their processes. The exact limits and requirements for these techniques

at each stage of the process need to be developed more precisely to target the exact

information required by the company. This also has the capability to be utilised in

the identification of new impurities introduced in bespoke batches of crop from a

range geographic areas and annual growth cycles to identify issues early in the

process.

Chapter 4 explored the identification of a few select impurities of interest

as this is of significant importance to Sun Pharmaceuticals. The future work in

relation to this portion is to develop techniques to identify the chemical composition

and the potential impact they have on the final product. In particular the impurities

such as those at 285 amu for example are of great importance to identify. It has

been pinpointed from this research where the largest concentration of the 285 amu

impurity is in the process and as such this sample may be explored in order to isolate

enough of this compound for NMR analysis. Other impurities of interest are also

able to be pinpointed as to where they are most concentrated in the stream and as

such also have a greater potential to be isolated for identification. Newly identified

impurities have the potential to be isolated in the process and utilised as future

important medications or as a precursor to make compounds of interest more

straightforward as was the case with the now indispensable compound, thebaine.

The main aim of the synthetic cannabinoids study presented in chapter 5 is

to work towards an at-scene detection test for these herbal substrates. The chemical

background to allow this device to be made has been achieved however placing it

into a device is dependent on engineering as there is a fine line as to what can be

162

achieved with both chemistry and engineering. The future aim of the chemistry

developed here is to fit it into a microfluidic lab on a chip that would allow at scene

detection of these compounds. In theory these chips would be hand held disposable

and cheap devices that could connect to a smart phone or similar to give quick and

reliable results.

Also in chapter 5, Electrochemiluminescence (ECL) was shown to react

with some synthetic cannabinoids, however only the surface of this technique has

been touched. There are a large variety of complexes that can be used in ECL

including different ruthenium and iridium complexes. Whilst work has previously

been done on these complexes, their reactions with the synthetic cannabinoids

could lead to a more targeted approach to these compounds. A fundamental study

of why some of these cannabinoids elicit light with ECL/chemiluminescence and

others don't is also of interest.

In chapter 6, solid state NMR was shown to detect synthetic cannabinoids

on the surface of herbal blends, which may have implications to detect coatings on

other plants. To develop this technique further, a larger sample set of synthetic

cannabinoid standards and different herbal mixtures could be explored to

potentially build a library of different herbal substrates and synthetic cannabinoid

spectra. The non-destructive nature this technique offers has implications in

forensic science and could potentially be used as a screening technique to test if

synthetic cannabinoids are present and if so the potential class of the compounds.

To achieve faster and cleaner results the parameters surrounding these could be

explored and modified to identify the most efficient method. Future work on with

this technique could also include exploring areas such as farming where plants that

chemicals have been applied to could be tested in order to find what remains on the

surface.

Importantly continuation of the work in this thesis is currently underway by

both a PhD and Honours student highlighting the importance of the information

generated in this thesis affords further scientific endeavours. I wish the students the

best of luck in their endeavours as I have greatly enjoyed this study program.

163

REFERENCES

1. Wu, H., J. Guo, S. Chen, X. Liu, Y. Zhou, X. Zhang, and X. Xu, Recent

developments in qualitative and quantitative analysis of phytochemical

constituents and their metabolites using liquid chromatography–mass

spectrometry. Journal of Pharmaceutical and Biomedical Analysis, 2013.

72: p. 267-291.

2. Li, J.W.-H. and J.C. Vederas, Drug Discovery and Natural Products: End

of an Era or an Endless Frontier? Science, 2009. 325(5937): p. 161-165.

3. Vuorela, P., M. Leinonen, P. Saikku, P. Tammela, J.-P. Rauha, T.

Wennberg, and H. Vuorela, Natural products in the process of finding new

drug candidates. Current medicinal chemistry, 2004. 11(11): p. 1375-1389.

4. Verkouteren, J.R. and J.L. Staymates, Reliability of ion mobility

spectrometry for qualitative analysis of complex, multicomponent illicit

drug samples. Forensic Science International, 2011. 206(1–3): p. 190-196.

5. Truta, L., A.L. Castro, S. Tarelho, P. Costa, M.G.F. Sales, and H.M.

Teixeira, Antidepressants detection and quantification in whole blood

samples by GC–MS/MS, for forensic purposes. Journal of Pharmaceutical

and Biomedical Analysis, 2016. 128: p. 496-503.

6. Drugs, Poisons and Controlled Substances Act 1981, in 9719. 2017:

Australia. p. 398.

7. Schiff Jr, P.L., Opium and its alkaloids. American Journal of

Pharmaceutical Education, 2002. 66(2): p. 186.

8. Askitopoulou, H., I.A. Ramoutsaki, and E. Konsolaki, Archaeological

evidence on the use of opium in the Minoan world. International Congress

Series, 2002. 1242: p. 23-29.

9. Berridge, V.a.E., G., Opium and the people. 1 ed. 1981, Great Britain: Allen

lane/St Martin's press. 370.

10. Brownstein, M.J., A brief history of opiates, opioid peptides, and opioid

receptors. Proceedings of the National Academy of Sciences, 1993. 90(12):

p. 5391-5393.

164

11. Leonard, E., W. Runguphan, S. O'Connor, and K.J. Prather, Opportunities

in metabolic engineering to facilitate scalable alkaloid production. Nature

chemical biology, 2009. 5(5): p. 292-300.

12. Nyman, U.L.F., Selection for high thebaine/low morphine content (cpv.

Morph: The) in Papaver somniferum L. Hereditas, 1978. 89(1): p. 43-50.

13. Nyman, U. and O. Hall, Some varieties of Papaver somniferum L. with

changed morphinane alkaloid content. Hereditas, 1976. 84(1): p. 69-76.

14. Chaterjee, A., S. Shukla, P. Mishra, A. Rastogi, and S.P. Singh, Prospects

of in vitro production of thebaine in opium poppy (Papaver somniferum L.).

Industrial Crops and Products, 2010. 32(3): p. 668-670.

15. Chatterjee, A., S. Shukla, B. Kishore Mishra, A. Rastogi, and S. Prasad

Singh, Induction of variability through mutagenesis in opium poppy

(Papaver somniferum L.). Turkish Journal of Agriculture & Forestry, 2012.

36(1): p. 1-11.

16. Alagoz, Y., T. Gurkok, B. Zhang, and T. Unver, Manipulating the

Biosynthesis of Bioactive Compound Alkaloids for Next-Generation

Metabolic Engineering in Opium Poppy Using CRISPR-Cas 9 Genome

Editing Technology. Scientific Reports, 2016. 6: p. 30910.

17. Acevska, J., A. Dimitrovska, G. Stefkov, K. Brezovska, M. Karapandzova,

and S. Kulevanova, Development and validation of a reversed-Phase HPLC

method for determination of alkaloids from Papaver somniferum

L.(Papaveraceae). Journal of AOAC International, 2012. 95(2): p. 399-405.

18. Walwyn, W.M., K.A. Miotto, and C.J. Evans, Opioid pharmaceuticals and

addiction: The issues, and research directions seeking solutions. Drug and

Alcohol Dependence, 2010. 108(3): p. 156-165.

19. Yoshimatsu, K., F. Kiuchi, K. Shimomura, and Y. Makino, A rapid and

reliable solid-phase extraction method for high-performance liquid

chromatographic analysis of opium alkaloids from Papaver plants.

Chemical and pharmaceutical bulletin, 2005. 53(11): p. 1446-1450.

20. Gossop, M. and F. Keaney, Research Note—Prescribing Diamorphine for

Medical Conditions: A Very British Practice. Journal of Drug Issues, 2004.

34(2): p. 441-450.

165

21. Anthony John Fist, C.J.B., Wayne Lyle Gerlach,, PRODUCTION OF

THEBAINE AND ORIPAVINE. 2004, Tasmanian Alkaloids PTY. Ltd.,

Westbury (AU).

22. Gómez-Serranillos, M.P., O.M. Palomino, E. Carretero, and A. Villar,

Analytical study and analgesic activity of oripavine from Papaver

somniferum L. Phytotherapy Research, 1998. 12(5): p. 346-349.

23. Den Bosch, J.J.K.-V., C.A. Salemink, and I. Khan, Biological activity of

the alkaloids of Papaver bracteatum lindl. Journal of Ethnopharmacology,

1981. 3(1): p. 21-38.

24. Duke, J.A., Utilization of Papaver. Economic Botany, 1973. 27(4): p. 390-

400.

25. Barber, R.B. and H. Rapoport, Synthesis of thebaine and oripavine from

codeine and morphine. Journal of medicinal chemistry, 1975. 18(11): p.

1074-1077.

26. GlaxoSmithKline. GSK GlaxoSmithKline. 2011 7/5/2014]; Available from:

http://www.gsk.com.au/default.aspx.

27. Rosenbaum, C.D., S.P. Carreiro, and K.M. Babu, Here Today, Gone

Tomorrow…and Back Again? A Review of Herbal Marijuana Alternatives

(K2, Spice), Synthetic Cathinones (Bath Salts), Kratom, Salvia divinorum,

Methoxetamine, and Piperazines. Journal of Medical Toxicology, 2012.

8(1): p. 15-32.

28. Hudson, S. and J. Ramsey, The emergence and analysis of synthetic

cannabinoids. Drug Testing and Analysis, 2011. 3(7-8): p. 466-478.

29. Vardakou, I., C. Pistos, and C. Spiliopoulou, Spice drugs as a new trend:

Mode of action, identification and legislation. Toxicology Letters, 2010.

197(3): p. 157-162.

30. Glass, M. and J.K. Northup, Agonist Selective Regulation of G Proteins by

Cannabinoid CB<sub>1</sub> and CB<sub>2</sub> Receptors.

Molecular Pharmacology, 1999. 56(6): p. 1362-1369.

31. Huffman, J.W., D. Dai, B.R. Martin, and D.R. Compton, Design, Synthesis

and Pharmacology of Cannabimimetic Indoles. Bioorganic & Medicinal

Chemistry Letters, 1994. 4(4): p. 563-566.

32. Denooz, R., J.-C. Vanheugen, M. Frederich, P. de Tullio, and C. Charlier,

Identification and Structural Elucidation of Four Cannabimimetic

166

Compounds (RCS-4, AM-2201, JWH-203 and JWH-210) in Seized

Products. Journal of Analytical Toxicology, 2013. 37(2): p. 56-63.

33. Coppola, M., R. Mondola, F. Oliva, R.L. Picci, D. Ascheri, and F. Trivelli,

Chapter 63 - Treating the Phenomenon of New Psychoactive Substances:

Synthetic Cannabinoids and Synthetic Cathinones A2 - Preedy, Victor R, in

Neuropathology of Drug Addictions and Substance Misuse. 2016,

Academic Press: San Diego. p. 679-686.

34. Lindigkeit, R., A. Boehme, I. Eiserloh, M. Luebbecke, M. Wiggermann, L.

Ernst, and T. Beuerle, Spice: A never ending story? Forensic Science

International, 2009. 191(1–3): p. 58-63.

35. Coulter, C., M. Garnier, and C. Moore, Synthetic Cannabinoids in Oral

Fluid. Journal of Analytical Toxicology, 2011. 35(7): p. 424-430.

36. Park, Y., C. Lee, H. Lee, J. Pyo, J. Jo, J. Lee, H. Choi, S. Kim, R.S. Hong,

Y. Park, B.Y. Hwang, S. Choe, and J.H. Jung, Identification of a new

synthetic cannabinoid in a herbal mixture: 1-butyl-3-(2-

methoxybenzoyl)indole. Forensic Toxicology, 2013. 31(2): p. 187-196.

37. Ciolino, L.A., Quantitation of Synthetic Cannabinoids in Plant Materials

Using High Performance Liquid Chromatography with UV Detection

(Validated Method). Journal of Forensic Sciences, 2015. 60(5): p. 1171-

1181.

38. Ogata, J., N. Uchiyama, R. Kikura-Hanajiri, and Y. Goda, DNA sequence

analyses of blended herbal products including synthetic cannabinoids as

designer drugs. Forensic Science International, 2013. 227(1–3): p. 33-41.

39. Luke, A.D. and C.V. Joris, Cannabis Concerns: Increased Potency,

Availability and Synthetic Analogues. Current Drug Abuse Reviews, 2014.

7(2): p. 67-68.

40. Müller, H.H., J. Kornhuber, and W. Sperling, The behavioral profile of

spice and synthetic cannabinoids in humans. Brain Research Bulletin, 2016.

126, Part 1: p. 3-7.

41. Squires, M., Geelong 'fake drugs' crisis, in Geelong Advertiser. 2013.

42. Castaneto, M.S., D.A. Gorelick, N.A. Desrosiers, R.L. Hartman, S. Pirard,

and M.A. Huestis, Synthetic cannabinoids: Epidemiology,

pharmacodynamics, and clinical implications. Drug and Alcohol

Dependence, 2014. 144: p. 12-41.

167

43. Maxwell, P., Synthetic cannabinomimetics and ‘legal highs’. Journal of

Pharmacy Practice and Research, 2014. 44(4): p. 238-239.

44. Banister, S.D., J. Stuart, R.C. Kevin, A. Edington, M. Longworth, S.M.

Wilkinson, C. Beinat, A.S. Buchanan, D.E. Hibbs, M. Glass, M. Connor,

I.S. McGregor, and M. Kassiou, Effects of Bioisosteric Fluorine in

Synthetic Cannabinoid Designer Drugs JWH-018, AM-2201, UR-144,

XLR-11, PB-22, 5F-PB-22, APICA, and STS-135. ACS Chemical

Neuroscience, 2015. 6(8): p. 1445-1458.

45. Banister, S.D., M. Moir, J. Stuart, R.C. Kevin, K.E. Wood, M. Longworth,

S.M. Wilkinson, C. Beinat, A.S. Buchanan, M. Glass, M. Connor, I.S.

McGregor, and M. Kassiou, Pharmacology of Indole and Indazole

Synthetic Cannabinoid Designer Drugs AB-FUBINACA, ADB-

FUBINACA, AB-PINACA, ADB-PINACA, 5F-AB-PINACA, 5F-ADB-

PINACA, ADBICA, and 5F-ADBICA. ACS Chemical Neuroscience, 2015.

6(9): p. 1546-1559.

46. Harris, C.R. and A. Brown, Synthetic Cannabinoid Intoxication: A Case

Series and Review. The Journal of Emergency Medicine, 2013. 44(2): p.

360-366.

47. Hopkins, C.Y. and B.L. Gilchrist, A case of cannabinoid hyperemesis

syndrome caused by synthetic cannabinoids. The Journal of emergency

medicine, 2013. 45(4): p. 544-546.

48. Gerostamoulos, D., O.H. Drummer, and N.W. Woodford, Deaths linked to

synthetic cannabinoids. Forensic science, medicine, and pathology, 2015.

11(3): p. 478-478.

49. Adamowicz, P., Fatal intoxication with synthetic cannabinoid MDMB-

CHMICA. Forensic Science International, 2016. 261: p. e5-e10.

50. (AIHW), A.I.o.H.a.W. National Drug Strategy Household Survey

(NDSHS). 2013 2017 [cited 2017; Available from:

http://www.aihw.gov.au/alcohol-and-other-drugs/data-sources/ndshs-

2013/.

51. Adam R Winstock, Monica Barratt, Jason Ferris, and Larissa J. Maier.

Global Drug Survey. 2014 [cited 2017; Available from:

https://www.globaldrugsurvey.com/past-findings/.

168

52. Government, V.S. health.vic Synthetic cannabinoids. 2017 [cited 2017;

Available from: https://www2.health.vic.gov.au/public-health/drugs-and-

poisons/drugs-poisons-legislation/synthetic-cannabinoids.

53. Andrews, D., Synthetic drugs banned and tough new ice laws L.

Maksimovic, Editor. 2017, Premier of Victoria.

54. Znaleziona, J., P. Ginterová, J. Petr, P. Ondra, I. Válka, J. Ševčík, J.

Chrastina, and V. Maier, Determination and identification of synthetic

cannabinoids and their metabolites in different matrices by modern

analytical techniques – a review. Analytica Chimica Acta, 2015. 874: p. 11-

25.

55. Francis, P.S., J.L. Adcock, J.W. Costin, S.D. Purcell, F.M. Pfeffer, and

N.W. Barnett, Chemiluminescence detection of opium poppy (Papaver

somniferum) alkaloids. Journal of pharmaceutical and biomedical analysis,

2008. 48(3): p. 508-518.

56. Siddiqui, M.R., Z.A. AlOthman, and N. Rahman, Analytical techniques in

pharmaceutical analysis: A review. Arabian Journal of Chemistry, 2017.

10, Supplement 1: p. S1409-S1421.

57. Altria, K.D. and D. Elder, Overview of the status and applications of

capillary electrophoresis to the analysis of small molecules. Journal of

Chromatography A, 2004. 1023(1): p. 1-14.

58. Gika, H., G. Kaklamanos, P. Manesiotis, and G. Theodoridis,

Chromatography: High-Performance Liquid Chromatography, in

Encyclopedia of Food and Health. 2016, Academic Press: Oxford. p. 93-

99.

59. Houck, M.M. and J.A. Siegel, Fundamentals of forensic science. 2009:

Academic Press.

60. Hanai, T., HPLC: a practical guide. Vol. 6. 1999: Royal Society of

Chemistry.

61. Kavanagh, P., A. Grigoryev, S. Savchuk, I. Mikhura, and A. Formanovsky,

UR-144 in products sold via the Internet: Identification of related

compounds and characterization of pyrolysis products. Drug Testing and

Analysis, 2013. 5(8): p. 683-692.

62. Mostafavi, A., G. Abedi, A. Jamshidi, D. Afzali, and M. Talebi,

Development and validation of a HPLC method for the determination of

169

buprenorphine hydrochloride, naloxone hydrochloride and

noroxymorphone in a tablet formulation. Talanta, 2009. 77(4): p. 1415-

1419.

63. Costin, J.W., S.W. Lewis, S.D. Purcell, L.R. Waddell, P.S. Francis, and

N.W. Barnett, Rapid determination of Papaver somniferum alkaloids in

process streams using monolithic column high-performance liquid

chromatography with chemiluminescence detection. Analytica Chimica

Acta, 2007. 597(1): p. 19-23.

64. Stoll, D.R., X. Li, X. Wang, P.W. Carr, S.E.G. Porter, and S.C. Rutan, Fast,

comprehensive two-dimensional liquid chromatography. Journal of

Chromatography A, 2007. 1168(1–2): p. 3-43.

65. Stevenson, P.G., M. Mnatsakanyan, G. Guiochon, and R.A. Shalliker, Peak

picking and the assessment of separation performance in two-dimensional

high performance liquid chromatography. Analyst, 2010. 135(7): p. 1541-

1550.

66. Bassanese, D.N., B.J. Holland, X.A. Conlan, P.S. Francis, N.W. Barnett,

and P.G. Stevenson, Protocols for finding the most orthogonal dimensions

for two-dimensional high performance liquid chromatography. Talanta,

2015. 134: p. 402-408.

67. Fairchild, J.N., K. Horváth, and G. Guiochon, Approaches to

comprehensive multidimensional liquid chromatography systems. Journal

of Chromatography A, 2009. 1216(9): p. 1363-1371.

68. Stevenson, P.G., D.N. Bassanese, N.W. Barnett, and X.A. Conlan,

Improved 2D-HPLC of red wine by incorporating pre-process signal-

smoothing algorithms. Journal of Separation Science, 2013. 36(21-22): p.

3503-3510.

69. Furton, K.G., M. Kusano, and M.S. Macias, GAS CHROMATOGRAPHY

| Forensic Applications A2 - Wilson, Ian D, in Encyclopedia of Separation

Science. 2007, Academic Press: Oxford. p. 1-8.

70. Mohr, S., J.A. Weiß, J. Spreitz, and M.G. Schmid, Chiral separation of new

cathinone- and amphetamine-related designer drugs by gas

chromatography–mass spectrometry using trifluoroacetyl-l-prolyl chloride

as chiral derivatization reagent. Journal of Chromatography A, 2012. 1269:

p. 352-359.

170

71. Singh, S., T. Handa, M. Narayanam, A. Sahu, M. Junwal, and R.P. Shah, A

critical review on the use of modern sophisticated hyphenated tools in the

characterization of impurities and degradation products. Journal of

Pharmaceutical and Biomedical Analysis, 2012. 69: p. 148-173.

72. Elder, D.P., D. Snodin, and A. Teasdale, Control and analysis of hydrazine,

hydrazides and hydrazones—Genotoxic impurities in active pharmaceutical

ingredients (APIs) and drug products. Journal of Pharmaceutical and

Biomedical Analysis, 2011. 54(5): p. 900-910.

73. Płonka, M., S. Walorczyk, and M. Miszczyk, Chromatographic methods for

the determination of active substances and characterization of their

impurities in pesticide formulations. TrAC Trends in Analytical Chemistry,

2016. 85, Part B: p. 67-80.

74. Eckre, D., L. Fadness, G. Rankin, and C. Bridge, Thermal Degradation of

Synthetic Cannabinoids Containing a Cyclopropyl Group.

75. Denis, C.M. and N.E. Baryla, Determination of piperazine in

pharmaceutical drug substances using capillary electrophoresis with

indirect UV detection. Journal of Chromatography A, 2006. 1110(1–2): p.

268-271.

76. Cruces-Blanco, C., L. Gámiz–Gracia, and A.M. García-Campaña,

Applications of capillary electrophoresis in forensic analytical chemistry.

TrAC Trends in Analytical Chemistry, 2007. 26(3): p. 215-226.

77. Hindson, B.J., P.S. Francis, S.D. Purcell, and N.W. Barnett, Determination

of opiate alkaloids in process liquors using capillary electrophoresis.

Journal of Pharmaceutical and Biomedical Analysis, 2007. 43(3): p. 1164-

1168.

78. Gerardi, R.D., N.W. Barnett, and S.W. Lewis, Analytical applications of

tris(2,2′-bipyridyl)ruthenium(III) as a chemiluminescent reagent. Analytica

Chimica Acta, 1999. 378(1–3): p. 1-41.

79. Adcock, J.L., C.J. Barrow, N.W. Barnett, X.A. Conlan, C.F. Hogan, and

P.S. Francis, Chemiluminescence and electrochemiluminescence detection

of controlled drugs. Drug Testing and Analysis, 2011. 3(3): p. 145-160.

80. Percy, D.W., J.L. Adcock, X.A. Conlan, N.W. Barnett, M.E. Gange, L.K.

Noonan, L.C. Henderson, and P.S. Francis, Determination of Citrus

171

aurantium protoalkaloids using HPLC with acidic potassium permanganate

chemiluminescence detection. Talanta, 2010. 80(5): p. 2191-2195.

81. McDermott, G.P., X.A. Conlan, L.K. Noonan, J.W. Costin, M.

Mnatsakanyan, R.A. Shalliker, N.W. Barnett, and P.S. Francis, Screening

for antioxidants in complex matrices using high performance liquid

chromatography with acidic potassium permanganate chemiluminescence

detection. Analytica Chimica Acta, 2011. 684(1–2): p. 134-141.

82. Khataee, A., R. Lotfi, A. Hasanzadeh, M. Iranifam, and S.W. Joo, Flow-

injection chemiluminescence analysis for sensitive determination of

atenolol using cadmium sulfide quantum dots. Spectrochimica Acta Part A:

Molecular and Biomolecular Spectroscopy, 2016. 157: p. 88-95.

83. Nalewajko-Sieliwoniuk, E., J. Malejko, M. Święczkowska, and A.

Kowalewska, A study on the selection of chemiluminescence system for the

flow injection determination of the total polyphenol index of plant-derived

foods. Food Chemistry, 2015. 176: p. 175-183.

84. Francis, P.S., J.W. Costin, X.A. Conlan, S.A. Bellomarino, J.A. Barnett,

and N.W. Barnett, A rapid antioxidant assay based on acidic potassium

permanganate chemiluminescence. Food Chemistry, 2010. 122(3): p. 926-

929.

85. Lenehan, C.E., N.W. Barnett, S.W. Lewis, and K.M. Essery, Preliminary

evaluation of dual acidic potassium permanganate and tris (2, 2′-bipyridyl)

ruthenium (II) chemiluminescence detection for the HPLC determination of

Papaver somniferum alkaloids. Australian journal of chemistry, 2004.

57(10): p. 1001-1004.

86. Brown, A.J., C.E. Lenehan, P.S. Francis, D.E. Dunstan, and N.W. Barnett,

Soluble manganese(IV) as a chemiluminescence reagent for the

determination of opiate alkaloids, indoles and analytes of forensic interest.

Talanta, 2007. 71(5): p. 1951-1957.

87. Li, Y.H. and J.R. Lü, Determination of three fluoroquinolone drugs by flow

injection chemiluminescence. Fenxi Huaxue/ Chinese Journal of Analytical

Chemistry, 2007. 35(5): p. 743-746.

88. Holland, B.J., P.S. Francis, B. Li, T. Tsuzuki, J.L. Adcock, N.W. Barnett,

and X.A. Conlan, Chemiluminescence detection of cannabinoids and

172

related compounds with acidic potassium permanganate. Drug Testing and

Analysis, 2012. 4(7-8): p. 675-679.

89. Camenzuli, M., J.M. Terry, R.A. Shalliker, X.A. Conlan, N.W. Barnett, and

P.S. Francis, Parallel segmented outlet flow high performance liquid

chromatography with multiplexed detection. Analytica Chimica Acta, 2013.

803: p. 154-159.

90. Theakstone, A.G., Z.M. Smith, J.M. Terry, N.W. Barnett, and P.S. Francis,

Chemiluminescence detection of MDMA in street drug samples using

tris(2,2′-bipyridine)ruthenium(III). Drug Testing and Analysis, 2015. 7(5):

p. 428-432.

91. McDermott, G.P., P. Jones, N.W. Barnett, D.N. Donaldson, and P.S.

Francis, Stable tris (2, 2′-bipyridine) ruthenium (III) for chemiluminescence

detection. Analytical chemistry, 2011. 83(13): p. 5453-5457.

92. Gorman, B.A., P.S. Francis, and N.W. Barnett, Tris(2,2'-

bipyridyl)ruthenium(II) chemiluminescence. Analyst, 2006. 131(5): p. 616-

639.

93. Waite, R.J., G.J. Barbante, N.W. Barnett, E.M. Zammit, and P.S. Francis,

Chemiluminescence detection of piperazine designer drugs and related

compounds using tris(2,2′-bipyridine)ruthenium(III). Talanta, 2013. 116: p.

1067-1072.

94. Adcock, J.L., P.S. Francis, and N.W. Barnett, Acidic potassium

permanganate as a chemiluminescence reagent—A review. Analytica

chimica acta, 2007. 601(1): p. 36-67.

95. Hindson, B.J. and N.W. Barnett, Analytical applications of acidic potassium

permanganate as a chemiluminescence reagent. Analytica chimica acta,

2001. 445(1): p. 1-19.

96. Adcock, J.L., N.W. Barnett, C.J. Barrow, and P.S. Francis, Advances in the

use of acidic potassium permanganate as a chemiluminescence reagent: a

review. Analytica chimica acta, 2014. 807: p. 9-28.

97. Adcock, J.L., Z.M. Smith, N.W. Barnett, G.J. Barbante, E.H. Doeven, and

P.S. Francis, A review of recent advances in chemiluminescence detection

using nano-colloidal manganese (IV). Analytica chimica acta, 2014. 848:

p. 1-9.

173

98. Brown, A.J., P.S. Francis, J.L. Adcock, K.F. Lim, and N.W. Barnett,

Manganese(III) and manganese(IV) as chemiluminescence reagents: A

review. Analytica Chimica Acta, 2008. 624(2): p. 175-183.

99. Terry, J.M., Z.M. Smith, G.P. McDermott, R.J. Waite, N.W. Barnett, L.C.

Henderson, J.M. Altimari, and P.S. Francis, Chemiluminescence detection

of amino acids and related compounds using acidic potassium

permanganate, manganese(IV) or tris(2,2′-bipyridine)ruthenium(III).

Talanta, 2012. 99: p. 1051-1056.

100. Nalewajko-Sieliwoniuk, E., I. Tarasewicz, and A. Kojło, Flow injection

chemiluminescence determination of the total phenolics levels in plant-

derived beverages using soluble manganese(IV). Analytica Chimica Acta,

2010. 668(1): p. 19-25.

101. Malejko, J., E. Nalewajko-Sieliwoniuk, J. Nazaruk, J. Siniło, and A. Kojło,

Determination of the total polyphenolic content in Cirsium palustre (L.)

leaves extracts with manganese(IV) chemiluminescence detection. Food

Chemistry, 2014. 152: p. 155-161.

102. Richter, M.M., Electrochemiluminescence (ECL). Chemical Reviews,

2004. 104(6): p. 3003-3036.

103. Tokel, N.E. and A.J. Bard, Electrogenerated chemiluminescence. IX.

Electrochemistry and emission from systems containing tris (2, 2′-

bipyridine) ruthenium (II) dichloride. J. Am. Chem. Soc, 1972. 94(8): p.

2862-2863.

104. Kirschbaum, S.E.K. and A.J. Baeumner, A review of

electrochemiluminescence (ECL) in and for microfluidic analytical devices.

Analytical and Bioanalytical Chemistry, 2015. 407(14): p. 3911-3926.

105. Miao, W., Electrogenerated Chemiluminescence and Its Biorelated

Applications. Chemical Reviews, 2008. 108(7): p. 2506-2553.

106. Hosseini, M., M.R.K. Pur, P. Norouzi, M.R. Moghaddam, and M.R.

Ganjali, An enhanced electrochemiluminescence sensor modified with a

Ru(bpy)32 +/Yb2O3 nanoparticle/nafion composite for the analysis of

methadone samples. Materials Science and Engineering: C, 2017. 76: p.

483-489.

107. McMurry, J., Organic chemistry. 2008: Belmont, CA : Thomson-

Brooks/Cole, c2008.7th ed.

174

108. Blackman, A., S.E. Bottle, S. Schmid, M. Mocerino, and U. Wille,

CHEMISTRY 3E. 2015, Melbourne, AUSTRALIA: Wiley.

109. Auwärter, V., S. Dresen, W. Weinmann, M. Müller, M. Pütz, and N.

Ferreirós, ‘Spice’ and other herbal blends: harmless incense or cannabinoid

designer drugs? Journal of Mass Spectrometry, 2009. 44(5): p. 832-837.

110. Ernst, L., K. Krüger, R. Lindigkeit, H.-M. Schiebel, and T. Beuerle,

Synthetic cannabinoids in “spice-like” herbal blends: First appearance of

JWH-307 and recurrence of JWH-018 on the German market. Forensic

Science International, 2012. 222(1–3): p. 216-222.

111. Uchiyama, N., M. Kawamura, R. Kikura-Hanajiri, and Y. Goda,

Identification of two new-type synthetic cannabinoids, N-(1-adamantyl)-1-

pentyl-1H-indole-3-carboxamide (APICA) and N-(1-adamantyl)-1-pentyl-

1H-indazole-3-carboxamide (APINACA), and detection of five synthetic

cannabinoids, AM-1220, AM-2233, AM-1241, CB-13 (CRA-13), and AM-

1248, as designer drugs in illegal products. Forensic Toxicology, 2012.

30(2): p. 114-125.

112. Uchiyama, N., R. Kikura-Hanajiri, J. Ogata, and Y. Goda, Chemical

analysis of synthetic cannabinoids as designer drugs in herbal products.

Forensic Science International, 2010. 198(1–3): p. 31-38.

113. Ammann, J., J.M. McLaren, D. Gerostamoulos, and J. Beyer, Detection and

Quantification of New Designer Drugs in Human Blood: Part 1 – Synthetic

Cannabinoids. Journal of Analytical Toxicology, 2012. 36(6): p. 372-380.

114. Dresen, S., S. Kneisel, W. Weinmann, R. Zimmermann, and V. Auwärter,

Development and validation of a liquid chromatography–tandem mass

spectrometry method for the quantitation of synthetic cannabinoids of the

aminoalkylindole type and methanandamide in serum and its application to

forensic samples. Journal of Mass Spectrometry, 2011. 46(2): p. 163-171.

115. Dziadosz, M., J.-P. Weller, M. Klintschar, and J. Teske, Scheduled multiple

reaction monitoring algorithm as a way to analyse new designer drugs

combined with synthetic cannabinoids in human serum with liquid

chromatography–tandem mass spectrometry. Journal of Chromatography

B, 2013. 929: p. 84-89.

116. Hutter, M., S. Kneisel, V. Auwärter, and M.A. Neukamm, Determination

of 22 synthetic cannabinoids in human hair by liquid chromatography–

175

tandem mass spectrometry. Journal of Chromatography B, 2012. 903: p.

95-101.

117. Salomone, A., E. Gerace, F. D'Urso, D. Di Corcia, and M. Vincenti,

Simultaneous analysis of several synthetic cannabinoids, THC, CBD and

CBN, in hair by ultra-high performance liquid chromatography tandem

mass spectrometry. Method validation and application to real samples.

Journal of Mass Spectrometry, 2012. 47(5): p. 604-610.

118. Kavanagh, P., A. Grigoryev, A. Melnik, S. Savchuk, A. Simonov, and V.

Rozhanets, Detection and tentative identification of urinary phase I

metabolites of phenylacetylindole cannabimimetics JWH-203 and JWH-

251, by GC–MS and LC–MS/MS. Journal of Chromatography B, 2013.

934: p. 102-108.

119. Grigoryev, A., A. Melnik, S. Savchuk, A. Simonov, and V. Rozhanets, Gas

and liquid chromatography–mass spectrometry studies on the metabolism

of the synthetic phenylacetylindole cannabimimetic JWH-250, the

psychoactive component of smoking mixtures. Journal of Chromatography

B, 2011. 879(25): p. 2519-2526.

120. Antelo-Domínguez, Á., J. Ángel Cocho, M. Jesús Tabernero, A. María

Bermejo, P. Bermejo-Barrera, and A. Moreda-Piñeiro, Simultaneous

determination of cocaine and opiates in dried blood spots by electrospray

ionization tandem mass spectrometry. Talanta, 2013. 117: p. 235-241.

121. Rana, S., R.K. Garg, and A. Singla, Rapid analysis of urinary opiates using

fast gas chromatography–mass spectrometry and hydrogen as a carrier gas.

Egyptian Journal of Forensic Sciences, 2014. 4(3): p. 100-107.

122. Imbert, L., S. Dulaurent, M. Mercerolle, J. Morichon, G. Lachâtre, and J.M.

Gaulier, Development and validation of a single LC–MS/MS assay

following SPE for simultaneous hair analysis of amphetamines, opiates,

cocaine and metabolites. Forensic Science International, 2014. 234: p. 132-

138.

123. Blair, B., A. Nikolaus, C. Hedman, R. Klaper, and T. Grundl, Evaluating

the degradation, sorption, and negative mass balances of pharmaceuticals

and personal care products during wastewater treatment. Chemosphere,

2015. 134: p. 395-401.

176

124. Archer, E., B. Petrie, B. Kasprzyk-Hordern, and G.M. Wolfaardt, The fate

of pharmaceuticals and personal care products (PPCPs), endocrine

disrupting contaminants (EDCs), metabolites and illicit drugs in a WWTW

and environmental waters. Chemosphere, 2017. 174: p. 437-446.

125. Boleda, M.R., M.T. Galceran, and F. Ventura, Trace determination of

cannabinoids and opiates in wastewater and surface waters by ultra-

performance liquid chromatography–tandem mass spectrometry. Journal of

Chromatography A, 2007. 1175(1): p. 38-48.

126. Ziegler, J., P.J. Facchini, R. Geißler, J. Schmidt, C. Ammer, R. Kramell, S.

Voigtländer, A. Gesell, S. Pienkny, and W. Brandt, Evolution of morphine

biosynthesis in opium poppy. Phytochemistry, 2009. 70(15–16): p. 1696-

1707.

127. Furuya, T., A. Ikuta, and K. Syōno, Alkaloids from callus tissue of Papaver

somniferum. Phytochemistry, 1972. 11(10): p. 3041-3044.

128. Stranska, I., M. Skalicky, J. Novak, E. Matyasova, and V. Hejnak, Analysis

of selected poppy (Papaver somniferum L.) cultivars: Pharmaceutically

important alkaloids. Industrial Crops and Products, 2013. 41: p. 120-126.

129. Macomber, R.S., A complete introduction to modern NMR spectroscopy.

1998: Wiley New York.

130. Rossini, A.J., A. Zagdoun, M. Lelli, J. Canivet, S. Aguado, O. Ouari, P.

Tordo, M. Rosay, W.E. Maas, C. Copéret, D. Farrusseng, L. Emsley, and

A. Lesage, Dynamic Nuclear Polarization Enhanced Solid-State NMR

Spectroscopy of Functionalized Metal–Organic Frameworks. Angewandte

Chemie, 2012. 124(1): p. 127-131.

131. Wang, W., J. Jiao, Y. Jiang, S.S. Ray, and M. Hunger, Formation and

Decomposition of Surface Ethoxy Species on Acidic Zeolite Y.

ChemPhysChem, 2005. 6(8): p. 1467-1469.

132. Dybowski*, C. and S. Bai, Solid-state NMR spectroscopy. Analytical

chemistry, 2008. 80(12): p. 4295-4300.

133. Uchiyama, N., M. Kawamura, R. Kikura-Hanajiri, and Y. Goda,

Identification and quantitation of two cannabimimetic phenylacetylindoles

JWH-251 and JWH-250, and four cannabimimetic naphthoylindoles JWH-

081, JWH-015, JWH-200, and JWH-073 as designer drugs in illegal

products. Forensic Toxicology, 2011. 29(1): p. 25-37.

177

134. Simolka, K., R. Lindigkeit, H.-M. Schiebel, U. Papke, L. Ernst, and T.

Beuerle, Analysis of synthetic cannabinoids in “spice-like” herbal highs:

snapshot of the German market in summer 2011. Analytical and

Bioanalytical Chemistry, 2012. 404(1): p. 157-171.

135. Moosmann, B., S. Kneisel, U. Girreser, V. Brecht, F. Westphal, and V.

Auwärter, Separation and structural characterization of the synthetic

cannabinoids JWH-412 and 1-[(5-fluoropentyl)-1H-indol-3yl]-(4-

methylnaphthalen-1-yl)methanone using GC–MS, NMR analysis and a

flash chromatography system. Forensic Science International, 2012.

220(1–3): p. e17-e22.

136. Fowler, F., B. Voyer, M. Marino, J. Finzel, M. Veltri, N.M. Wachter, and

L. Huang, Rapid screening and quantification of synthetic cannabinoids in

herbal products with NMR spectroscopic methods. Analytical Methods,

2015. 7(18): p. 7907-7916.

137. Wick, A., M. Wagner, and T.A. Ternes, Elucidation of the Transformation

Pathway of the Opium Alkaloid Codeine in Biological Wastewater

Treatment. Environmental Science & Technology, 2011. 45(8): p. 3374-

3385.

138. Hagel, J.M. and P.J. Facchini, Plant metabolomics: analytical platforms and

integration with functional genomics. Phytochemistry Reviews, 2008. 7(3):

p. 479-497.

139. Plumb, R.S., C.L. Stumpf, M.V. Gorenstein, J.M. Castro-Perez, G.J. Dear,

M. Anthony, B.C. Sweatman, S.C. Connor, and J.N. Haselden,

Metabonomics: the use of electrospray mass spectrometry coupled to

reversed-phase liquid chromatography shows potential for the screening of

rat urine in drug development. Rapid Communications in Mass

Spectrometry, 2002. 16(20): p. 1991-1996.

140. Zhao, Y.Q., X. Li, X. Liu, P.F. Dong, L.L. Wang, and X.Q. Wang, Research

on water quality analysis model with PCA method and UV absorption

spectra. Guang Pu Xue Yu Guang Pu Fen Xi/Spectroscopy and Spectral

Analysis, 2016. 36(11): p. 3592-3596.

141. Bicchi, C.P., O.M. Panero, G.M. Pellegrino, and A.C. Vanni,

Characterization of Roasted Coffee and Coffee Beverages by Solid Phase

178

Microextraction−Gas Chromatography and Principal Component Analysis.

Journal of Agricultural and Food Chemistry, 1997. 45(12): p. 4680-4686.

142. Cevallos-Cevallos, J.M., J.I. Reyes-De-Corcuera, E. Etxeberria, M.D.

Danyluk, and G.E. Rodrick, Metabolomic analysis in food science: a

review. Trends in Food Science & Technology, 2009. 20(11–12): p. 557-

566.

143. Kallithraka, S., I.S. Arvanitoyannis, P. Kefalas, A. El-Zajouli, E. Soufleros,

and E. Psarra, Instrumental and sensory analysis of Greek wines;

implementation of principal component analysis (PCA) for classification

according to geographical origin. Food Chemistry, 2001. 73(4): p. 501-514.

144. Rodrı́guez-Delgado, M.-Á., G. González-Hernández, J.-E.a. Conde-

González, and J.-P. Pérez-Trujillo, Principal component analysis of the

polyphenol content in young red wines. Food Chemistry, 2002. 78(4): p.

523-532.

145. Snyder, L.R., J.J. Kirkland, and J.W. Dolan, Introduction to modern liquid

chromatography. 2011: John Wiley & Sons.

146. Stachniuk, J., P. Kubalczyk, P. Furmaniak, and R. Głowacki, A versatile

method for analysis of saliva, plasma and urine for total thiols using HPLC

with UV detection. Talanta, 2016. 155: p. 70-77.

147. Li, G., S. Liu, Z. Sun, L. Xia, G. Chen, and J. You, A simple and sensitive

HPLC method based on pre-column fluorescence labelling for multiple

classes of plant growth regulator determination in food samples. Food

Chemistry, 2015. 170: p. 123-130.

148. Ma, J., Z. Yao, L. Hou, W. Lu, Q. Yang, J. Li, and L. Chen, Metal organic

frameworks (MOFs) for magnetic solid-phase extraction of

pyrazole/pyrrole pesticides in environmental water samples followed by

HPLC-DAD determination. Talanta, 2016. 161: p. 686-692.

149. Guiochon, G., N. Marchetti, K. Mriziq, and R.A. Shalliker,

Implementations of two-dimensional liquid chromatography. Journal of

Chromatography A, 2008. 1189(1–2): p. 109-168.

150. Murahashi, T., F. Tsuruga, and S. Sasaki, An automatic method for the

determination of carcinogenic 1-nitropyrene in extracts from automobile

exhaust particulate matter. Analyst, 2003. 128(11): p. 1346-1351.

179

151. Regalado, E.L., J.A. Schariter, and C.J. Welch, Investigation of two-

dimensional high performance liquid chromatography approaches for

reversed phase resolution of warfarin and hydroxywarfarin isomers.

Journal of Chromatography A, 2014. 1363: p. 200-206.

152. Król-Kogus, B., D. Głód, M. Krauze-Baranowska, and I. Matławska,

Application of one- and two-dimensional high-performance liquid

chromatography methodologies for the analysis of C-glycosylflavones from

fenugreek seeds. Journal of Chromatography A, 2014. 1367: p. 48-56.

153. Andrighetto, L.M., P.G. Stevenson, J.R. Pearson, L.C. Henderson, and X.A.

Conlan, DryLab® optimised two-dimensional high performance liquid

chromatography for differentiation of ephedrine and pseudoephedrine

based methamphetamine samples. Forensic Science International, 2014.

244: p. 302-305.

154. Köhne, A.P., U. Dornberger, and T. Welsch, Two-dimensional high

performance liquid chromatography for the separation of complex mixtures

of explosives and their by-products. Chromatographia, 1998. 48(1-2): p. 9-

16.

155. Gilar, M., J. Fridrich, M.R. Schure, and A. Jaworski, Comparison of

Orthogonality Estimation Methods for the Two-Dimensional Separations of

Peptides. Analytical Chemistry, 2012. 84(20): p. 8722-8732.

156. Watson, N.E., J.M. Davis, and R.E. Synovec, Observations on

“Orthogonality” in Comprehensive Two-Dimensional Separations.

Analytical Chemistry, 2007. 79(20): p. 7924-7927.

157. Li, X., D.R. Stoll, and P.W. Carr, Equation for Peak Capacity Estimation in

Two-Dimensional Liquid Chromatography. Analytical Chemistry, 2009.

81(2): p. 845-850.

158. Gilar, M., P. Olivova, A.E. Daly, and J.C. Gebler, Orthogonality of

Separation in Two-Dimensional Liquid Chromatography. Analytical

Chemistry, 2005. 77(19): p. 6426-6434.

159. Lee, M.J., I.-M. Chung, H. Kim, and M.Y. Jung, High resolution LC–ESI-

TOF-mass spectrometry method for fast separation, identification, and

quantification of 12 isoflavones in soybeans and soybean products. Food

Chemistry, 2015. 176: p. 254-262.

180

160. McLafferty, F.W. and F. Tureček, Interpretation of mass spectra. 1993: Mill

Valley, Calif. : University Science Books, [1993] Fourth edition.

161. Saranjit S, T.H., Mallikarjun N, Archana S, Mahendra J, Ravi S, A critical

review on the use of modern sophisticated hyphenated tools in the

characterization of impurities and degradation products. Journal of

Pharmaceutical and Biomedical Analysis, 2012. 69: p. 26.

162. Kohlbacher, O., K. Reinert, C. Gröpl, E. Lange, N. Pfeifer, O. Schulz-

Trieglaff, and M. Sturm, TOPP—the OpenMS proteomics pipeline.

Bioinformatics, 2007. 23(2): p. e191-e197.

163. Sturm, M., A. Bertsch, C. Gropl, A. Hildebrandt, R. Hussong, E. Lange, N.

Pfeifer, O. Schulz-Trieglaff, A. Zerck, K. Reinert, and O. Kohlbacher,

OpenMS - An open-source software framework for mass spectrometry.

BMC Bioinformatics, 2008. 9(1): p. 163.

164. Dioumaeva, I., S.-B. Choi, B. Yong, D. Jones, and R. Arora, Understanding

Orthogonality in Reversed-Phase Liquid Chromatography for Easier

Column Selection and Method Development. Application Note, Agilent

Technologies,

http://www. chem. agilent. com/Library/applications/SI-02425. pdf.

165. Mnatsakanyan, M., P.G. Stevenson, X.A. Conlan, P.S. Francis, T.A.

Goodie, G.P. McDermott, N.W. Barnett, and R.A. Shalliker, The analysis

of café espresso using two-dimensional reversed phase–reversed phase high

performance liquid chromatography with UV-absorbance and

chemiluminescence detection. Talanta, 2010. 82(4): p. 1358-1363.

166. Gritti, F. and G. Guiochon, Adsorption mechanism in reversed-phase liquid

chromatography: Effect of the surface coverage of a monomeric C18-silica

stationary phase. Journal of Chromatography A, 2006. 1115(1–2): p. 142-

163.

167. Gray, M., G.R. Dennis, P. Wormell, R. Andrew Shalliker, and P. Slonecker,

Two dimensional reversed-phase–reversed-phase separations: Isomeric

separations incorporating C18 and carbon clad zirconia stationary phases.

Journal of Chromatography A, 2002. 975(2): p. 285-297.

181

168. Catchpoole, H.J., R. Andrew Shalliker, G.R. Dennis, and G. Guiochon,

Visualising the onset of viscous fingering in chromatography columns.

Journal of Chromatography A, 2006. 1117(2): p. 137-145.

169. Shalliker, R.A., H.J. Catchpoole, G.R. Dennis, and G. Guiochon,

Visualising viscous fingering in chromatography columns: High viscosity

solute plug. Journal of Chromatography A, 2007. 1142(1): p. 48-55.

170. Mayfield, K.J., R.A. Shalliker, H.J. Catchpoole, A.P. Sweeney, V. Wong,

and G. Guiochon, Viscous fingering induced flow instability in

multidimensional liquid chromatography. Journal of Chromatography A,

2005. 1080(2): p. 124-131.

171. Stevenson, P.G., D.N. Bassanese, X.A. Conlan, and N.W. Barnett,

Improving peak shapes with counter gradients in two-dimensional high

performance liquid chromatography. Journal of Chromatography A, 2014.

1337: p. 147-154.

172. Samuelsson, J., R.A. Shalliker, and T. Fornstedt, Viscosity contrast effects

in analytical scale chromatography - Evidence and impact. Microchemical

Journal, 2017. 130: p. 102-107.

173. Ayre, A., D. Varpe, R. Nayak, and N. Vasa, Impurity profiling of

pharmaceuticals. Advance Research in Pharmaceuticals and Biologicals,

2011. 1: p. 76-90.

174. Purcell, D.S.D., Personal correspondance. 2017.

175. Yin, S., X. Zhu, and H.R. Karimi, Quality Evaluation Based on Multivariate

Statistical Methods. Mathematical Problems in Engineering, 2013. 2013: p.

10.

176. Custers, D., B. Krakowska, J.O. De Beer, P. Courselle, M. Daszykowski, S.

Apers, and E. Deconinck, Chromatographic impurity fingerprinting of

genuine and counterfeit Cialis® as a means to compare the discriminating

ability of PDA and MS detection. Talanta, 2016. 146: p. 540-548.

177. Hopfer, H., J. Nelson, T.S. Collins, H. Heymann, and S.E. Ebeler, The

combined impact of vineyard origin and processing winery on the elemental

profile of red wines. Food Chemistry, 2015. 172: p. 486-496.

178. Pap, T.L. and Z. Pápai, Application of a new mathematical function for

describing chromatographic peaks. Journal of Chromatography A, 2001.

930(1–2): p. 53-60.

182

179. Nielsen, N.-P.V., J.M. Carstensen, and J. Smedsgaard, Aligning of single

and multiple wavelength chromatographic profiles for chemometric data

analysis using correlation optimised warping. Journal of Chromatography

A, 1998. 805(1–2): p. 17-35.

180. Stevenson, P.G. and G. Guiochon, Cumulative area of peaks in a

multidimensional high performance liquid chromatogram. Journal of

Chromatography A, 2013. 1308: p. 79-85.

181. Ahuja, S., Assuring quality of drugs by monitoring impurities. Advanced

Drug Delivery Reviews, 2007. 59(1): p. 3-11.

182. Guideline, I.H.T. Impurities in new drug substances Q3A (R2). in

Proceedings of the International Conference on Harmonization of

Technical Requirements for Registration of Pharmaceuticals for Human

Use, Geneva, Switzerland. 2006.

183. Basak, A.K., A.S. Raw, A.H. Al Hakim, S. Furness, N.I. Samaan, D.S. Gill,

H.B. Patel, R.F. Powers, and L. Yu, Pharmaceutical impurities: Regulatory

perspective for Abbreviated New Drug Applications. Advanced Drug

Delivery Reviews, 2007. 59(1): p. 64-72.

184. van Beers, M.M.C. and M. Bardor, Minimizing immunogenicity of

biopharmaceuticals by controlling critical quality attributes of proteins.

Biotechnology Journal, 2012. 7(12): p. 1473-1484.

185. Rice, K.C. and A.H. Newman, Total synthesis of northebaine, normophine,

noroxymorphone enantiomers and derivatives via N-Nor intermediates.

1997, Google Patents.

186. Andrighetto, L.M., N.K. Burns, P.G. Stevenson, J.R. Pearson, L.C.

Henderson, C.J. Bowen, and X.A. Conlan, In-silico optimisation of two-

dimensional high performance liquid chromatography for the determination

of Australian methamphetamine seizure samples. Forensic Science

International, 2016. 266: p. 511-516.

187. Frick, S., J.A. Chitty, R. Kramell, J. Schmidt, R.S. Allen, P.J. Larkin, and

T.M. Kutchan, Transformation of opium poppy (Papaver somniferumL.)

with antisense berberine bridge enzyme gene (anti-bbe) via somatic

embryogenesis results in an altered ratio of alkaloids in latex but not in

roots. Transgenic research, 2004. 13(6): p. 607-613.

183

188. Kelly, S.S., P.M. Glynn, S.J. Madden, and D.H. Grayson, Impurities in a

morphine sulfate drug product identified as 5-(hydroxymethyl)-2-furfural,

10-hydroxymorphine and 10-oxomorphine. Journal of Pharmaceutical

Sciences, 2003. 92(3): p. 485-493.

189. Smith, K.H., T. Bowser, and G.W. Stevens, Performance and Scale-up of

Karr Reciprocating Plate Extraction Columns. Industrial & Engineering

Chemistry Research, 2008. 47(21): p. 8368-8375.

190. Stella, A., K.H. Mensforth, T. Bowser, G.W. Stevens, and H.R.C. Pratt,

Mass Transfer Performance in Karr Reciprocating Plate Extraction

Columns. Industrial & Engineering Chemistry Research, 2008. 47(11): p.

3996-4007.

191. Wiegand, T.J. and F. Lovecchio, Designer drugs: Focus on cathinones (bath

salts) and synthetic cannabinoids (K2 or spice). Emergency Medicine

Reports, 2012. 33(26): p. 305-315.

192. Arnold, C., The new danger of synthetic drugs. The Lancet. 382(9886): p.

15-16.

193. Wilkinson, S.M., S.D. Banister, and M. Kassiou, Bioisosteric Fluorine in

the Clandestine Design of Synthetic Cannabinoids. Australian Journal of

Chemistry, 2015. 68(1): p. 4-8.

194. Merola, G., Z. Aturki, G. D’Orazio, R. Gottardo, T. Macchia, F. Tagliaro,

and S. Fanali, Analysis of synthetic cannabinoids in herbal blends by means

of nano-liquid chromatography. Journal of Pharmaceutical and Biomedical

Analysis, 2012. 71: p. 45-53.

195. Uchiyama, N., M. Kawamura, R. Kikura-Hanajiri, and Y. Goda, URB-754:

A new class of designer drug and 12 synthetic cannabinoids detected in

illegal products. Forensic Science International, 2013. 227(1–3): p. 21-32.

196. Burns, N., Synthetic Cannabinoids: Extraction, identification and

purification, in School of Life and Environmental Sciences. 2013, Deakin

University.

197. Rasanen, I., M. Kyber, I. Szilvay, J. Rintatalo, and I. Ojanperä,

Straightforward single-calibrant quantification of seized designer drugs by

liquid chromatography–chemiluminescence nitrogen detection. Forensic

Science International, 2014. 237: p. 119-125.

184

198. Żwir-Ferenc, A. and M. Biziuk, Solid Phase Extraction Technique - Trends,

Opportunities and Applications. Polish Journal of Environmental Studies,

2006. 15(5): p. 677-690.

199. Grigoryev, A., S. Savchuk, A. Melnik, N.j. Moskaleva, J. Dzhurko, M.

Ershov, A. Nosyrev, A. Vedenin, B. Izotov, I. Zabirova, and V. Rozhanets,

Chromatography–mass spectrometry studies on the metabolism of synthetic

cannabinoids JWH-018 and JWH-073, psychoactive components of

smoking mixtures. Journal of Chromatography B, 2011. 879(15–16): p.

1126-1136.

200. Kavanagh, P., A. Grigoryev, A. Melnik, and A. Simonov, The identification

of the urinary metabolites of 3-(4-methoxybenzoyl)-1-pentylindole (RCS-

4), a novel cannabimimetic, by gas chromatography–mass spectrometry.

Journal of analytical toxicology, 2012. 36(5): p. 303-311.

201. Jang, M., W. Yang, I. Shin, H. Choi, H. Chang, and E. Kim, Determination

of AM-2201 metabolites in urine and comparison with JWH-018 abuse.

International Journal of Legal Medicine, 2014. 128(2): p. 285-294.

202. Rigdon, A., P. Kennedy, and T. Kahler, LC/MS/MS analysis of metabolites

of synthetic cannabinoids JWH-018 and JWH-073 in urine. Clin. Forensic

Toxicol., 2011.

203. Batista, C., M. Berisha, M. Karst, K. Salim, U. Schneider, and R.

Brenneisen, Determination of ajulemic acid and its glucuronide in human

plasma by gas chromatography–mass spectrometry. Journal of

Chromatography B, 2005. 820(1): p. 77-82.

204. de Castro, A., B. Piñeiro, E. Lendoiro, A. Cruz, and M. López-Rivadulla,

Quantification of selected synthetic cannabinoids and Δ9-

tetrahydrocannabinol in oral fluid by liquid chromatography–tandem mass

spectrometry. Journal of Chromatography A, 2013. 1295: p. 99-106.

205. Amaratunga, P., C. Thomas, B.L. Lemberg, and D. Lemberg, Quantitative

measurement of XLR11 and UR-144 in oral fluid by LC–MS-MS. Journal

of analytical toxicology, 2014. 38(6): p. 315-321.

206. Chimalakonda, K.C., C.L. Moran, P.D. Kennedy, G.W. Endres, A. Uzieblo,

P.J. Dobrowolski, E.K. Fifer, J. Lapoint, L.S. Nelson, R.S. Hoffman, L.P.

James, A. Radominska-Pandya, and J.H. Moran, Solid-Phase Extraction

and Quantitative Measurement of Omega and Omega-1 Metabolites of

185

JWH-018 and JWH-073 in Human Urine. Analytical Chemistry, 2011.

83(16): p. 6381-6388.

207. Adcock, J.L., J.M. Terry, C.J. Barrow, N.W. Barnett, D.C. Olson, and P.S.

Francis, Chemiluminescence detectors for liquid chromatography. Drug

Testing and Analysis, 2011. 3(3): p. 139-144.

208. Holland, B.J., P.S. Francis, B. Li, T. Tsuzuki, J.L. Adcock, N.W. Barnett,

and X.A. Conlan, Chemiluminescence detection of cannabinoids and

related compounds with acidic potassium permanganate. Drug Test. Anal.,

2012. 4(7-8): p. 675-679.

209. McDermott, G.P., P. Jones, N.W. Barnett, D.N. Donaldson, and P.S.

Francis, Stable Tris(2,2′-bipyridine)ruthenium(III) for Chemiluminescence

Detection. Analytical Chemistry, 2011. 83(13): p. 5453-5457.

210. Ong, E.S., Extraction methods and chemical standardization of botanicals

and herbal preparations. Journal of Chromatography B, 2004. 812(1–2): p.

23-33.

211. Oulton, S. Scientific working group for the analysis of seized drugs. 2017

[cited 2016 16/04/2016]; Available from: http://www.swgdrug.org/.

212. Pandohee, J., B.J. Holland, B. Li, T. Tsuzuki, P.G. Stevenson, N.W.

Barnett, J.R. Pearson, O.A.H. Jones, and X.A. Conlan, Screening of

cannabinoids in industrial-grade hemp using two-dimensional liquid

chromatography coupled with acidic potassium permanganate

chemiluminescence detection. Journal of Separation Science, 2015. 38(12):

p. 2024-2032.

213. Dash, S., S. Patel, and B.K. Mishra, Oxidation by permanganate: synthetic

and mechanistic aspects. Tetrahedron, 2009. 65(4): p. 707-739.

214. Adcock, J.L., P.S. Francis, T.A. Smith, and N.W. Barnett, The characteristic

red chemiluminescence from reactions with acidic potassium

permanganate: further spectroscopic evidence for a manganese(ii) emitter.

Analyst, 2008. 133(1): p. 49-51.

215. Chen, G.N., F.X. Huang, X.P. Wu, Z.F. Zhao, and J.P. Duan,

Chemiluminescence determination of melatonin and some of its derivatives

using potassium permanganate and formaldehyde system. Analytical and

Bioanalytical Chemistry, 2003. 376(6): p. 873-878.

186

216. Han, S., Chemiluminescence determination of indole derivatives in human

body fluids and soil by flow injection analysis using potassium

permanganate. Microchimica Acta, 2010. 168(1): p. 169-175.

217. Nie, F., J. Lu, Y. He, and J. Du, Determination of indomethacin in urine

using molecule imprinting-chemiluminescence method. Talanta, 2005.

66(3): p. 728-733.

218. Hindson, C.M., P.S. Francis, G.R. Hanson, J.L. Adcock, and N.W. Barnett,

Mechanism of Permanganate Chemiluminescence. Analytical Chemistry,

2010. 82(10): p. 4174-4180.

219. Slezak, T., Z.M. Smith, J.L. Adcock, C.M. Hindson, N.W. Barnett, P.N.

Nesterenko, and P.S. Francis, Kinetics and selectivity of permanganate

chemiluminescence: A study of hydroxyl and amino disubstituted benzene

positional isomers. Analytica Chimica Acta, 2011. 707(1–2): p. 121-127.

220. Smith, Z.M., J.M. Terry, N.W. Barnett, L.J. Gray, D.J. Wright, and P.S.

Francis, Enhancing permanganate chemiluminescence detection for the

determination of glutathione and glutathione disulfide in biological

matrices. Analyst, 2014. 139(10): p. 2416-2422.

221. Smith, Z.M., J.M. Terry, C.M. Hindson, N.W. Barnett, and P.S. Francis,

On-line generation of a colloidal manganese(IV) reagent for

chemiluminescence detection. Microchemical Journal, 2013. 111: p. 67-73.

222. Dronova, M., E. Smolianitski, and O. Lev, Electrooxidation of New

Synthetic Cannabinoids: Voltammetric Determination of Drugs in Seized

Street Samples and Artificial Saliva. Analytical Chemistry, 2016. 88(8): p.

4487-4494.

223. Knight, A.W. and G.M. Greenway, Relationship between structural

attributes and observed electrogenerated chemiluminescence (ECL) activity

of tertiary amines as potential analytes for the tris(2,2-

bipyridine)ruthenium(II) ECL reaction. A review. Analyst, 1996. 121(11):

p. 101R-106R.

224. Uchikura, K. and M. Kirisawa, Electrochemiluminescence Determination

of D-, L-Tryptophan Using Ligand-Exchange High-Performance Liquid

Chromatography. Analytical Sciences, 1991. 7(6): p. 971-973.

225. Mervartová, K., M. Polášek, and J.M. Calatayud, Sequential injection

analysis (SIA)-chemiluminescence determination of indomethacin using

187

tris[(2,2′-bipyridyl)]ruthenium(III) as reagent and its application to

semisolid pharmaceutical dosage forms. Analytica Chimica Acta, 2007.

600(1–2): p. 114-121.

226. Tortajada-Genaro, L.A. and P. Campíns-Falcó, Multivariate standardisation

for non-linear calibration range in the chemiluminescence determination of

chromium. Talanta, 2007. 72(3): p. 1004-1012.

227. Kelly, T.J., C.W. Spicer, and G.F. Ward, An assessment of the luminol

chemiluminescence technique for measurement of NO2 in ambient air.

Atmospheric Environment. Part A. General Topics, 1990. 24(9): p. 2397-

2403.

228. Logan, B.K., L.E. Reinhold, A. Xu, and F.X. Diamond, Identification of

Synthetic Cannabinoids in Herbal Incense Blends in the United States.

Journal of Forensic Sciences, 2012. 57(5): p. 1168-1180.

229. Kneisel, S., F. Westphal, P. Bisel, V. Brecht, S. Broecker, and V. Auwärter,

Identification and structural characterization of the synthetic cannabinoid

3-(1-adamantoyl)-1-pentylindole as an additive in ‘herbal incense’. Journal

of Mass Spectrometry, 2012. 47(2): p. 195-200.

230. Zuba, D., B. Byrska, and M. Maciow, Comparison of “herbal highs”

composition. Analytical and Bioanalytical Chemistry, 2011. 400(1): p. 119-

126.

231. Marino, M.A., B. Voyer, R.B. Cody, A.J. Dane, M. Veltri, and L. Huang,

Rapid Identification of Synthetic Cannabinoids in Herbal Incenses with

DART-MS and NMR. Journal of Forensic Sciences, 2016. 61: p. S82-S91.

232. Ernst, L., H.-M. Schiebel, C. Theuring, R. Lindigkeit, and T. Beuerle,

Identification and characterization of JWH-122 used as new ingredient in

“Spice-like” herbal incenses. Forensic Science International, 2011. 208(1–

3): p. e31-e35.

233. Namespetra, A.M., D.A. Hirsh, M.P. Hildebrand, A.R. Sandre, H. Hamaed,

J.M. Rawson, and R.W. Schurko, 35Cl solid-state NMR spectroscopy of

HCl pharmaceuticals and their polymorphs in bulk and dosage forms.

CrystEngComm, 2016. 18(33): p. 6213-6232.

234. Hofstetter, A. and L. Emsley, Positional Variance in NMR Crystallography.

Journal of the American Chemical Society, 2017. 139(7): p. 2573-2576.

188

235. Bardet, M. and A. Pournou, Chapter Two - NMR Studies of Fossilized

Wood, in Annual Reports on NMR Spectroscopy, A.W. Graham, Editor.

2017, Academic Press. p. 41-83.

236. Ashbrook, S.E., D.M. Dawson, and V.R. Seymour, Recent developments in

solid-state NMR spectroscopy of crystalline microporous materials.

Physical Chemistry Chemical Physics, 2014. 16(18): p. 8223-8242.

237. Lee, D., G. Monin, N.T. Duong, I.Z. Lopez, M. Bardet, V. Mareau, L.

Gonon, and G. De Paëpe, Untangling the Condensation Network of

Organosiloxanes on Nanoparticles using 2D 29Si–29Si Solid-State NMR

Enhanced by Dynamic Nuclear Polarization. Journal of the American

Chemical Society, 2014. 136(39): p. 13781-13788.

238. Baldus, M., Solid-State NMR Spectroscopy: Molecular Structure and

Organization at the Atomic Level. Angewandte Chemie International

Edition, 2006. 45(8): p. 1186-1188.

239. Berendt, R.T., D.M. Sperger, E.J. Munson, and P.K. Isbester, Solid-state

NMR spectroscopy in pharmaceutical research and analysis. TrAC Trends

in Analytical Chemistry, 2006. 25(10): p. 977-984.

240. Paradowska, K. and I. Wawer, Solid-state NMR in the analysis of drugs and

naturally occurring materials. Journal of Pharmaceutical and Biomedical

Analysis, 2014. 93: p. 27-42.

241. Schaefer, J. and E.O. Stejskal, Carbon-13 nuclear magnetic resonance of

polymers spinning at the magic angle. Journal of the American Chemical

Society, 1976. 98(4): p. 1031-1032.

242. Albert, K. and E. Bayer, Characterization of bonded phases by solid-state

NMR spectroscopy. Journal of Chromatography A, 1991. 544: p. 345-370.

243. Rossini, A.J., A. Zagdoun, M. Lelli, A. Lesage, C. Coperet, and L. Emsley,

Dynamic nuclear polarization surface enhanced NMR spectroscopy.

Accounts of chemical research, 2013. 46(9): p. 1942-1951.

244. Bower, P.V., E.A. Louie, J.R. Long, P.S. Stayton, and G.P. Drobny, Solid-

State NMR Structural Studies of Peptides Immobilized on Gold

Nanoparticles. Langmuir, 2005. 21(7): p. 3002-3007.

245. Linton, R.W., M.L. Miller, G.E. Maciel, and B.L. Hawkins, Surface

characterization of chemically modified (trimethylsily) silicas by 29Si solid

189

state NMR, XPS, and IR photoacoustic spectroscopy. Surface and Interface

Analysis, 1985. 7(4): p. 196-203.

246. Pallister, P.J., S.C. Buttera, and S.T. Barry, Quantitative Surface Coverage

Calculations via Solid-State NMR for Thin Film Depositions: A Case Study

for Silica and a Gallium Amidinate. The Journal of Physical Chemistry C,

2014. 118(3): p. 1618-1627.

247. Ladizhansky, V. and S. Vega, Doping of CdS Nanoparticles by Co2+ Ions

Studied by NMR. The Journal of Physical Chemistry B, 2000. 104(22): p.

5237-5241.

248. Jäger, C., T. Welzel, W. Meyer-Zaika, and M. Epple, A solid-state NMR

investigation of the structure of nanocrystalline hydroxyapatite. Magnetic

Resonance in Chemistry, 2006. 44(6): p. 573-580.

249. Eckman, R.R., P.M. Henrichs, and A.J. Peacock, Study of Polyethylene by

Solid State NMR Relaxation and Spin Diffusion. Macromolecules, 1997.

30(8): p. 2474-2481.

identification and isolation of drugs in complex...

Documents