UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

Fullerene nanoparticles in soil: Analysis, occurrence and fate

Carboni, A.

Link to publication

Citation for published version (APA):Carboni, A. (2016). Fullerene nanoparticles in soil: Analysis, occurrence and fate.

General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.

Download date: 27 Jul 2020

1

Fullerene Nanoparticles in Soil:

Analysis, Occurrence and Fate

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctor

aan de Universiteit van Amsterdam

op gezag van de Rector Magnificus

prof. dr. ir. K. I. J. Maex

ten overstaan van een door het College voor Promoties ingestelde

commissie, in het openbaar te verdedigen in de Agnietenkapel

op dinsdag 18 oktober, te 14:00 uur

door

Andrea Carboni

geboren te Narni, Italië

2

Promotiecommissie:

Promotors: Prof. Dr. W.P. de Voogt - Universiteit van Amsterdam

Prof. Dr. K. Kalbitz - Universiteit van Amsterdam /

Technische Universität Dresden

Copromotor: Dr. J. R. Parsons - Universiteit van Amsterdam

Overige leden: Prof. Dr. Ir. P. J. Schoenmakers - Universiteit van Amsterdam

Prof. Dr. B. de Bruin - Universiteit van Amsterdam

Prof. Dr. A.P. van Wezel - Universiteit Utrecht

Dr. W. Th. Kok - Universiteit van Amsterdam

Dr. Ir. N.W. van den Brink - Wageningen UR

Faculteit der Natuurwetenschappen, Wiskunde en Informatica

Cover by Alessandro Cavalletti, Archinit

Printed by Leoni Tipografia Grafiche s.n.c. - Amelia, Italy

Fullerene nanoparticles in soil: analysis, occurrence and fate by Andrea Carboni

Proefschrift Universiteit van Amsterdam, FNWI, IBED, 2016

ISBN 978-94-91407-36-9 / Copyright © 2016

This work was funded by NanoNextNL, a micro and nanotechnology consortium of the Government of The Netherlands and 130 partners

3

“Pollution is nothing but the resources we are not harvesting. We allow

them to disperse because we've been ignorant of their value.”

R. Buckminster Fuller

4

5

Table of contents

Chapter 1. General Introduction ................................................................. 7

Chapter 2. Optimization of the extraction procedure for fullerenes into

soil ...............................................................................................................17

Chapter 3. A HPLC-UV method for the analysis of fullerenes in soils ........31

Chapter 4. An UHPLC-HRMS method for the analysis of fullerenes in

soils ..............................................................................................................53

Chapter 5. Analysis of fullerenes in soils from The Netherlands ...............75

Chapter 6. Incubation of solid state C60 fullerene under environmentally

relevant conditions .....................................................................................91

Chapter 7. Synthesis..................................................................................109

Appendices ...............................................................................................119

Summary ...................................................................................................143

Samenvatting ............................................................................................145

References ................................................................................................147

List of papers used in this thesis ..............................................................165

Curiosities .................................................................................................166

Acknowledgements ..................................................................................169

6

7

Chapter 1

Introduction

This introduction gives an overview of the terminology and classification of

nanomaterials and nanotechnology (1.1) before proceeding to focus on the

fullerenes (1.2), their sources and fate in the environment (1.3) and the

analysis of fullerenes in the environment (1.4). Finally the justification and

objectives of the thesis (1.5) are discussed.

1.1 Engineered Nanomaterials and Nanotechnology

Nanomaterials (NMs) consist of a group of chemical substances defined by

their external size between 1 and 100 nm in at least one dimension, and can

be considered a bridge between atomic or molecular structures and bulk

materials (EU Commission, 2011; Henglein et al., 1993). This is a

heterogeneous group of chemicals that can be further classified depending

on their composition (e.g. metallic, organic) as well as on the basis of their

origin (e.g. natural or anthropogenic). Although particles in the nano-size

range have likely been always present on Earth (Nowack et al., 2007), only

recently they have attracted a lot of attention due to their innovative

properties. In particular, due to the small size, NMs display a larger surface

to volume ratio in comparison with bulk materials that determines changes

in the physico-chemical properties related to electrical conductivity, colour,

solubility, mechanical strength, and catalytic activity (Reed, 1993; Ebbesen

et al., 1996; Pal et al., 1997; Powell et al., 1998; Pan et al., 2012). This makes

them appealing for novel applications and uses and in the last decades,

8

large efforts were dedicated to the synthesis of engineered nanomaterials

(ENMs) aimed to enhance technological and industrialized processes. In this

context, nanotechnology refers to the characterization, production and

application of NMs that can be achieved by manipulating their composition

and arrangement. This field is very broad, covering a wide range of different

techniques, scientific and commercial applications and products (RS & RAE,

2004). According to some, nanotechnology is expected to determine the

“next industrial revolution” and current and future applications could hold

societal benefits including economic development and employment and

advancements in material, environmental and medical sciences (RS & RAE,

2004; Roco et al., 2005). Nowadays, nanotechnology is estimated to be

worth $2.6 trillion in manufactured goods (Lux Research 2006) and the

production of engineered nanomaterials is expected to be 58,000 tons in

the 2011-2020 period (Maynard, 2006). The increasing production and use

(Wijnhoven et al., 2010; Hendren et al., 2011; Maynard et al., 2012) will

inevitably result in a larger release of these chemicals into the environment.

However, the implications of a larger occurrence and exposure to ENMs are

not fully understood and several studies raised concern about their fate,

transport, and potential adverse effects for the environment and human

health (Christian et al., 2008; Peralta-Videa et al., 2011). Nonetheless it is

difficult for regulatory agencies to develop standards on the usage,

manufacture and regulation because NMs may display a different

behaviour in comparison with “traditional” materials (Bhatt et al., 2011).

For instance, in toxicity test particles number and/or surface area may have

a more important role than concentration, raising questions about the

validity of current dose-metric methodologies and risk assessment in

general (Pakaninen, 2013). This is further complicated by the fact that not

only the size but also the shape and surface properties are important in

affecting the behaviour of nanostructured matter (Yadav et al., 2008). In

this context, key limitations include the lack of data on the environmental

fate and behaviour of engineered nanomaterials and of direct

measurements of their occurrence in aquatic or terrestrial systems (Klaine

et al., 2008; Gottschalk et al., 2010). These are in turn determined by the

lack of analytical methodologies capable of quantifying trace

concentrations of NPs in environmental matrices (Hasselhov et al., 2008;

Mueller et al., 2008; Gottschalk et al., 2009; Pan et al., 2012).

9

1.2 Fullerenes

The synthesis of fullerene in 1985 by Kroto et al., represented a milestone

in the development of nanotechnology and signed the beginning of one of

its most prominent fields, the carbon nanotechnology. Fullerenes, Cn, are

an allotrope of carbon. However, unlike other forms of carbon such as

diamond and graphite, which consist of the repetition of atomic structures,

fullerenes are discrete molecules with a defined number of atoms (Fig. 1.1).

Fig. 1.1. Structures of carbon allotropes. A) C60 fullerene, B) diamond, C) graphene and D) graphite (layers of graphene).

They display a closed-cage molecular shape where the carbon atoms are

interconnected in five- and six- membered rings. Depending on the number

of hexagonal rings that combine with 12 pentagonal rings, fullerenes can

present a theoretically unlimited number of structures. Among these, C60

was the first to be discovered and consists of 12 pentagonal and 20

hexagonal rings arranged in a football-like structure of 0.72 nm in diameter.

C60 is the most stable fullerene because of energetic reasons (Zhang et al.,

1992) and by far the most studied to date. All the carbon atoms are

equivalent in the molecule and present a sp2 hybridization, with lengths of

the bonds of 1.38 Å and 1.45 Å, for six-six and five-six bonds, respectively.

Due to the π-electrons in the rings, fullerenes could be described as three-

dimensional aromatic molecules. However, since double bonds do not

locate in the pentagonal rings, resulting in poor electrons delocalisation,

they are not “super aromatic” as initially proposed (Buhl et al., 2001; Yadav

10

et al., 2008). Fullerenes are more electronegative than most hydrocarbons,

behave like electron deficient alkenes and react readily with electron rich

species (Hirsch et al., 2005). In general, these compounds are extremely

versatile and can undergo a large number of reactions including redox,

nucleophilic attack, addition, cycloaddition and photochemical reaction as

photocycloaddiction (Kroto et al., 1994).

These reactions also give the possibility to derivatize the closed-cage

structure with consequent production of novel fullerenic materials with

modified physico-chemical properties that provide new features and

chemical characteristics (e.g. higher water solubility or conductivity). In

particular, the derivatization of fullerenes can be achieved with the

inclusion of atoms other than carbon within the structure (heterohedral

fullerenes), the inclusion of chemical species (e.g. metals, water) in the

inner space (endohedral fullerenes) and / or the functionalization of the

outer space (exohedral fullerenes) (Chai et al., 1991; Hummelen et al., 1995)

(Fig. 1.2). The unique molecular nano-size structure, combined with the

possibly unlimited number of species that may be derived, make fullerenes

appealing for a large number of possible applications. According to the

Nanotechnology Consumer Products Inventory, as of March 2006 carbon

nanomaterials (fullerenes and carbon nanotubes) were the most widely

used nanoparticles with regard to the number of products on the market.

Current applications include personal care products (e.g. as antioxidants in

cosmetics) and photovoltaics (as electron acceptors) as well as electronics

and optics (Tagmatarchis et al., 2001; Guldi et al., 2002; Burangulov et al.,

2005; Kim et al.,2006; Benn et al., 2011; Li et al., 2012). Fullerenes are also

expected to play a role in novel medical and environmental strategies

(Bakry et al., 2007; Cantrill, 2011).

11

Fig. 1.2. Examples of fullerenes derivatives structures. Left side: endohedral fullerenes displaying A) a water molecule (H2O@C60) and B) a potassium atom (K@C60) included into the C60 structure. Right side: exohedral fullerenes, C) [6,6]-Phenyl-C61-butyric acid methyl ester ([60]PCBM) and D) poly-hydroxylated fullerenes also known as fullerol or fullerenol (e.g. C60(OH)n).

1.3 Sources and environmental fate of fullerenes

The understanding of fullerenes occurrence and fate in the environment is

limited and complicated by the multiple sources from which they may

derive. Similar to any industrially produced chemical, the increasing

production and application will inevitably mean a release of these

compounds in the environment at larger concentration and in a wider

variety of ecosystem than is currently the case (Nowack et al., 2007; Tiwari

et al., 2014). In addition, fullerenes can be formed naturally during wildfire,

lightening, meteor impact and combustion processes in general where

carbonaceous materials are consumed (Daly et al., 1993; Heyman et al.,

12

1994; Tiwari et al., 2016). Thus, unintentional emissions of anthropogenic

origin may be expected due to industrial processes and transportation that

involve coal, fuel and organic matrices (e.g. as by-product of combustion,

oil refinery; Sanchis et al., 2013). A precise picture of the current worldwide

production of fullerenes is hindered by the lack of data concerning their

production and a complete inventory of the products they are included in.

However, manufacturing of fullerenes is currently estimated to be in the

range of tens of tons per year and natural/incidental contributions could

exceed that of engineered nanomaterials (ENMs) (Hendren et al., 2011;

Tiwari et al., 2016).

Once in the environment, little is known about the environmental fate of

fullerenes. The assessment of their behaviour is further complicated by

their dualistic character in water solubility. Indeed, although extremely non

polar, fullerenes are well known to create stable aqueous suspensions with

the formation of nanometer-sized agglomerates that are negatively

charged and whose water solubility is much higher than that of the pristine

structures (Deguchi et al., 2001; Javfert et al., 2008). Although it is not clear

yet to which extent these colloidal structures will form in the environment,

such phenomena of homo-aggregation, as well as the hetero-aggregation

with natural components (e.g. humic and fulvic acids), will likely determine

mobility in water and soil matrices (Wang et al., 2012; Haftka et al., 2015).

The colloidal stability can be further affected by other environmental

conditions such as ionic strength and type of ions present in the water and

the zeta potential of the fullerenes (Haftka et al., 2015). Upon release,

fullerenes will presumably interact with solar radiation, water, natural

materials (e.g. organic matter), soil, air and biota and may be transformed

or degraded as a consequence (e.g. Avanasi et al., 2014; Tiwari et al., 2014).

In this context, some studies suggest that fullerenes will be functionalized

upon interaction with ozone, water and light, and will undergo oxidative

pathways that could lead to mineralization (Hou et al., 2009; Lee et al.,

2009; Hwang et al., 2010; Tiwari et al., 2014). However, other studies

highlighted the stability of C60 in soil matrices that may determine their

accumulation in the environment (Jehlicka et al., 2005; Parthasarathy et al.,

2008). Thus, soil may act as a sink (Gottschalk et al., 2009) receiving

fullerenes through several pathways such as the direct release of ENMs, the

deposition of incidental species from the atmosphere, water transport and

13

use of biosolids (Utsunomoya et al., 2002; Sanchis et al., 2012; Navarro et

al., 2013).

1.4 Environmental analysis of fullerenes

Understanding the occurrence and fate of nanoparticles necessarily relies

on the collection of empirical data. At the time this thesis project started,

several studies had highlighted the lack of analytical methodology able to

detect and quantify fullerenes in environmental samples (Hasselhov et al.,

2008; Gottschalk et al., 2009; Isaacson et al., 2009), a task which is

complicated by the low concentrations expected and the difficulties in the

extraction from environmental matrices (Jehlicka et al., 2005). The existing

techniques presented several limitations. For instance, the majority of the

studies addressed the analysis of C60 only, whereas other pristine fullerenes

(e.g. C70 and to a lesser extent larger fullerenes such as C84) and

functionalized structures were seldom included (e.g. Bouchard et al., 2008).

Furthermore, most of the studies were focussed on water and wastewater

media, whereas soil and sediments, where these chemicals are expected to

accumulate to a larger extent, received little or no attention. Eventually,

although several methods were developed for the extraction and analysis

of fullerenes in soil and sediment matrices, these were not suited for the

extraction of samples presenting high organic carbon content, the detection

at environmentally relevant concentrations and were not able to distinguish

between particles of engineered or natural origin (Jehlicka et al., 2005; Vitek

et al., 2009; Shareef et al., 2010). These issues likely hindered the analysis

of fullerenes in several studies and made the reproducibility of positive

results a difficult task. For instance, the concentration detected in samples

from the Sudbury impact ranged from part per mil to nothing, in three

different studies on the same material (Becker et al., 1994; Heymann et al.,

1999; Elsila et al., 2005).

14

In the last few years several efforts have been made to overcome these

issues and analytical methodologies are now available that allow the

determination of fullerenes and functionalized structures in environmental

samples.

In detail, fullerenes are extremely non-polar chemicals and their analysis

typically makes use of organic solvents, such as toluene, where they are

more soluble (Ruoff et al., 1993). Robust methods for their extraction

include liquid-liquid extraction and solid-phase extraction (for water

samples, e.g. Bouchard et al., 2008; van Wezel et al., 2011) as well as

soxhlet, ultrasonication, microwave assisted extraction (MAE) and

accelerated solvent extraction (ASE) (for solid samples, e.g. Jehlicka et al.,

2005). With regard to the analysis, although separation techniques such as

electrophoresis, size exclusion chromatography and field flow fractionation

were tested, the general opinion is that the combination of liquid

chromatography with ultraviolet-visible and mass spectrometric detection

offers the greatest potential for routine analysis (Isaacson et al., 2009). This

usually consisted of non-aqueous reverse phase methodologies employing

non-polar mobile phases (e.g. toluene or toluene-acetonitrile mixtures)

(e.g. Bouchard et al., 2008, Shareef et al., 2010). The separation of the

fullerenes has been achieved with both standard octadecyl silica as well as

functionalized silica columns and several stationary phases such as 5PBB

and pyrenylpropyl silica that were specifically designed for the analysis of

fullerenes (e.g. Hou et al., 2009).

Concerning the detection methodologies, fullerenes possess strong light

absorptivity in the UV range (specifically at circa 330 nm) and UV detection

has been employed in their analysis in water, sediments, soil and biological

samples (Moussa et al., 1997; Xia et al., 2006; Bouchard et al., 2008; Shareef

et al., 2010; Wang et al, 2011). However, it must be noted that, although

close to some mass spectrometric techniques in terms of sensitivity (Wang

et al., 2010), UV detection lacks the specificity needed for an unambiguous

identification, especially in complex matrices where it can suffer from the

presence of co-extractants. Mass spectrometry can overcome these issues

due to the higher selectivity provided by the m/z signals and the larger

sensitivity, especially in the analysis of exohedral fullerenes (i.e.

functionalized structures that undergo fragmentation in MS/MS) (e.g. van

Wezel et al., 2011; Kolkman et al., 2013; Astefanei et al., 2014b). Although

15

ionization is usually achieved with electrospray interfaces operating in

negative mode, other techniques, such as heated electrospray (H-ESI) and

atmospheric pressure chemical and photo ionization systems (APCI and

APPI, respectively), have been tested and can improve the ionization

efficiency (Nunez et al., 2012; Astefanei et al, 2014b; Emke et al., 2015;

Sanchis et al., 2015). Eventually, identifications of fullerenes in

environmental matrices at concentrations in the part per billion range were

achieved with high resolution MS instruments such as Orbitrap and FTICR

that can provide better identification due to higher mass and isotopic

cluster distribution accuracies (Astefanei et al., 2014b; Emke et al., 2015;

Sanchis et al., 2015).

1.4 Justification and objectives of the thesis

The last two decades have been characterized by the rise of

nanotechnology. However, the interest toward the novel and possible

applications of nanomaterials has been accompanied by the concern about

the implications that these may have for humans and the environment. In

this context, the NanoNextNL project was started with the aim of studying

several aspects of micro- and nano-materials, from engineering to

environmental and life science as well as regulatory and societal aspects.

Within this framework, research was started at the University of

Amsterdam (UvA) and Watercycle Research Institute (KWR) and focussed

on a class of carbon-based nanomaterials, the fullerenes, whose

characterization in the environment was still largely unknown and limited

by the lack of analytical methodologies.

Thus, the general objective of the present thesis was to study the

occurrence and fate of fullerenes in the terrestrial environment, which is

done by first developing analytical methods that allow their analysis in soil

16

samples. First, the extraction of several fullerenes and functionalized

structures was tested from soil samples differing in properties such as the

texture and composition. To this end a non-selective UV detection method

was developed that enables one to determine the totality of fullerenes

species in the samples with an extraction methodology that is potentially

applicable in routine analysis (Chapter 2 and 3). Next, a more sensitive and

selective methodology was developed employing mass spectrometric

detection, in order to fulfil identification criteria and provide unambiguous

determination of the fullerenes in soil samples. This included the

optimization of the method for the analysis of fullerenes at environmentally

relevant concentrations and in complex soil matrices (Chapter 4). These

methodologies were then applied for the study of fullerenes occurrence

and fate in the environment. First, an environmental survey was carried out

for the detection of fullerenes in soils collected in the Netherlands. This also

allowed a partial understanding of their sources and transformation in the

environment (Chapter 5). Following, in Chapter 6 incubation studies were

carried out in order to shed light on the fullerenes fate upon release in the

environment. This included the irradiation with UVA light at

environmentally relevant conditions and the characterization of

transformation products that may derive from their degradation. Chapter 7

finally provides a synthesis of the results obtained.

17

Chapter 2

Optimization of the extraction

procedures for fullerenes in soil

18

Abstract

Two extraction techniques, accelerated solvent extraction (ASE) and a

combination of ultra-sonication and shaking extractions (SSh) were

compared for the recovery of fullerenes spiked into sandy soil matrices.

Consecutive steps of extractions were tested at different concentrations and

the use of polar solvents was tested for the enhancement of the

performance. The results indicate that average recoveries were in the range

of 70% or higher with both techniques at all the concentrations tested.

Furthermore, although SSh delivered higher recoveries at larger

concentrations, when fullerenes and functionalized structures were spiked

at the concentration of 5 μg/kg, both techniques had similar performance.

Eventually, more polar solvents were not applicable in the extraction or in

clean-up of the samples. In general, although ASE is an automatize process

and requires one cycle of extraction, a procedure involving two cycles of SSh

is recommended as more feasible for the routine analysis of a large number

of samples.

19

2.1 Introduction

The presence of fullerenes in natural samples reflects very specific

conditions and natural events in which these chemicals are formed (Buseck

et al., 1992; Becker et al., 1995). However, their increasing manufacturing

and application in emerging nanotechnology is expected to lead to a higher

occurrence in the environment and soil could act as a sink for their

accumulation (Piccinno et al., 2012; Gotthschalk et al., 2009). The lack of

proper analytical methodologies likely hindered both the analysis of these

nanoparticles in several studies (Vitek et al., 2009) and the reproducibility

of positive results (Decker et al., 1994; Heymann et al., 1999; Elsila et al.,

2005). In particular, extraction methodologies are needed for monitoring

studies and routine analysis, which are critical for the understanding of

fullerenes occurrence and fate in the environment. In this context,

extraction techniques such as Soxhlet and supercritical CO2 extraction can

recover fullerenes from solid matrices (soil, carbon soot) but can be

expensive and / or time-consuming and are not suitable for routine analysis

or screening of a large number of samples in general (Issacson et al., 2009).

Thus, techniques such as ultrasonication and accelerated solvent extraction

can provide faster analysis with similar or better performance. In particular,

ultrasonication is a robust method whose performance for C60 extraction

from soil samples was already investigated by Jehlicka et al. (2005) and

Vitek et al. (2009) who reported acceptable recoveries of extraction (>

75%). On the other hand, accelerated solvent extraction (ASE) presents the

advantage of being automated and was recently employed by Shareef et al.

(2010), who reported good recoveries of extraction (>80%) in several soil

matrices. However, when these techniques were compared for the

extraction of fullerenes from carbonaceous matrices (e.g. bituminous coal,

shungite), both techniques presented low recoveries (usually < 5%) (Jehlicka

et al., 2005). In these studies a comparison of the two methodologies was

not performed for soil samples that may be representative of the matrices

in which fullerenes are present in the environment. Furthermore, only C60

was investigated and no other fullerenes and functionalized structures were

included. Thus, in the present work the extraction of sandy soil samples

spiked with fullerenes was performed with both ASE and a combination of

ultra-sonication and shaking extraction. A first experiment was aimed to

20

compare these techniques for the extraction of C60 and C70 at different

concentrations and to assess the contribution of consecutive steps of

extraction to the overall recovery. Then, a second experiment compared

the techniques for the extraction of fullerenes and functionalized fullerenes

at lower concentrations. In this case, the effect of a pre-extraction and pre-

washing with polar solvents was tested.

2.2 Materials and methods

2.2.1 Reagents and chemicals

Toluene, methanol and acetonitrile (Biosolve, Dieuze, France) were all

analytical grade. C60 fullerene (purity >99.9%) was obtained from MER

Corporation. (Tuscon, Arizona, United States). C70 fullerene (purity >99%),

was purchased from Sigma-Aldrich (Steinheim, Germany). [6,6]-Phenyl-C61-

butyric acid methyl ester ([60]PCBM) (purity >99) and [6,6]-Phenyl C71

butyric acid methyl ester ([70]PCBM) (purity >99%) were purchased from

Solenne B.V. (Groningen the Netherlands). Stock solutions of the individual

fullerenes were prepared in toluene at a concentration of 500 mg/L

according to the method described by Kolkman et al. (2013). The solutions

were placed in the dark overnight on a rotary shaker to achieve complete

dissolution of the fullerenes. Diluted solutions for the individual fullerenes

and their mixture were obtained by diluting aliquots from the individual

stock solutions. The solutions were stored at 4°C in the dark and sonicated

for 2 min before use. Sandy soil was collected in the Flevopark area,

Amsterdam, the Netherlands (52°21'55.09"N, 4°57'3.88"E) and consisted in

a top-soil A horizons (top 10-15 cm from the surface) whose texture was

assessed according to WRB 2006. The sample was placed in a freezer at -

20°C overnight and lyophilized with a Scanvac Coolsafe freeze-dryer

(Labogene, Lynge, Denmark) in order to remove traces of water. The dried

sample was finely ground with an agate mortar and sieved. Amendment of

the soil was achieved as follows: three samples (200 g each) of soil were

placed into glass jars and fullerenes were added by spiking a C60 and C70

21

stock solution in toluene to obtain a final concentration of 80 µg/kg, 15

µg/kg and 5 µg/kg each. Following, a sample was made with the same

procedure but spiking both fullerenes and functionalized fullerenes to

obtain a final concentration of 5 µg/kg. The soils were then homogenized

by stirring and left in the dark for 24 h to allow the solvent to evaporate.

2.2.2 Extraction procedures

Two sets of experiments were carried out to compare ASE with a

combination of ultrasonic and shaking extraction. In the first, sandy soil was

spiked at three different levels: 80, 15 and 5 µg/kg, respectively and toluene

was used as the extraction solvent. For ultrasonication and shaking

extraction, 15 g of soil were weighed and placed into a glass centrifuge tube,

15 ml of toluene were added and the samples were placed open into a

Bransonic 12 ultrasonic bath (Branson, Danbury CT, United states)

operating at 50 kHz for 30 min. Then, the tubes were closed with a glass

stopper and shaking extraction was performed with an orbital shaker at 160

rpm for 90 min. Subsequently, the samples were centrifuged at 2000 rpm

and the toluene supernatant was filtered through a 4-7 µm pore size

prepleated paper filter (Whatman, Maidstone, United Kingdom) into 60 ml

amber glass vials. The filter was rinsed with 3 ml of toluene and the

extraction was repeated a second time with the same procedure. For the

first experiment, the extracts were collected separately.

For accelerated solvent extraction (ASE), the soil samples (circa 15 g) were

weighed and transferred directly into a 66-mL stainless steel extraction cell

(Dionex, Sunnyvale, CA, USA). PTFE filters were placed at the bottom and

the top of the extraction cell. Extraction was carried out using an ASE

(Dionex, Sunnyvale, CA, USA) operating with the following parameters:

1500 psi, 50°C, static time 15 min and 120 sec purge with N2. For the first

experiment, three cycles of extraction with toluene were carried out with

the same parameters and the extracts collected separately.The extracts

were evaporated in a water bath at 60 °C under a gentle nitrogen flow until

approximately 5 ml. Finally, the extracts were filtered with 0.45 µm

regenerated cellulose filters and concentrated to a final volume of ~ 1-2 ml.

22

All experiments were performed in triplicate and non-spiked soils were

extracted with the same protocol as reference.

For the second set of experiments, a sandy soil sample spiked at the low

level (5µg/kg) was extracted using three solvents or solvent combinations.

In the first subset an extraction using toluene only was carried out based on

the results of the first experiment. ASE was used to this end with a single

extraction cycle, whereas sonication combined with shaking was carried out

using two consecutive extractions (two cycles). The next subset consisted

of ASE extraction using two consecutive cycles with acetonitrile and

toluene, respectively (the two extracts were then combined), and

sonication/shaking using three consecutive extractions: a first one with

acetonitrile, and the next two employing toluene (combining the three

extracts). Finally, the third subset consisted of an ASE extraction of the

spiked soil using two consecutive cycles: first a methanol extraction

(assuming that methanol would remove polar constitutents from the

sample; this extract was discarded), followed by a toluene extraction.

2.2.3 HPLC-MS analysis

Analyses were performed with a hybrid LTQ Orbitrap mass spectrometer

(Thermo Electron, Bremen, Germany) provided with an ESI interface and

interfaced to a Surveyor HPLC system (Thermo Electron, Bremen, Germany)

for the chromatographic separation. The separation was achieved with a

Cosmosil® Buckyprep column consisting of 3-(1-pyrenyl)propyl groups

stationary phase (4.6 mm ID x 250 mm, Nacalai-Tesque, Kyoto, Japan)

equipped with a C18 silica pre-column. The isocratic elution was obtained

with a mobile phase composition of toluene/acetonitrile (80/20 v/v) and a

flow rate of 0.8 ml/min. Post-column addition of methanol was performed

to improve ionization efficiency in ESI by means of a peek high pressure

mixing-tee at a flow rate of 200 µl/min. The injection volume was 20 µl.

With these settings, the retention times of the pristine fullerenes was 20.5

and 28.5 min for C60 and C70, respectively, whereas the functionalized

structures [60]PCBM and [70]PCBM eluted at 11.0 and 17.0 min,

respectively. The mass spectrometer operated according to the settings

described by Kolkman et al. (2013). Briefly, analysis was performed using

23

negative electrospray ionization and the capillary used was a metal needle

maintained at a temperature of 400°C. The sheath, sweep and auxiliary

gases were set to arbitrary units of respectively 30, 10, 10. A source voltage

of 3 kV and a capillary voltage of -80 V were used. The tube lens was set to

-200 V. Full scan high accuracy mass spectra were acquired in the range of

300-2000 m/z with the resolution set at 30,000 (FWHM). External

calibration curves were obtained analysing standard solutions in toluene at

concentrations ranging from 1 µg/L to 128 µg/L and quantification was

based on the sum of the peak areas of the accurate masses of the fullerene

compound and all its related adducts as described by van Wezel et al.

(2011).

2.3 Results and discussion

The experiments performed in this work were aimed to compare two

extraction techniques: accelerated solvent extraction (ASE in the further

manuscript) and a combination of ultrasonication and shaking (SSh in the

further manuscript).

The first consisted in the comparison of these techniques for the extraction

of pristine fullerenes spiked into soil at three concentrations: 80, 15 and 5

μg/kg which will be referred as high, medium and low, respectively. The

results are presented in Figure 2.1 and in the Table 2.1. In general, similar

to what was reported in previous studies (Vitek et al. 2009, Shareef et al.,

2010) the overall recoveries were acceptable (≥ 70%) for both C60 and C70 at

all the concentrations tested with the exception of one (SSh of C70 at low

concentration, 65%). Variability within the same sample was also

acceptable and generally below 10%. Furthermore the results indicate that

(I) consecutive steps of extraction had a more relevant impact on the overall

recovery in SSh than in ASE. In detail, the contribution of further extractions

after the first was 6.4%, on average, for SSh and 1.4%, on average, for ASE.

(II) For the high and medium concentrations, the overall performance of

24

SSh was better than ASE. Especially in the extraction of C60, SSh showed

recoveries of 96% on average, 25% higher than that of ASE. (III) However,

when samples at the low concentration were extracted, ASE showed slightly

better performance, 8.5% higher than SSh on average. Eventually it must be

noted that ASE showed similar performances at all the concentrations

tested whereas SSh delivered lower recoveries when tested for the

extraction of the low treatments.

Table 2.1. Recoveries of consecutive steps extraction for C60 and C70 spiked

at different concentrations into sandy soil.

ASE: Accelerated solvent extraction (toluene). SSh: Ultrasonication and shaking extraction (toluene).

25

Fig. 2.1. Performance of accelerated solvent extraction (ASE, orange) and a combination of ultrasonication and shaking (SSh, green) in the extraction of (a) C60 and (b) C70 from sandy soil samples. SSh and ASE extraction were carried out with toluene; consecutive extraction steps of ASE and SSh are indicated by different colors according to the legend.

26

From the results of the first experiment it was concluded that when using

toluene, one extraction cycle sufficed in the case of ASE, whereas for SSh

two cycles appeared to be necessary. This information was used in the

setup of the second set of experiments, where - in addition to a toluene-

only extraction with a single cycle for ASE or two consecutive cycles in the

case of SSh, the extracts of which were then combined - the influence of

additional solvents with different polarities were tested on sandy soil

samples spiked at the low level. In these second set of experiments, also

functionalized fullerenes, specifically [60]PCBM and [70]PCBM were spiked

into the soil.

In general, the recoveries obtained for the four fullerenes when using only

toluene were acceptable with both the techniques (~70%), confirming the

results observed for C60 and C70 in the first experiment (Table 2.2). For the

functionalized fullerenes recoveries were between 64 and 89%.

Table 2.2. Recovery of extraction for fullerenes and functionalized fullerenes spiked at 5 μg/kg into sandy soil.

ASE: Accelerated solvent extraction. SSh: Ultrasonication and shaking

extraction.

Although ASE, when using toluene only, provided higher recoveries for all

the compounds except for C60, the differences between the two

methodologies were within the variability of the treatment, with the

27

exception of [60]PCBM. With regard to the functionalized structures,

[60]PCBM was extracted to a higher extent in comparison with the non-

functionalized C60 and the other structures. This could be due to the higher

polarity of the functionalized C60 that resulted in a larger extractability of

the compound due to (I) a higher solubility in toluene or (II) a lower binding

to the soil particles. However, [70]PCBM was recovered to the same extent

as the pristine fullerenes, suggesting that the functional group did not play

a role in determining the overall extractability of the fullerenes in the

present study. The employment of a more polar solvent (acetonitrile), in

combination with the toluene, was expected to enhance the extractability

of the fullerenes, but resulted in a decrease of the recoveries of all the

compounds under investigation (Fig. 2.2b and Table 2.2). This can be

explained with the fact that, in addition to not recovering the fullerene

itself, the acetonitrile prevented the toluene entering the soil matrix. Pre-

washing of the samples with methanol was tested as a possible clean-up

procedure, i.e. aimed to remove the polar components of the sandy soil

that could interfere with both extraction and detection of the fullerenes.

However, in this case a large loss in the recoveries was observed (Table 2.2).

Obviously methanol also extracted a part of the fullerens from the soils

samples. The methanol pre-wash procedure was not further investigated.

In general, although SSh showed better performances for higher

concentrations and ASE was slightly better in recovering fullerenes at lower

concentrations, the two techniques provided similar results and are

therefore good options for future studies on the topic. It must be noted that

this work did not take full advantage of one of the main ASE characteristics,

which is that of operating at high temperature while organic solvents can

be kept at the liquid state, due to the high pressure generated in the ASE

cells. Since fullerene solubility is maximum around 25°C (Ruoff et al., 1993),

in this work the temperatures were relatively low in comparison with other

ASE operational procedures.

With regard to the operational procedures, which in the framework of this

study should be optimized for routine analysis, best SSh required two cycles

of extraction whereas ASE achieved good results with one cycle of

extraction only. Furthermore, ASE is an automated process and, in principle,

more reproducible. However, during this work, ASE was more time-

28

consuming and solvent requiring than SSh with regard to both sample

preparation and cleanup. Furthermore, extract volumes, which cannot be

controlled by the operator, were always larger in the case of ASE than those

obtained in SSh and resulted in further time and materials losses (e.g.

toluene, nitrogen) during the subsequent concentration step. Due to these

drawbacks, and because the performance of the two techniques are very

similar, SSh is recommended for routine analysis since it provides an overall

faster and more environmental friendly procedure. Eventually, further

tests will be required in order to assess the performance of these

procedures on soil matrices differing in physico-chemical properties such as

texture and chemical composition.

29

Fig. 2.2. Performance of accelerated solvent extraction (ASE, orange) and a

combination of ultrasonication and shaking (SSh, green) in the extraction of

fullerenes and functionalized fullerenes from sandy soil. (a) Samples

extracted with toluene and (b) samples that underwent pre-extraction with

acetonitrile.

30

2.4 Conclusions

The aim of this work was to test two extraction techniques, and specifically

accelerated solvent extraction (ASE) and a combination of ultrasonication

and shaking (SSh), for the analysis of fullerenes in soil samples. In general,

when fullerenes were spiked at the concentration of 80 and 15 μg/kg, SSh

showed higher recoveries than ASE but at the lower concentration of 5

μg/kg the two techniques were comparable. Furthermore, optimal

extraction required toluene only and could not be enhanced with the use

of more polar solvents in pre-extraction or clean-up steps. In addition of

being automated, ASE had the advantages of achieving the best results with

one cycle of extraction, whereas SSh required a double-step procedure.

Nonetheless, SSh resulted in an overall faster procedure and required less

materials and energy to be accomplished. Thus, the authors recommend

the use of SSh for the analysis of fullerenes in soils and especially for

monitoring studies where a large batch of samples is expected.

31

Chapter 3

A HPLC-UV method for the

analysis of fullerenes in soils

Published as:

Carboni A., Emke E., Parsons J. R., Kalbitz K., de Voogt P. 2013. An analytical

method for determination of fullerenes and fullerene derivatives in soil with

high performance liquid chromatography and UV detection. Analitica

Chimica Acta 807, 159-165.

32

Abstract

Fullerenes are carbon-based nanomaterials expected to play a major role in

emerging nanotechnology and produced at an increasing rate for industrial

and household applications. In the last decade a number of novel

compounds (i.e. fullerene derivatives) is being introduced into the market

and specific analytical methods are needed for analytical purposes as well

as environmental and safety issues. In the present work eight fullerenes (C60

and C70) and functionalized fullerenes (C60 and C70 exohedral-derivatives)

were selected and a novel liquid chromatographic method was developed

for their analysis with UV absorption as a method of detection. The resulting

HPLC-UV method is the first one suitable for the analysis of all eight

compounds. This method was applied for the analysis of fullerenes added to

clayish, sandy and loess top-soils at concentrations of 20, 10 and 5 µg/kg

and extracted with a combination of sonication and shaking extraction. The

analytical method limits of detection (LoD) and limits of quantification (LoQ)

were in the range of 6-10 µg/L and 15-24 µg/L respectively for the analytical

solutions. The extraction from soil was highly reproducible with recoveries

ranging from 47 ± 5 to 71 ± 4% whereas LoD and LoQ for all soils tested were

of 3 µg/kg and 10 µg/kg respectively. No significant difference in the

extraction performance was observed depending of the different soil

matrices and between the different concentrations. The developed method

can be applied for the study of the fate and toxicity of fullerenes in complex

matrices at relatively low concentrations and in principle it will be suitable

for the analysis of other types of functionalized fullerenes that were not

included in this work.

33

3.1 Introduction

Since their discovery in 1985 by Kroto et al. (Kroto et al., 1985), fullerenes

have attracted a lot of interests due to their unique structure and innovative

properties and are nowadays considered as some of the most promising

materials in nanotechnology. Fullerenes are very versatile compounds

already applied in several fields such as optics and electronics as well as

cosmetics and in medical research (Tagmatarchis et al., 2001; Guldi et al.,

2002; Burangulavet al., 2005; Kim et al., 2006) with a worldwide production

estimated in tens of tons per year, that is expected to increase in the near

future (Hendren et al., 2011; Piccinno et al., 2012). Furthermore, the

possibility to functionalize the closed cage structure by the covalent binding

of external groups to the fullerene’s surface (i.e. exohedral fullerene

derivatives) increase the solubility of these compounds in organic as well as

polar solvents and consequently widens their range of applications and

uses. Contrary to pristine compounds such as C60 and C70, that can be

naturally produced during highly energetic events such as lightening (Daly

et al., 1993) and massive wildfires (Heymann et al., 1994), functionalized

fullerenes are in all respect engineered nano-materials (ENMs). Firstly

described by Hummelen et al. in 1995 (Hummelen et al., 1995), the

fullerene derivative 1-(3-methoxycarbonyl)propyl-1-phenyl[6,6]C61, better

known as [60]PCBM, is to date one of the most studied in the field of organic

photovoltaic (OPV) materials (Dang et al., 2011) and has been proposed for

the construction of organic field-effect transistors (OFETs) (Tiwari et al.,

2007) and photo detectors (Baierl et al., 2010). In the last decade, a number

of PCBM-like chemicals differing in the substituent group (e.g. thienyl

analog of [60]PCBM, Popescu et al., 2006), number of substituents (e.g.

bisadducts, Lenes et al., 2008) or the functionalization of fullerenes other

than C60 (e.g. 70[PCBM], Wienk et al., 2003) as well as compounds with

different functionalization (e.g. C60-pyrrolidines, Marchesan et al., 2005) are

being produced and studied for their use in novel applications. Despite the

broad interest in the development of new engineered nanomaterials,

knowledge on the human safety and environmental issues of fullerenes and

their derivatives is scarce. Fullerenes entering the environment as

consequence of their production and use will presumably accumulate in soil

34

and sediments. Although functionalized fullerene derivatives have been

recently included in environmental monitoring (Sanchis et al., 2011 and

2013), most of the research so far has been focused on C60 only and no

chromatographic methods have been developed yet for the analysis of the

functionalized fullerenes structures. Among the analytical techniques that

have been applied to the analysis of fullerenes, liquid chromatography

appears to be the most feasible method for routine analysis and the main

advancements in this field have already been reviewed elsewhere (Baena

et al., 2002; Isaacson et al., 2009). In general, although octadecil silica (ODS)

stationary phases can be used to separate compounds such as C60 and C70,

better performance is achieved with other materials that offer a higher

surface for the interaction and therefore retention of fullerenes (e.g. 2-(1-

pyrenil)ethylsilica or 3-(pentabromobenzyl)oxy-propylsilylsilica) particularly

when more compounds are analyzed in a mixture.

Toluene is the most common mobile phase applied due to the high solubility

of fullerenes in this solvent at room temperature (Ruoff et al., 1993) and

can be used as only eluent when C60 is the only analyte under investigation.

When other fullerenes (e.g. C70) or functionalized fullerenes such as

[60]PCBM were included in the study, more polar solvents such as

acetonitrile (Bouchard et al., 2008), hexane or isopropanol (Deye et al.,

2008) have been used as modifiers to enhance the separation. Fullerenes

absorb light in the 300-350 nm range and UV-vis detection is a powerful tool

for their analysis in combination with HPLC because of the broad linearity

range and high sensitivity. In a recent study, Wang et al. (Wang et al., 2010)

compared UV-vis and mass spectrometry (MS) for the detection of C60 in

HPLC and concluded that, despite the higher selectivity of MS based on the

m/z ratio, the two techniques are comparable in terms of sensitivity and

UV-vis offers a larger linear range. HPLC-UV methods have been used for

the analysis of fullerenes in different matrices such as soil (Shareef et al.,

2010; Perez et al., 2013), artificial sediments (Wang et al., 2011), surface

and groundwaters (Bouchard et al., 2008) and biological matrices (Moussa

et al., 1997; Xia et al., 2006) but most of these studies were focused on C60

and occasionally higher fullerenes whereas functionalized structures were

seldom included. Furthermore, fullerenes and fullerene derivatives have

also shown to emit fluorescence at room temperature when dissolved in

35

organic solvents (Lin et al., 1995; Zhao et al., 2006) but no data are available

of fluorescence detection coupled to HPLC.

In the present study we developed a HPLC method with UV detection for

the determination of eight selected fullerenes and functionalized

fullerenes. After optimization the method was tested for the analysis of the

fullerenes in environmental matrices. Soil and sediments might act as a sink

for the accumulation of hydrophobic fullerenes after their release into the

environment but few studies have addressed yet the issue of analyzing

these compounds in these matrices (e.g. Vitek et al., 2009; Shareef et al.,

2010; Perez et al., 2013). Furthermore, in the majority of the studies that

have addressed the issue, the concentrations tested were relatively high

(hundreds µg/kg and above) with the exception of a recent study from Perez

et al. (2013). None of these studies included functionalized structures other

than [60]PCBM. Thus, in the present work three top-soils differing in their

properties as texture and organic matter content, namely sandy, clayey and

loess soils were spiked with toluene standard solutions containing all the

fullerenes under investigation to a final concentration of 20, 10 and 5 µg/kg

for each compound and analyzed using the HPLC-UV method.

3.2 Materials and Methods

3.2.1 Reagents and chemicals

Table 3.1 presents characteristics of the fullerenes in the present study.

Toluene and Acetonitrile (Biosolve, Dieuze, France) were both analytical

grade. Stock solutions of the individual fullerenes were prepared in toluene

at a concentration of 500 mg/L according to the method described by

Kolkman et al. (2013). The solutions were placed in the dark overnight on a

rotary shaker to achieve complete dissolution of the fullerenes. Diluted

solutions for the individual fullerenes and their mixture were obtained by

diluting aliquots from the individual stock solutions. The solutions were

stored at 4°C in the dark and sonicated for 2 min before use.

36

Table 3.1. Fullerene standards used in the present study.

3.2.2 Soil sampling, soil characterization and sample treatment

Sandy soil was collected in the Flevopark area, Amsterdam, the Netherlands

(52°21'55.09"N, 4°57'3.88"E), the loess soil was collected from an

agricultural field in south Limburg, the Netherlands (50°53'58"N, 5°53'16"E)

and the clayish soil was collected in Dikkebuiksweg, the Netherlands

(50°50'03"N, 5°54'27.7"E). All the soils in the present study were sampled

from top soils A horizons within the first 10-15 cm from the surface and

their texture was assessed according to WRB 2006. The samples were

placed in a freezer at -20°C overnight and lyophilized with a Scanvac

Coolsafe freeze-dryer (Labogene, Lynge, Denmark). The dried samples were

finely ground with an agate mortar and sieved.

At first we obtained an aqueous extract to measure dissolved organic

carbon (DOC). Samples of 20 g for each soil were placed in 200 ml

polyethylene bottles and 100 ml of ultrapure water were added (dilution

1:5) before to undergo shaking extraction for 2 h at 120 rpm with a

Laboshake orbital shaker (Gerhardt, Königswinter, Germany). The samples

were then transferred into 50 ml plastic tubes, centrifuged for 15 min at

2000 rpm with a Rotofix 32A (Hettich, Tuttlingen, Germany) and the

supernatants were transferred into plastic syringes and filtered with 0.2 µm

cellulose ester membrane filters (Whatman, Maidstone, United Kingdom)

37

previously rinsed with ultrapure water. The pH of the final extracts was

measured with a Consort C831 electrode (Consort NV, Turnhout, Belgium)

and DOC and IC (inorganic carbon) were determined using a TOC-VCPH

(Shimadzu, Kyoto, Japan). The carbon and nitrogen contents in the dried soil

samples were measured using a Vario EL Cube (Elementar, Hanau,

Germany). All the experiments for the soils characterization were made in

triplicate. Three samples (200 g) for each soil were placed into glass jars and

fullerenes were added by spiking a fullerene stock solution in toluene to

obtain a final concentration of 20 µg/kg, 10 µg/kg and 5 µg/kg. The soils

were then homogenized by stirring and left in the dark for 48 h to allow the

solvent to evaporate.

3.2.3 Extraction

10 g of soil from each jar were weighed and placed into a glass centrifuge

tubes, 10 ml of toluene were added and the samples were placed open into

a Bransonic 12 ultrasonic bath (Branson, Danbury CT, United states)

operating at 50 kHz for 30 min. Then, the tubes were closed with a glass

stopper and shaking extraction was performed with an orbital shaker at 160

rpm for 90 min. Subsequently, the samples were centrifuged at 2000 rpm

and the toluene supernatant was filtered through a 4-7 µm pore size

prepleated paper filter (Whatman, Maidstone, United Kingdom) into 40 ml

amber glass vials. The filter was rinsed with 3 ml of toluene and the

extraction was repeated a second time by adding 8 ml of toluene. In this

latter procedure, the samples were not centrifuged and the soil samples

were transferred directly to the paper filters. After elution of the solvent,

each sample was rinsed with 5 ml of toluene and the extracts were

evaporated in a water bath at 60 °C under a gentle nitrogen flow until

approximately 3 ml. Finally, the extracts were filtered with 0.45 µm

regenerated cellulose filters and concentrated to a final volume of ~ 0.5-1

ml. All experiments were performed in triplicate and non-spiked soils were

extracted with the same protocol as reference.

38

3.2.4 HPLC with UV and fluorescence detection

UV-vis and fluorescence spectra of the fullerenes were obtained analyzing

stock solutions of the single compounds in quartz cuvettes with an Olis DW-

2000 spectrophotometer and an Olis DM45 spectrofluorimeter (Olis, Bogart

GA, United States), both equipped with Olis SpectralWorks software. Liquid

chromatography was performed with a Shimadzu Prominence system

(Shimadzu, Kyoto, Japan) equipped with a diode-array detector and a

fluorescence detector. The wavelengths monitored for UV detection were

305 nm and 332 nm. For fluorescence detection, emission wavelengths at

400 nm, 550 nm and 700 nm were monitored with excitation wavelength

set at 286 nm, 332 nm, 400 nm or 463 nm. The data were collected with the

LCsolution software. The separation was achieved with a Cosmosil®

Buckyprep column consisting of 3-(1-pyrenil)propyl groups stationary phase

(4.6 mm ID x 250 mm, Nacalai-Tesque, Kyoto, Japan) equipped with a C18

silica pre-column at a flow rate of 1 ml/min and an injection volume of 20

µl. External calibration curves were obtained analyzing standard solutions

in toluene at concentrations ranging from 4 µg/L to 1 mg/L and

quantification was based on chromatographic peak areas whereas limits of

detection (LoD) and quantification (LoQ) were assessed observing the signal

to noise ratio (S/N) and considering LoD as the concentration with S/N=3

and LoQ as the concentration with S/N=10.

3.3 Results and discussion

3.3.1 HPLC-UV method

The separation of fullerenes in liquid chromatography necessitates the use

of an apolar mobile phase able to dissolve and elute the compounds in a

relatively short time. In this study toluene was applied as the main eluent

in the mobile phase in combination with a specific stationary-phase,

composed of pyrenyl-propyl functionalized silica (Buckyprep), that

enhances the retention of fullerenes as a result of the large ligand that can

interact with the aromatic structure of the fullerenes. This non-aqueous

39

chromatographic system can be nominally referred to as normal-phase

liquid chromatography due to the apolarity of both the mobile and

stationary phases. Since the isocratic elution with toluene as only eluent

resulted in a partial or total co-elution of some of the compounds in the

mixtures a more polar solvent, in this case acetonitrile, was added to the

mobile phase in different percentages to enhance the separation of the

analytes. The final optimized method (fig. 3.1) consisted in a gradient

elution starting with 75:25, toluene:acetonitrile (% volume) and the gradual

conversion after 6 min to 100% toluene to allow a faster elution of the more

apolar compounds. With these settings the elution of all the analytes is

obtained within 25 min while the whole method lasted 32 min to allow the

system to equilibrate prior to the next analysis.

Figure 3.1. HPLC-UV chromatogram of fullerenes and functionalized

fullerenes in toluene containing 20 ng of each analyte. [1]: bis[60]PCBM,

[2]: [60]PCBO, [3]: [60]PCBB, [4]: [60]PCBM, [5]: [60]ThCBM, [6]:

[70]PCBM, [7]: C60 and [8]: C70.

As shown in fig. 3.1, the order of elution is correlated with: 1) the number

and presence of functionalization on the cage, that increase the solubility

of the compounds in the mobile phase and decrease the surface available

for the interaction with the pyrenyl-propyl groups in the stationary phase

(functionalized elute earlier than pristine fullerenes) and 2) the size of the

40

cage (C60 structures elute earlier than C70). Thus, the double functionalized

bis[60]PCBM (Fig. 3.1, peak 1) is the first compound to elute with a relatively

broad peak, between 3.5 and 4.7 min. Its jagged peak-shape might be due

to either the presence of different isomers (60 positions are available on

the structure for the attachment of the two functional groups) or by the

formation of micelles in the solution. The four C60 derivatives (fig. 3.1, peaks

2, 3, 4 and 5), which are not baseline resolved, eluted in a cluster between

6 to 7.5 min followed by the [70]PCBM (fig. 3.1, peak 6) at time 11.8 min.

This latter peak has a shoulder that might be due to oxidized products or

the presence of different isomers. C60 and then C70 (fig. 3.1, peaks 7 and 8

respectively) are fully resolved and elute in the end of the analysis after 17.0

and 24.8 min respectively.

The separation of the mono-functionalized C60 fullerenes is challenging

because of the high similarity in the structures (Table A.1 in appendix A)

that results in the co-elution of the compounds in between 6 and 7.5 min

as shown in fig. 3.1 (peaks 2-5). Since the absorption spectra of the

compounds is very similar (discussed below) and because of the lack of

selectivity, UV detection alone cannot help in the determination of these

non-fully resolved peaks. If a more selective detection method such as mass

spectrometry is not available, the determination of the respective

compounds that are co-eluting must be achieved by improving the

chromatographic separation. In general, the elution time of the

functionalized C60 structures in this study is correlated with the aromatic

ring in the functionalizing group (phenyl-functionalized eluted before

thienyl-functionalised) and is inversely proportional with the length of the

alkyl chain in it. Therefore [60]PCBO and [60]PCBB (octyl ester and butyl

ester respectively) eluted before than [60]PCBM and [60]ThCBM (both

methyl esters but with different aromatic rings).

When a mobile phase composition of 75:25, toluene:acetonitrile was

applied (fig. 3.2B), the four compounds created two clusters, the first one

including [60]PCBO and [60]PCBB (Rs < 1) which was fully resolved from the

second one composed by [60]PCBM and [60]ThCBM (also Rs < 1). A better

resolution of the peaks in one of the clusters was obtained by modification

of the mobile phase composition, i.e the ratio between acetonitrile and

toluene, but resulted in a lower resolution in the other cluster. For instance,

41

increasing the percentage of acetonitrile (fig. 3.2A) and therefore the

polarity of the eluent resulted in a better separation of the compounds

based on the different aromatic rings (Rs ≥ 1 for [60]PCBM and [60]ThCBM)

but decreased the resolution between [60]PCBO and [60]PCBB. On the

contrary, increasing the percentage of toluene in the mobile phase (fig.

3.2C) allowed a better separation of the compounds depending on the alkyl

length, thus improving the resolution between [60]PCBO, [60]PCBB but

resulted in the co-elution of [60]PCBM and [60]ThCBM. In addition, the

variation in the polarity of the mobile phase affected the peak shape and

retention times of the analytes. A larger percentage of acetonitrile (fig.

3.2A) enhanced the separation of the jagged peaks of bis[60]PCBM but also

caused a slower elution of all the compounds whereas a more apolar eluent

(fig. 3.2C) leaded to a faster elution of all the compounds. These results

suggest that a complete separation of very similar structures such as the

ones included in this study might be achieved by the variation of the

physical parameters (e.g. length of the column, particles size) more than

chemical parameters such as the polarity of the mobile phase.

42

Figure 3.2. Chromatographic separation of fullerene derivatives at mobile phase composition of Toluene:acetonitrile; 65:15 (A), 75:25 (B) and 85:15 (C) (volume %). The fullerenes structures are numbered according to the caption of fig. 3.1.

When fullerenes are dissolved in organic solvents such as toluene,

spectrophotometric detection is a powerful tool for their analysis owing to

the strong absorption of these compounds in the UV range. The absorption

spectra of the functionalized fullerenes included in this study (fig. 3.3) are

43

comparable to those of the pristine C60 and C70 fullerenes from which they

are derived.

Figure 3.3. UV-vis absorption spectra of functionalized fullerenes at concentrations ranging from 1.5 to 2 mg/L.

As reported by Bouchard et al. (2008), the wavelength selected for the

detection during the chromatographic runs were 332 nm for C60, C70 and the

C60 mono-derivatives whereas the optimum for bis[60]PCBM and [70]PCBM

was found at 305 nm (fig. A.2 in appendix A), despite the fact that the

maximum absorbance for all the compounds was recorded at 286 nm. This

latter wavelength was not applied in the measurements owing to the

toluene absorbance in the same range that resulted in a greater baseline

noise. Thus, at these wavelengths selected the detector response was linear

(correlation coefficients > 0.99) over more than two orders of magnitude of

44

mass. The analytical method limits of detection (LoD) and limits of

quantification (LoQ) were assessed to be 120 pg (LoD) and 300 pg (LoQ) for

C60 and the C60 mono-derivatives respectively and 200 pg (LoD) and 480 pg

(LoQ) for C70 and its derivative. The presence of an interference peak at

retention time 4.0 min precluded an accurate detection and quantification

of bis[60]PCBM below 2 ng injected. Fluorescence emission spectra were

collected for all the compounds dissolved in toluene and the wavelengths

tested for the excitation were 286 nm, 332 nm and 463 nm, because of the

absorption of fullerenes at these values and 400 nm which had been

reported to excite C60 with consequent emission at 700 nm (Zhao et al.,

2006). C70 and [70]PCBM showed a weak fluorescence emission at 700 nm

when excited at 463 nm while the other compounds did not display any

clear emission signal at none of the excitation wavelengths tested.

Furthermore, the chromatograms collected recording the fluorescence

emission at 700 nm for the excitation wavelengths tested showed a very

high baseline noise and no clear chromatographic peak. Therefore,

fluorescence detection was not considered for further analysis in this study.

3.3.2 Application of the method

The analysis of fullerenes in environmental matrices can be problematic

because of the presence of matrix components in the extracts that can

absorb in the same range of wavelengths affecting the detection. The

properties of the soils used in the present study are reported in table 3.2

whereas the HPLC chromatograms corresponding to the analysis of the

three soil matrices spiked at 20 µg/kg are shown in figure 3.4.

45

Table 3.2. Physico-chemical properties of the soils used in the present study

(DOC, dissolved organic carbon; IC, inorganic carbon).

All soil extracts analyzed showed different but consistent matrix

interferences which are probably due to the extracted constituents of the

soil (e.g. hydrophobic fraction of the organic matter) that are not retained

in the column and eluted with the void peak in the beginning of the

chromatograms. The matrix effect was particularly evident in sandy soil

extracts (fig 3.4A) where the co-extractants eluted until 8 minutes and in

loess soil extracts (Fig 3.4B) with a number of small signals in the first 12

minutes of elution. In clay soil extracts (fig. 3.4C) the fullerenes peaks were

relatively clear (with higher S/N ratios) in comparison with the other two

matrices although this soil had the highest content in organic matter, DOC

and clay. The matrix constituents seemed not to affect the order of elution,

separation and retention times of the compounds. However, they

interfered with the detection of the fullerenes eluting in the beginning, i.e.

several of the functionalized C60s. Thus, while the last three fullerenes to

elute, [70]PCBM, C60 and C70, seemed not to be affected by any strong

interference in comparison with the chromatogram obtained running pure

standard solutions (fig. 3.1), bis[60]PCBM (Rt: ~4 min.) was not detected in

any of the soil extracts and the C60 mono-functionalized peaks were difficult

to quantify. Except for bis[60]PCBM, all the analytes could be detected (S/N

≥ 3) and quantified (S/N ≥ 10) in the extracts for all the three soils spiked at

46

20 and 10 µg/kg. When samples were spiked at 5 µg/kg, detection of the

compounds was still possible but no quantification could be made. In

general, C60 and C60 derivatives were more easy to detect than C70 and

[70]PCBM owing to the lower sensitivity of the detector for these latter

compounds and because of the slope in the baseline that affected the

determination in the end of the chromatograms. These results suggest that

the use of the optimized method is suitable for the analysis of fullerenes in

soils differing in clay and organic matter content especially for pristine

fullerenes such as C60 and C70. A sample clean-up before the injection may

help to remove the impurities in the extract that interfere with the analysis

of the fullerene derivatives in the beginning of the chromatograms.

47

Figure 3.4. Chromatograms of fullerenes extracts from sandy (A), loess (B) and clay (C) soils spiked at 20 µg/kg (blue, continues line). The red non-continues lines represent the non-spiked soils. The fullerenes structures are numbered according to the caption of fig. 3.1. Note that [70]PCBM was detected at 305 nm.

48

The recoveries of extraction for all the compounds spiked in the three soils

tested are reported in table 3.3. Several methods of extraction have already

been applied for the extraction of fullerenes from soil samples (e.g.

microwave-assisted extraction, sonication, soxhlet and accelerated solvent

extraction). However, it is not possible to establish which, among these

techniques, is the better because of the differences in the experimental

settings (e.g. kind of soil, concentrations) reported. Ultrasonication is a

robust method that was already investigated by Jehlicka et al. (2005), Vitek

et al. (2009) and Perez et al. (2013) and was applied in the present study in

combination with shaking extraction. As shown in table 3.3 the recoveries

for all other compounds were acceptable with good repeatability (n = 3)

except for bis[60]PCBM, that could not be recovered in any of the samples

because of the co-extracted interferences. The good repeteabilities (on

average less than 5% for the 20 µg/kg level and less than 6% for the 10 µg/kg

level) demonstrate that the method developed in the present study is

robust. Increasing the injection volume in the HPLC or extracting a larger

sample intake could further improve the recovery of extractions because of

the higher amount of fullerenes in the extracts and the limited interference

of the co-extractants for the other compounds except for bis[60]PCBM.

Fullerenes are expected to absorb to the soil matrix (Jehlicka et al., 2005)

and different soil components (e.g. clay minerals, organic carbon etc.) may

affect the extraction efficiency. In the present work, all analytes except for

one were recovered from the three soils to similar extents and this is

consistent with what already reported by Shareef et al. (2010) who did not

observe any real difference in the recovery from six soils tested in their

study. Statistical analysis performed (two way ANOVA) on the mean

recoveries (combining all concentrations tested) from the present study

revealed that there is no significant difference (P > 0.05) between the three

soils tested. Jehlicka et al., (2005) highlighted the role of fullerenes

concentration in the extraction efficiency of fullerenes from carbonaceous

matrices. They observed that the extraction efficiency decreased at

decreasing concentration of the C60 in soil and concluded that the possible

reasons for the reduction might be a decomposition or transformation of

the compounds and/or the absorption of fullerenes to the soil components.

This effect was not observed in the present study where the difference

between the recoveries for all the compounds at the two quantifiable

49

concentrations tested (table 3.3) is not significant at the 5% confident level

(P > 0.05).

Table 3.3. Comparison of the extraction recoveries of fullerenes from sandy, loess and clay soil at the concentrations of 20 µg/kg (left) and 10 µg/kg (right).

50

This could be explained by the fact that the concentrations in our study are

relatively similar (factor of 2 difference) in comparison with those tested by

Jehlicka et al. (factor of 10) and that the effect of the fullerenes

concentration on the extraction recovery is not appreciable in this small

range. Since the concentrations tested in the present work were lower than

those reported in the majority of previous studies, absorption and general

losses of the compounds during the sample treatments are a possible

explanation for the lower recoveries (from 47% to 71%) in comparison with

those already reported (e.g. 83-107% recovery with ASE, Shareef et al.,

2010). Recently Perez et al. (2013) reported a temperature dependency on

the recoveries of extraction for C60 and C70 spiked into soil at concentrations

similar to those tested in the present study. The spiking of the samples at

low concentrations in the present study allowed an accurate determination

of the LoDs and LoQs, estimated to be 3 µg/kg and 10 µg/kg respectively for

all the soil tested. Although more sensitive LC-MS methodologies have been

developed very recently, that allow the determination of fullerenes in

environmental matrices at even lower concentrations (Kolkman et al., 2013;

Sanchis et al 2013), the HPLC-UV method in the present study is a valid

alternative, cheaper and easier to interpret. Finally it must be noted that,

despite the fact that spiking of concentrated solutions in toluene is a

common procedure, it also represents a limitation because it could not

reproduce the real conditions at which fullerenes are present in the soil

environment and further efforts will be needed in the development of

alternative and more representative spiking techniques.

3.4 Conclusions

In this study a new chromatographic method was developed and optimized

for the analysis of fullerenes and functionalized fullerenes in soil using HPLC

with UV as method of detection. UV-detection showed very high linearity

for the compounds under investigation and allowed their detection at

concentrations as low as 6 µg/kg whereas fluorescence detection did not

fulfil the prerequisite for the analysis when coupled with HPLC in the

51

present study. This is the first time that such a number of functionalized and

non functionalized fullerenes are analyzed by HPLC-UV in a single run and

in principle, other functionalized structures, similar to those included in this

study, can be analyzed with the method. The analytical settings can be

optimized depending on the analyte(s) of interest e.g. by modification of

the mobile-phase composition or wavelength of detection.

The analysis of fullerenes including the functionalized derivatives extracted

from real soil samples spiked with the compounds, showed that the method

is robust and suitable for the determination of these compounds in complex

environmental matrices at concentrations in the range of µg/kg. The

extraction of the compounds with a combination of sonication and shaking

in two steps, with toluene as extracting solvent, is highly reproducible and

relatively efficient (from 47% to 71% recovery) and the method limit of

detection and limit of quantification (3 µg/kg and 10 µg/kg respectively)

are lower than those already reported by other works.

This method would allow the study of the fate and toxicity of fullerenes and

their functionalized derivatives in environmental samples at concentrations

close to those expected in the real environment. However, since the

predicted environmental concentration of fullerenes is expected to be in

the range of ng/kg (Gottshalk et al., 2009), further developments are

needed in order to apply the method to environmental monitoring,

especially for the functionalized structures whose UV detection in the

present study was affected by the matrix components. The current

limitations of the HPLC-UV method developed in the present study may be

overcome by applying preconcentration methods in combination with

clean up or using more sensitive but also more expensive detection

methods such as high resolution mass spectrometry.

52

53

Chapter 4

An UHPLC-HRMS method for the

analysis of fullerenes in soil

Published as:

Carboni A., Helmus R., Parsons J. R., Kalbitz K., de Voogt P. 2016. A method

for the determination of fullerenes in soil and sediment matrices using

ultra-high performance liquid chromatography coupled with heated

electrospray quadrupole time of flight mass spectrometry. Journal of

Chromatography A, 1433, 123–130.

54

Abstract

The increasing production of fullerenes likely means a release of these

chemicals in the environment. Since soils and sediments are expected to act

as a sink, analytical tools are needed to assess the presence of fullerenes in

these matrices. In the present work, a method was developed for the

determination of fullerenes at environmental relevant levels employing

Ultra High Performance Liquid Chromatograph coupled with High

Resolution Mass Spectrometry (UHPLC-HRMS). Chromatographic

separation was achieved with a core-shell biphenyl stationary phase that

provided fast analysis with complete baseline separation. Ion Booster

Electro Spray Ionization (IB–ESI) resulted in higher ionization efficiency and

was much less susceptible to adduct formation in comparison with standard

ESI, whereas Quadrupole Time of Flight (QTOF) MS granted high resolution

mass spectra used for accurate identification. The Instrumental method

limits of detection (ILoD) and quantification (ILoQ) were 6 and 20 fg,

respectively, for C60 and 12 and 39 fg, respectively, for C70. Matrix effects

related to co-extractants were systematically investigated in soil and

sediments extracts through standard addition method (SAM) and

monitoring the signal response during the chromatographic run of these

samples. Consequently, minor chromatographic modifications were

necessary for the analysis of matrices with high organic carbon content. The

method limit of detection (MLoD) ranged from 84 pg/kg to 335 pg/kg,

whereas limit of quantification (MLoQ) ranged from 279 pg/kg to 1.1 ng/kg.

Furthermore, the method was successfully applied for the analysis of

functionalized fullerenes (i.e. methanofullerenes). To the best of our

knowledge, this is the first analytical method for the analysis of fullerenes in

soils and sediments that employ core-shell biphenyl stationary phase as well

as IB-ESI-QTOF MS hyphenated with UHPLC.

55

4.1 Introduction

Since their discovery (Kroto et al., 1985), fullerenes and especially C60 have

been subject of a large number of studies mostly focused on their

innovative physical-chemical properties and possible applications.

Fullerenes are already used in cosmetics (as antioxidants) and in

photovoltaics (as electron acceptors) and possible fields of application

include electronics and optics as well as biomedical engineering

(Tagmatarchis et al., 2001; Guldi et al., 2002; Burangulov et al., 2006; Kim

et al., Maynard et al., 2006). Furthermore, the possibility of derivatization

of their structure e.g. with surface functionalization (Hummelen et al.,

1995) as well as the encapsulation of other chemical species (Chai et al.,

1991), is likely to increases the range of applications of these chemicals.

Precise data regarding the manufactured amounts are missing but the

worldwide production was estimated in tens of tons per year and is

expected to increase in the near future (Hendren et al., 2011; Piccinno et

al., 2012). This will inevitably mean a release into the environment and

there are several concerns about their possible accumulation and toxic

effects. Besides, fullerenes can also be naturally produced (e.g. in energetic

events such as volcanic eruptions (Heymann et al., 1994) or flame

generation (Howard et al., 1991)) but data are limited and often

contradictory (Elsila et al., 2005). Thus, the presence of fullerenes in the

environment may be due to both natural and anthropogenic origins as well

as resulting from transformation of engineered nanoparticles (e.g. fullerene

derivatives). In this context, fullerenes are nowadays considered as

emerging contaminants and analytical methods are needed for both

environmental and toxicological assessment. However, only in the last few

years have environmental monitoring and modeling studies been carried

out and the knowledge is still limited. Data modeled based on the estimated

production suggested that the concentrations (Predicted Environmental

Concentration, PEC) of C60 in the environment should range between parts

per billion and parts per trillion with higher concentrations in soil and

sediments than water and air (Gottschalk et al., 2009). Thus, soil may act as

a sink for the fullerenes whose source may be direct release and

atmospheric deposition but also water transport (in the form of nC60 and/or

56

associated with organic matter) and application of biosolids (Navarro et al.,

2013). In the last few years, a number of studies have been published

focused on the extraction of fullerenes and methods are now available that

allow the extraction of these chemicals in a large number of matrices with

acceptable if not excellent recoveries (e.g. Carboni et al., 2013; Kolkman et

al., 2013; Perez et al., 2013). High Performance Liquid Chromatography

(HPLC) in combination with UV-vis and/or mass spectrometry (MS) is the

most widely used technique for the determination of fullerenes and issues

related to the chromatographic separation of the fullerenes have been

recently reviewed elsewhere (Saito et al., 2004; Astefanei et al., 2014a).

Briefly, mobile phases usually consisted of toluene as a main eluent in

combination with a more polar solvent, such as acetonitrile or methanol,

whereas stationary phases commonly employed include standard octadecyl

silica (ODS) as well as functionalized silica. However, these methods are

often time consuming and seldom optimized for the analysis of complex

environmental samples (e.g. Carboni et al., 2013; Kolkman et al., 2013). In

this context, core-shell biphenyl is a novel stationary phase, never applied

for the analysis of fullerenes, which may provide a retention mechanism

similar to that of the pyrenylpropyl silica (Lomas et al., 2015). Regarding the

detection, although UV-vis showed promising performance (Wang et al.,

2010) for the determination of fullerenes in environmental matrices

(Carboni et al., 2013), its relatively low sensitivity and its lack of selectivity

in comparison with mass spectrometric detection make it unsuitable for the

analysis of the low concentrations expected in the environment. Thus, mass

spectrometry is considered as the detection method of choice and low as

well as high resolution methods have been described. Instruments

commonly employed make use of triple quadrupole (Sanchis et al., 2013;

Astefanei et al., 2014b), LTQ Orbitrap (van Wezel et al., 2011; Nunez et al.,

2012; Kolkman et al., 2013; Emke et al., 2015), and FTICR (Nunez et al.,

2012) coupled with HPLC systems and Quadrupole time of flight (QTOF) MS

with direct injection (Song et al., 2007). Atmospheric pressure ionization

(API) methods were usually employed in negative mode. Specifically, the

use of ESI in negative mode allowed the detection of fullerenes in soils and

water at relatively low concentrations (Kolkman et al., 2013; Sanchis et al.,

2013) but APPI was recently proposed as a better option due to the higher

sensitivity and less complex mass spectra (Nunez et al., 2012; Astefanei et

57

al., 2014b; Emke et al., 2015). Heated interfaces such as H-ESI were seldom

applied and showed efficiency of ionization generally higher than that of

standard ESI, but lower in comparison with other systems such as

atmospheric pressure photo (APPI) and chemical (APCI) ionizaton (Nunez et

al., 2012; Astefanei et al., 2014b). In this context, the Ion Booster-ESI (IB-

ESI) consists of a modified ESI source that was recently employed in the

screening of drugs in serum, showing better performance than standard ESI

(Huppertz et al., 2014; Kempf et al., 2014). The additional soft voltage and

vaporizer temperature are expected to enhance the ionization of the

fullerenes in comparison with standard ESI, i.e. to improve the sensitivity.

The analysis of fullerenes in the environment is challenging because of the

low concentrations expected and only recently monitoring studies have

highlighted the presence of these chemicals in environmental samples (e.g.

soil (Sanchis et al., 2013 and 2015), sediments (Sanchis et al., 2013;

Astefanei et al., 2014b), surface waters (Nunez et al., 2012; Astefanei et al.,

2014b), wastewater (Farre et al., 2010; Emke et al., 2015) and air (Sanchis

et al., 2012)). Furthermore, the complexity of some of the matrices that

have to be analyzed (e.g. WWTPs, soil) is an issue during both the extraction

and analysis of fullerenes (Jehlicka et al., 2005). The presence of co-

extractants, with special regard to organic matter content, could hamper a

correct quantification and it may have hindered the detection in other

recent works focused on soil (Perez et al., 2013) and water (van Wezel et

al., 2011). To date a few studies addressed the challenges of a clean-up

procedure (Sanchis et al., 2013) or accounted for the matrix effect during

the detection and quantification of fullerenes. Although fortification (i.e.

standard addition) of environmental samples under investigation was

recently included in monitoring studies, this issue was never systematically

approached and analytical methodologies such as matrix matched

calibration were recommended (Astefanei et al., 2014b; Sanchis et al.,

2015). Thus, in the present work we developed an analytical method that is

specifically aimed for the analysis of fullerenes in complex environmental

samples such as soil and sediment. The method included the use of a novel

stationary phase, i.e. biphenyl functionalized silica, and of a heated

interface (IB-ESI) in combination with high resolution mass spectrometry

(QTOF-MS).

58

4.2 Materials and Methods

4.2.1 Reagents and chemicals

Toluene, methanol, isopropanol and acetonitrile (Biosolve, Valkenswaard,

The Netherlands) were all analytical grade (LC-MS). The mobile phase

modifiers tested were acetic acid (Merck, Darmstadt, Germany), formic acid

(Biosolve, Valkenswaard, The Netherlands) and ammonium acetate (Sigma-

Aldrich, Zwijndrecht, The Netherlands) whereas sodium hydroxide was

purchased by Merck KGaA (Amsterdam, The Netherlands). Fullerenes C60

(CAS: 99685-96-8) and C70 (CAS: 115383-22-7) were purchased from Sigma-

Aldrich (Zwijndrecht, The Netherlands). Stock solutions of the individual

fullerenes were prepared in toluene, at a concentration of 5 mg/L, and

placed on an orbital shaker (Laboshake orbital shaker, Gerhardt,

Königswinter, Germany) in the dark overnight. Further solutions needed for

the experiments, including mixtures of the fullerenes, were obtained by

dilution of the stock solutions, stored at 4°C in the dark and sonicated for 2

min before use (Bransonic 12, Branson, Danbury CT, United states). The

analyses were performed using an UHPLC system (Nexera, Shimadzu, Den

Bosch, The Netherlands) equipped with a binary pump, autosampler and

column oven. Retention of the chemicals was achieved with a core-shell

Kinetex 2.6 µm biphenyl 100 Å chromatographic column (Phenomenex,

Utrecht, The Netherlands) consisting of a biphenyl stationary phase. The

UHPLC system was coupled to a high resolution Quadrupole-Time of Flight

mass spectrometer (Q-TOF; maXis 4G upgraded with HD collision cell,

Bruker Daltonics, Wormer, The Netherlands) equipped with either an

electrospray (ESI) or an Ion Booster electro spray (IB-ESI) ionization sources

(Bruker Daltonics, Wormer, The Netherlands). High purity nitrogen was

supplied by a N2 generator (Avilo, Dirksland, The Netherlands) and used for

ionization and collision gas.

59

4.2.2 Analytical method settings

Internal mass calibration was performed automatically during each analysis

in order to assure good mass accuracy for all the samples independently of

the total analysis time. This was achieved at the beginning of the analysis

(0.1-0.4 min) by infusing a 2 mM sodium acetate solution in a water

isopropanol mixture (1:1, v:v), with a loop injection of 20 µl and a loop rinse

of 20 µl. A temperature gradient was employed for the ionization source

because of the different optimal temperatures needed for the analysis of

the mass calibration solution and the analytes under investigation.

Specifically, the temperature was set at 325°C at the beginning of the run

and was gradually increased after one minute in order to reach 450°C at 2

min (Figure B.1 in appendix B). The sodium acetate cluster provided ten

points of calibration ranging from m/z 387 to m/z 1207 (Table B.2 in

appendix B) of which at least eight points (standard deviation ≤ 0.3 ppm)

were taken for the mass calibration of each sample. The final

chromatographic methods were optimized with regard to the methanol

(eluent A) and toluene (eluent B) amounts employed and consisted of the

following programs. Method A: flow rate of 400 µl, 40 sec at 100% eluent A

(focusing step), a linear gradient up to 50% eluent B in 0.3 min, an isocratic

step of 3.50 min at 50% eluent B (elution step) and then a linear gradient of

20 sec to reach 100% eluent B which was maintained for 2.5 min (cleaning

step). Method B: flow rate of 600 µl starting with 40 sec at 100% eluent A

(focusing step), a linear gradient up to 35% eluent B in 20 sec, an isocratic

step of 4 min at 35% eluent B followed by a linear gradient of 1.5 min to

reach 60% B (elution step), a linear gradient of 10 sec in order to reach 100%

eluent B which was maintained for 2.5 min (cleaning step). The

chromatograms were divided into four segments: segment 1 (from 0 min to

0.1 min) to assure the correct position of the loop, segment 2 (from 0.1 min

to 0.4 min) dedicated to the mass calibration, segment 3 (from 0.4 min to

3.4 min in method A and 4.4 min in method B) for the SRM analysis of C60

(m/z 720.0005) and segment 4 (from 3.4 min to 7.6 min for method A and

from 4.4 min to 9.6 min for method B) for the SRM analysis of C70 (m/z

840.0005). After the mass calibration was achieved, the LC flow was

diverted to the waste (0.5 min) in order to avoid the exposure of the

interface to the more polar co-extractants that eluted at the beginning of

the chromatogram. The LC flow was redirected to the MS (1.5 min in

60

method A and 2.5 min in method B) to allow the interface to equilibrate

before the elution of the chemicals and then switched again to waste after

the elution of the analytes (4.5 min in method A and 6.5 min in method B).

Fullerenes ESI and IB-ESI spectra were obtained by infusing both individual

and mixture standard solutions in toluene:methanol (1:1,v:v) at varying

concentrations. The optimized working conditions for the ESI and IB-ESI in

negative mode were: capillary voltage 1000 V, end plate offset -400 V,

charging voltage 300 V, nebulizer gas 4.1, dry gas 3.0 l/min and dry heater

200°C. Further settings optimized for the MS analysis were: funnel radio

frequency (RF) at 325.0 Vpp (voltage point to point), Multipole RF at 300.0

Vpp, collision cell RF at 1600.0 Vpp, transfer time at 50 µs and prepulse

storage time at 25 µs. The mass range analyzed in MS1 was 300 – 2000 m/z

whereas the spectra rate frequency was set at 2 Hz during all the analysis

performed. Collision energy for MS/MS was set at 100 V whereas the

isolation width was 8.00 m/z. The present settings allowed a mass resolving

power up to 80000 (0.03 m/z FWHM). The [M]-•and [M+1]-• abundances

relative to C60 and C70 were determined by analysing a standard solution at

500 ng/L repeatedly with both method A (n=7) and method B (n=7). More

details about the molecular ions and isotopic clusters during the detection

are provided in the discussion section. The Bruker Compass 1.7 software

was employed in both the data collection and data processing.

4.2.3 Samples collection and treatment

In table 4.1, the properties of the soil and sediment samples included in the

present study are reported.

61

Table 4.1. Properties of the soil and sediment samples included in the present study.

n.d. not detected

The Loess soil (matrix 1) was collected from an agricultural field in South

Limburg, The Netherlands (50°53'58"N, 5°53'16"E). Matrix 2 consisted of an

urban park soil sample collected in Vondelpark, Amsterdam, The

Netherlands (52.3580° N, 4.8680° E). River clay and sea clay (matrix 3 and

matrix 4 respectively) were available from the collection of the Earth

Surface Science research group at the University of Amsterdam. The sandy

top soil rich in organic carbon (matrix 5) was collected in Flevopark,

Amsterdam, The Netherlands (52.3611° N, 4.9492° E) whereas the dune

sand (matrix 6) was sampled in a natural park near Castricum, The

Netherlands (52.5500° N, 4.6667° E). The samples were placed in a freezer

at -20°C overnight and lyophilized with a Scanvac Coolsafe freeze-dryer

(Labogene, Lynge, Denmark) in order to remove traces of water. The dried

samples were finely ground with an agate mortar and sieved. The total

carbon, sulfur and nitrogen contents of the dried soil samples were

measured using a Vario EL Cube (Elementar, Hanau, Germany). The

inorganic carbon content was determined as reported by Wesemael (1955)

as follows: 1 g of soil was weighed into a 250 ml Erlenmeyer provided with

a silica gel lid and an excess of HCl 4 M was added. Control samples

consisted of 0.250 g of CaCO3. The samples were weighed again after 24 h

of shaking at 60 rpm (Laboshake orbital shaker, Gerhardt, Königswinter,

Germany) and the inorganic carbon content was calculated accordingly. The

soil and sediment samples were extracted with the protocol already

62

reported by our group (Carboni et al., 2013) with minor modifications, i.e.

the temperature was set at 20 ± 2 °C as recommended by Perez et al. (Perez

et al., 2013). Briefly, 10 to 15 g of each sample underwent two cycles of

extraction with toluene with each cycle consisting of a combination of

sonication for 20 min with a ultrasonic bath (Bransonic 12, Branson,

Danbury CT, United states) and shaking for 2 h (Laboshake orbital shaker,

Gerhardt, Königswinter, Germany). The extracts were concentrated

evaporating the solvent under a gentle nitrogen flow (enrichment factor

20x) and stored in a refrigerator at 4 ºC in the dark. A repeatability test for

the extraction procedure was achieved by spiking and extracting the

fullerenes into matrix 6 at the concentration of 100 ng/kg (n=8) as reported

by Carboni et al. (2013).

4.2.4 Sample analysis

Methanol (25% in volume) was added to each sample and the analyses were

performed with an injection volume of 10 µl. External calibration curves

were obtained by analyzing standard solutions at concentrations ranging

from 1 to 500 ng/L and quantification was based on chromatographic peak

areas. Instrumental method limits of quantifications (ILOQs) were assumed

as the lowest measured concentration in the linear range with a deviation

less than 30% of the theoretical concentration injected (EC commission,

2009). The instrumental method limit of detection (ILoD) was calculated as

3/10 of the ILoQ observed. The method limits of detection (MLoDs) and

quantification (MLoQs) in soil and sediment samples were extrapolated

from the signal-to-noise ratio observed in the chromatograms assuming a

70% recovery of extraction (Carboni et al., 2013). In case of absence of

matrix effect and noise, these values were assumed to be equal to the ILoD

and ILoQ. The criteria used for the identification of the analytes were: 1) the

expected chromatographic retention time (± 0.2 min), 2) a mass accuracy

threshold (≤ 5 ppm) and 3) an isotopic pattern fit threshold (≤ 50 mSigma),

where mSigma represents the goodness of fit (the smaller the better)

between the measured and theoretical isotopic pattern (Gago-Ferrero et

al., 2015). For the standard addition method (SAM) experiments, each

63

matrix was spiked with both C60 and C70 fullerenes at increasing

concentrations (25, 50, 100, 250 and 500 ng/L) in order to obtain

concentration versus response curves (fig. B.3 and B.4 in appendix B).

Standard solutions at the same concentration were also analyzed to allow

a comparison in terms of linearity, slope and intercept. In addition, the ion

suppression/enhancement due to the matrix effect was evaluated in terms

of response recovery by comparing the peak areas obtained for C60 and C70

spiked into the extracts with those obtained analyzing standard solutions at

the same concentrations. The matrix effect (ME) was further investigated

by continuously infusing a fullerenes standard (1 µg/L in toluene) into the

MS source while injecting the matrices extracts. This allowed us to monitor

the detector response during the chromatographic run, i.e. the matrix

suppression/enhancement during the analysis. For this purpose a 5 ml glass

chromatographic syringe and syringe pumps (KD Scientific, Holliston MA,

United States) operating at 0.2 ml/h were employed.

4.3 Results and discussion

4.3.1 UHPLC-HRMS

The first aim of the present work was to assess the performance of the

biphenyl stationary phase selected for the separation of the fullerenes.

When method A was applied, C60 and C70 eluted at 3.1 min and 3.8 min

respectively (fig. 4.1) with fully resolved (Rs>1.5) and highly symmetric

chromatographic peaks.

64

Fig. 4.1. Chromatographic separation of C60 (left) and C70 (right) standard at

500 ng/L. The continuous and dashed lines represent the MS1 and MS2

chromatograms respectively.

Similar to what is observed with columns that are specifically designed for

the separation of fullerenes, e.g. the pyrenylpropyl functionalized silica

(Buckyprep), with biphenyl groups the main interactions are expected to be

the pi-pi-interactions and pi-stacking between the aromatic rings in the

fullerenes and those of the stationary phase. However, as suggested by

Nunez et al. (2012) the size of the buckyballs may play a role in the retention

mechanism. The biphenyl stationary phase was an optimal compromise for

selectivity and rapidity when compared with other materials that were

usually employed providing higher retention than the standard octadecyl

silica and shorter analysis time in comparison with the pyrenylpropyl

functionalized silica. It must be noticed that the analysis time may be

shortened when using a shorter column with smaller particle size (i.e. 1.7

µm) that is currently available on the market. The “focusing step” i.e. the

application of a 100% methanol at the beginning of the chromatographic

run, resulted in a better peak shape and sensitivity and may be explained

with the fact that the analytes accumulated at the start of the column

(focused) until a stronger solvent was provided. The use of ESI-HRMS was

already investigated by some of the authors of the present work (Kolkman

65

et al., 2013). In those studies, a consistent formation of adducts during the

ionization process was reported, consisting of both oxidized products ([M +

O]-•, [M + OH] -•) and methanol and/or toluene adducts (mostly [M + OCH3]-

• and [M + C7H7O2]-•). In contrast, IB-ESI-HRMS provided mass spectra that

were dominated by the isotope cluster of the molecular ions with an only

minor abundance of [M+16]-• adducts as shown in figure 4.2. Similar to what

has already been reported in other studies, the method was more sensitive

for C60 than C70 (a factor of 2) at all the concentrations tested. Furthermore,

in the present study the IB-ESI signal response was several orders of

magnitudes higher than that of ESI stand-alone for both C60 and C70 (data

not shown). Thus, ESI was not taken into consideration for further analysis.

In the present work, sodium acetate was selected for internal mass

calibration because it can stand higher temperatures in comparison with

other commonly applied calibration solutions (e.g. sodium formate). This

allowed the determination of the accurate masses of fullerenes with errors

lower than 5 ppm. In comparison with APPI, where toluene was used as a

dopant to improve the ionization efficiency, ESI analysis required the

presence of methanol during ionization that can be added post-column as

already described (Kolkman et al., 2013). The post column addition was not

needed in the present work since methanol was one of the constituents in

the mobile phase during the elution of the chemicals. It must be noted that

fragmentation of the fullerenes could not be obtained even at extremely

high collision energies (> 100 eV). Thus, although multiple reaction

monitoring (MRM) mode was applied in the present work, the MS2 analysis

corresponded to [M]-•[M]-• transitions (pseudo-MRM) and the MS2

spectra resembled the MS1 ones (fig. 4.2b and 4.2e) in which adducts and

background masses (e.g. m/z 731.4124 in fig 4.2a) were removed.

66

Fig. 4.2. (IB)ESI-QTOF mass spectra of C60 and C70 MS (a and d respectively) and MS2 (b and e respectively). In fig. 4.2c (C60) and 4.2f (C70) a comparison between measured (top) and expected (bottom) m/z clusters.

Nunez et al. (2012) reported higher than expected relative abundances of

the isotopic cluster ions when using APPI with special regard to the m/z

[M+1]-• and an enhanced effect on larger fullerenes (Nunez et al., 2012). In

their work they demonstrated that the addition of hydrogenated products

to the peaks corresponding to the 13C natural abundance may account for

the phenomenon. Furthermore, Emke et al. (2015) showed that the

unexpected isotopic distribution was due to the presence of methanol

during the ionization, a phenomenon not observed when toluene was the

only solvent employed. If highly resolving systems, such as FTICR MS, are

not available, anomalous isotopic abundances may hinder a qualitative

detection of fullerenes. Especially when analyzing non-functionalized

structures, such as C60 and C70, the lack of MS/MS fragmentation implies

that the main identification points must be provided by the accurate mass

and cluster distribution. In the present work the C60 spectra presented

67

molecular ions at m/z [M]-•, [M+1]-•, [M+2]-• and [M+3]-• consistent with the

theoretical isotopic patterns (fig. 4.1c) and those obtained by ESI, whereas

C70 showed slightly higher than expected m/z [M+1]-• abundance in MS1 fig

(4.2a and 4.2f). In detail, C60 presented a [M+1]-•/ [M]-• ratio of 0.64 ± 0.05

(versus the expected value of 0.65) whereas C70 presented a ratio of 0.77 ±

0.03 (versus the 0.74 expected).

Identification points in the present work were provided by the

chromatographic retention times and the accurate mass detection and

isotopic clusters in both MS1 and MS2. Calibration curves, obtained

analyzing standard solutions of the fullerenes, showed high linearity

(R2>0.998) and the ILoD and ILoQ were 0.6 and 2 ng/L, respectively, for C60

and 1.2 and 4 ng/L, respectively, for C70. These results indicate that the IB-

ESI-HRMS system allowed the determination of fullerenes at concentrations

lower than those of other ESI-MS and heated ESI methods applied (van

Wezel et al., 2011) and similar to those of APPI-MS methods recently

developed (Astefanei et al., 2014b). Finally, the method was also tested for

the determination of fullerene derivatives to assess the possibility to

include functionalized structures investigated in our previous works

(Kolkman et al., 2013; Carboni et al., 2013). The method was suitable for the

analysis of the methanofullerenes that were completely resolved (Rs > 1.5)

from the related non functionalized structures at the beginning of the

chromatogram (Fig. B.5 in appendix B).

4.3.2 Matrix effects

The method we developed was tested for the determination of fullerenes

spiked into extracts of soils and sediments. Soil and sediment samples were

chosen in order to represent a range of possible environmental matrices

and textures (e.g. clay, sand) and the selection included samples with

varying inorganic and organic carbon contents whereas nitrogen and sulfur

concentrations were similar between the samples (table 4.1). After the

extraction, the sample extracts were spiked with fullerenes in order to

investigate the analysis of the chemicals in presence of the co-extractants.

The use of fullerene standards in organic solvent for the spiking of samples

has often been the subject of debate since it is not expected to represent

68

the real conditions at which the chemicals are present in the environment

(Shareef et al., 2010; Carboni et al., 2013). However, in the present work,

standard solutions in toluene were directly added to the toluene extracts,

thus the results hereby presented will help in understanding the behavior

of fullerenes that are already present in the extracts with no effect of the

extraction procedure applied to the original sample. According to a previous

study, the texture of the soil samples should be of a minor concern when

analyzing fullerenes (Carboni et al., 2013). On the contrary, the presence of

carbon and specifically organic matter in the samples is expected to play a

major role, e.g. by hindering the detection and quantification of the

chemicals (matrix effects, ME). The ME may either cause the enhancement

or the decrease of the method’s accuracy and sensitivity. It must be noted

that during the sample preparation, an addition of methanol at percentages

higher than 25% of the final volume caused precipitation in the extracts

with the highest carbon content (matrix 2). This was due to the large

content of non-polar co-extractants resulting from the extraction with

100% toluene. Therefore, although the initial optimization was achieved

with a toluene:methanol ratio of 1:1 (v:v), the final composition of each

sample injected into the UHPLC was a toluene:methanol ratio of 3:1 (v:v).

The high collision energy values (100 V) applied during the analysis led to a

“Background clean-up”, i.e. the degradation of the co-extractants, with

resulting extremely clean MS2 spectra. This allowed the unambiguous

determination of the fullerenes in all the samples, even at concentrations

as low as 25 ng/L (fig. 4.3).

69

Fig. 4.3. MS1 (top) and MS2 (bottom) spectra of C60 spiked at the concentration of 25 ng/L in presence of high organic carbon content (matrix 2).

The results of the SAM experiments are reported in Table 4.2 and more

comprehensively in Appendix B (Table B.6). The response was linear (R2 >

0.99) for both C60 and C70 in all the matrices tested (with the only exception

of C60 in matrix 5 when analyzed in MS2), and no significant difference

(P>0.05, F-test for covariances) was observed between MS1 and MS2

measurements in the same matrix. It should be noted that, although

relative small intercepts were observed, these were not significantly

different from zero (P>0.05).

70

Table 4.2. Results of the standard addition experiment for C60 and C70 with the use of the method A (left) and method B (right)a. For matrix descriptions, see section 2.3

a r2 = regression coefficient of the standard addition lines; Rec% = recovery; Matrix 2 dil 10x = matrix 2 diluted ten times; method A and method B differ in the mobile phase composition during the isocratic step (50:50,MeOH:Tol and 65:35,MeOH, respectively) and flow rate (400 and 600 ul/min, respectively).

However, a significant difference (P<0.05) was present when comparing the

response recovery of different samples. Indeed, although the majority of

the matrices showed recoveries ranging from 100% to 122%, i.e. featuring

either no ME or a slight enhancement of the response, a large signal

suppression (specifically 6% recovery for C60 and 29% recovery for C70) was

found in the matrix with the highest carbon content (matrix 2). This effect

was not observed in other matrices with a high organic carbon content

(matrix 4 and matrix 5) indicating that the percentage of organic matter in

the samples was not the only property determining the ion suppression and

that other feature such as its chemical composition may play a role.

However, when the extract from matrix 2 was diluted ten times (matrix 2

dil 10x in table 4.2), an enhancement of the response was found, which is

consistent with that observed in matrices that have a lower organic carbon

content. This suggests that although the chemical composition of the co-

71

extractants played a role in the ion suppression, their concentration was

likely the main reason for the response loss observed in the extract of

matrix 2. The MLoDs were extrapolated assuming a 70% recovery of

extraction from the matrices (see materials and methods) and ranged from

84 pg/kg for C60 to 168 pg/kg for C70 whereas MLoQs ranged from 279 pg/kg

for C60 to 559 pg/kg for C70 (method A). However, in the case of matrix 2,

which showed a strong matrix effect, these values were a factor of 30 higher

for C60 and a factor of 5 higher for C70. The smaller suppression of the C70

signal in comparison with C60 may be explained by the fact that, being more

retained during the analysis, C70 eluted in a position of the chromatogram

that was less affected by the more polar co-extractants. The matrix effect

due to the co-extractants in the samples was monitored by injecting the

matrices while infusing the fullerenes in the MS system. The signal

suppression was due to the more polar fraction of the co-extractants,

eluting at the beginning of the chromatographic run where a higher

percentage of methanol was present.

Fig. 4.4. Response of C60 post-column infused directly into the IB-ESI-MS during chromatographic runs of the soil or sediments extracts using a gradient elution of methanol:toluene (method A). The black line represents a pure solvent injection in comparison with matrix 1 (light blue), matrix 2 (red), matrix 3 (orange), matrix 4 (yellow), matrix 5 (blue) and matrix 6 (green). The start of section b corresponds to the moment the gradient reaches 100% toluene, at the start of c the gradient program returns to initial solvent composition (see 6.2.2).

72

In particular, as shown in Figure 4.4, the matrix 2 (red line) suffered a large

signal suppression at the beginning of the analysis (Fig. 4.4a) whereas

signals from other matrices with a high organic carbon content were not

suppressed (matrix 4 in yellow) or were affected to a lesser extent (matrix

5 blue line). Flushing with toluene (100%) (Fig. 4.4b), in order to remove the

more non-polar fraction that may remain in the column and affect the next

injections, ultimately restored the response to the maximum levels as

shown in Figure 4.4c. During the flushing stage of the chromatogram (b),

however, no signal was present because of the absence of any methanol

and consequent lack of ionization of the fullerenes (Fig. 4.4b) in the ESI.

Matrices 2 and 5 were selected for further experiments aimed to improve

the analytical method for the determination of fullerenes in complex

matrices with mitigation of the ion suppression. Three mobile phase

modifiers: acetic acid, ammonium acetate and formic acid, were tested at

varying concentrations with the purpose of attenuating the ion

suppression. However, since the addition of formic acid resulted in a near

or complete loss of the response whereas both acetic acid and ammonium

acetate suppressed the signal to a large extent, these modifiers were not

taken into account for further analysis. The large suppression may be due

to the susceptibility of the ESI to buffer salts in comparison with APPI as

mentioned by Nunez et al. (2012). Also the modification of the mobile-

phase composition with increasing amounts of isopropanol and acetonitrile

did not lead to any improvement in recovering the response and, in the case

of acetonitrile, caused the formation of adducts at the m/z [M+40]-• for

both C60 and C70. Subsequently, the modification of the mobile phase

composition was tested with regards to the methanol:toluene ratio. In

detail, weakening the eluotropicity, i.e. increasing the amount of methanol,

delayed the elution of the fullerenes and enhanced the separation of

interfering compounds in the biphenyl column, with a general improvement

of the signal response (Fig. B.7 in appendix B). Optimal conditions were

found at 65:35 (methanol:toluene, v:v) whereas further addition of

methanol (e.g. 70:30, v:v) increased the retention time of the fullerenes and

broadened the peak, with no further enhancement of the response. Finally,

the original method (method A) was modified accordingly by changing the

methanol:toluene ratio in the isocratic step from 50:50 (method A) to 65:35

(method B). Further optimization of the methods consisted in the addition

73

of a washing step (100% toluene) at the end of the analysis and to divert

the LC flow to the waste in order to prevent the exposure of the source to

the co-extractants at the beginning of the analysis and provide higher

sensitivity and better peak shape of the fullerenes. In the resulting method

B, the retention times of the fullerenes were delayed, with C60 and C70

eluting at 4 and 6 min respectively (Fig. B.8 in appendix B) and the ILOD and

ILOQ were 12 and 39 fg, respectively, for C60 and 23 and 78 fg, respectively,

for C70. When method B was applied, a considerable improvement in the

response recovery of C60 was observed for the matrix 2 (from 6% to 67%)

and minor but consistent improvements were obtained for C70 (from 29 to

45%), as reported in Table 4.2. This corresponded to a MLoD and MLoQ of

168 pg/kg and 559 ng/kg, respectively, for C60 and 335 pg/kg and 1.1 ng/kg,

respectively, for C70. Furthermore, no significant difference (P<0.05) was

found when comparing the performances of the two methods for the

analysis of the other matrices. Eventually, we recommend the use of

method B (10 min) for monitoring studies, since it can provide a more

reliable measurement of the fullerenes in a wider range of matrices. In

particular, method B should be employed in the analysis of soils and

sediments as well as wastewater samples that can consist of complex

matrices possessing high organic carbon contents. However, method A can

provide a faster (8 min) and more sensitive analysis for the determination

of fullerenes in standard solutions and less complex samples such as sandy

soils with low organic carbon contents.

4.4 Conclusions

In the present work we developed a sensitive and fast method for the

determination of fullerenes in soils and sediments. The biphenyl-coated

stationary LC phase provides an adequate retention and represents a valid

alternative to other stationary phase materials currently applied with

promising perspectives for the analysis of complex mixtures including

functionalized fullerenes. We also showed how heated ESI, i.e. the IB-ESI

employed in the present work, can grant an ionization efficiency higher than

74

that of standard ESI and comparable to that of other recently developed

APPI methods, with the advantage of producing isotopic patterns that

resemble the theoretical ones. In general, the high resolution and the

clarity of the mass spectra allowed an unambiguous determination of the

fullerenes in all the samples under investigation, without the need of an

additional clean up, and at low concentrations that are environmentally

relevant. The matrix effect due to the presence of co-extractants was

investigated and the method is robust and flexible with regard to the

analysis of very complex matrices such as soils having high organic carbon

contents. However, the matrix effect and specifically the ion

suppression/enhancement during the analysis remains an issue and the use

of internal standards (e.g. isotopically labeled), if available, as well as matrix

matched calibration are recommended for a more precise determination.

Due to its favorable qualitative and quantitative features, the method

developed in the present study is a valid tool for the monitoring of

fullerenes in soils and sediments and for the study of the fate of these novel

contaminants at environmentally relevant concentrations.

75

Chapter 5

Analysis of fullerenes in soils

from The Netherlands

Submitted to Environmental Pollution as:

Carboni A., Helmus R, Emke E., van den Brink N., Parsons J. R., Kalbitz K. and

de Voogt P. Analysis of fullerenes in soils samples collected in The

Netherlands. Environmental Pollution (in revision).

76

Abstract

Fullerenes are carbon based nanoparticles that may enter the environment

as a consequence of both natural processes and human activities. Although

little is known about the presence of these chemicals in the environment,

recent studies suggested that soil may act as a sink. The aim of the present

work was to investigate the presence of fullerenes in soils collected in The

Netherlands. Samples (n=91) were taken from 6 locations, and included

highly trafficked and industrialized as well as urban and natural areas and

analyzed using a LC-QTOF-MS method. In general, C60 was the most

abundant species found in the environment and detected in almost a half of

the samples, at concentrations in the range of ng/kg. Other fullerenes such

as C70 and an unknown structure containing a C60 cage were detected to a

lower extent. The highest concentrations were found in the proximity of

combustion sites such as a coal power plant and an incinerator, suggesting

that the nanoparticles were unintentionally produced during combustions

processes and reached the soil through atmospheric deposition. Consistent

with other recent studies, these results show that fullerenes are widely

present in the environment and that the main route for their entrance may

be due to human activities. These data will be helpful in the understanding

the distribution of fullerenes in the environment and for the study of their

behaviour and fate in soil.

77

5.1 Introduction

Fullerenes are carbon based nanomaterials widely researched due to their

unique properties that make them useful for a large number of applications,

including electronics and optics as well as medicine and personal care

products (Murayama et al., 2004, Mauter et al., 2008, Osawa et al., 2002).

Moreover, the possibility to derivatize the closed cage carbon structure,

with consequent production of novel nanomaterials, widens the range of

uses of these chemicals (Hummelen et al., 1995). Although the annual

production is estimated to be tens of tons per year, a large increase is

expected in the near future (Hendren et al., 2011; Piccinno et al. 2012),

which is likely to imply higher emissions into the environment. Thus,

fullerenes are nowadays referred to as emerging contaminants and there is

concern about the environmental and health implications that may arise. In

particular, fullerenes may be released directly during production or they

may leach from materials as a consequence of their use and disposal.

Furthermore, they may be produced in combustion processes and

therefore associated with both natural phenomena (e.g. forest fires,

volcanic eruptions; Howard et al., 1991; Heymann et al., 1994) and human

activities such as industrial processes (e.g. as a byproduct during

combustion of carbonaceous materials) and transport (e.g. fuel

combustion; Murr et al., 2004). Little is known about the fate of these

chemicals once they end up in the environment. Although some studies

suggested that fullerenes may be degraded due to both biotic and abiotic

processes (e.g. Avanasi et al., 2014; Tiwari et al., 2014), others highlighted

the stability of C60 that may result in their accumulation in the environment

(Jehlicka et al., 2000; Parthasarathy et al., 2008). However, fullerenes have

been shown to affect soil organisms (e.g. earthworms, van der Plog et al.,

2011) and monitoring studies are crucial in order to provide valuable

information about the occurrence of these chemicals and to determine the

potential environmental risk associated with them. In general, pristine

structures such as C60 and C70 are expected to be more abundant in the

environment because of the higher production volumes, their natural

occurrence and formation resulting from the degradation of functionalized

structures. Indeed, several studies reported the presence C60 and C70

78

fullerenes in environmental matrices (Sanchis et al., 2012, 2013 and 2015;

Nunez et al., 2012; Astefanei et al., 2014b), whereas functionalized

structures (i.e. engineered nanomaterials) were seldom detected (Astefanei

et al., 2014b). Soil has been predicted to be a sink for fullerenes (Gottschalk

et al., 2009) that may enter the terrestrial environment via atmospheric

deposition, direct release, water transport and amendment of soils (Sanchis

et al., 2012; Navarro et al., 2013). However, most of the research so far has

been focused on water and wastewater samples (van Wezel et al., 2009;

Farré et al., 2011; Nunez et al., 2012; Kolkman et al., 2013; Astefanei et al.,

2014b; Emke et al., 2015) and data regarding the occurrence of fullerenes

in soil are limited to a few studies. In particular, Sanchis et al. (2013)

reported the presence of fullerenes in soils from Saudi Arabia associated

with the petroleum refinery activities in the area. In their study, C60 was the

only compound detected, and was found in 19% of the samples at

concentrations in the low µg/kg range. Fullerenes were detected to a larger

extent (67.6% of the agricultural soils and 91% of the urban soils) during a

monitoring study in Brazil where both C60 and C70 were found at

concentrations up to 154 ng/kg (Sanchis et al., 2015). In this case, their

presence was also correlated to combustion activities, particularly the

deposition of atmospheric particulate resulting from combustion processes.

Fullerenes were also found in aerosol particulates (Sanchis et al., 2011) and

urban atmosphere (Laitinen et al., 2014) as well as resulting from coal

combustion (Utsunomiya et al., 2002), suggesting that atmospheric

deposition could represent a main route for their entrance in the terrestrial

environment. In the present work, an environmental survey was carried out

with the aim of investigating the presence of fullerenes in soil samples

collected in The Netherlands. For this purpose, a method employing Ultra

High Performance Liquid Chromatography (UHPLC) coupled with High

Resolution Mass Spectrometric (HRMS) detection was used (Carboni et al.,

2016). The main goal was to investigate the presence of C60 and C70 in highly

trafficked and industrialized areas, i.e. those areas where the generation of

fullerenes may be related to human activities, in order to assess the

influence that these possible sources may have in the occurrence of these

chemicals in the environment. The selection of the samples was done in

order to represent diverse situations. Thus, soil samples were collected in

79

proximity of expected sources such as a coal power plant (Utsunomiya et

al., 2002) as well as in uncontaminated areas (i.e. natural park).

5.2 Materials and Methods

5.2.1 Reagents and chemicals

Toluene, methanol, isopropanol and acetonitrile (Biosolve, Valkenswaard,

The Netherlands) were all analytical grade (LC-MS). Fullerenes C60 (CAS:

99685-96-8) and C70 (CAS: 115383-22-7) standards were purchased by

Sigma-Aldrich (Steinheim, Germany). Stock solutions of the individual

fullerenes, at the concentration of 5 mg/L, were prepared in toluene and

placed on an orbital shaker (Laboshake orbital shaker, Gerhardt,

Königswinter, Germany) in the dark overnight. Further solutions needed for

the experiments, were obtained by dilution of the stock solutions, and

stored at 4°C in the dark and sonicated for 2 min before use (Bransonic 12,

Branson, Danbury CT, United States).

5.2.2. Sample collection and treatment

The surface soil samples (top 5 cm, at least 125 g each) were collected in

250 ml amber glass bottles. Each sample consisted of the composite of five

sub-samples (25-30 g each) collected at the corners and centre of a 1 m2

squared area. A total of 91 surface soils were sampled in 6 locations in The

Netherlands:

Location A was an urban area in proximity of a coal power plant in

Amsterdam (15 samples, labelled A_01 - A_15); Location B was a rural and

urban area nearby the AVR incinerator in Duiven (16 samples, labelled B_01

– B_16); Location C was an area surrounding the Amsterdam’s “Ring”

highway (26 samples, labelled C_01 – C_25); Location D was in green areas

(e.g. parks, flower beds) in the city of Amsterdam (16 samples, labelled

D_01 – D_16); Location E was a natural park situated near Castricum (6

80

samples, labelled E_01 – E_06) whereas Location F samples were collected

in proximity of a runway at the Eindhoven airport (12 samples, labelled F_01

– F_12). Additional information regarding the distance of the sampling

locations from the expected sources and their coordinates are reported in

the appendix C (Table C.1). The wind direction was assumed as the main

wind direction recorded during the three days before the sampling was

performed. The samples were placed in a freezer at -20 °C overnight and

lyophilized with a Scanvac Coolsafe freeze-dryer (Labogene, Lynge,

Denmark) in order to remove traces of water. The dried samples were

ground and homogenized with an agate mortar and sieved before

extraction with the protocol already reported by our group (Carboni et al.

2013) with minor modifications, i.e. the temperature was set at 20 ± 2 °C as

recommended by Perez et al. (Perez et al., 2013) and the extraction was

performed in 40 ml custom made glass vial in order to allow the extraction

of larger volumes. Briefly, circa 25 g of each sample underwent two cycles

of extraction with toluene with each cycle consisting of a combination of

sonication for 20 min with a Bransonic 12 ultrasonic bath (Branson, Danbury

CT, United states) and shaking for 2 h (Laboshake orbital shaker, Gerhardt,

Königswinter, Germany). Negative controls consisted of quartz sand

samples (25 g each) extracted and analyzed with the same procedure. The

extracts were stored in a refrigerator at 4 °C in the dark and sonicated for

30 sec before to be analyzed.

5.2.3 Sample analysis

The analyses were performed with the method recently developed in our

research group for the identification of fullerenes in soil and sediment

samples as described previously (Carboni et al., 2016), using an UHPLC

system (Nexera, Shimadzu, Den Bosch, The Netherlands) equipped with a

binary pump, autosampler and column oven. Retention of the chemicals

was achieved with a core-shell Kinetex 2.6 µm biphenyl 100 Å

chromatographic column (Phenomenex, Utrecht, the Netherlands)

consisting of a biphenyl stationary phase. The UHPLC system was coupled

to a high resolution Quadrupole-Time of Flight mass spectrometer (Q-TOF;

maXis 4G, Bruker Daltonics, Wormer, The Netherlands) equipped with an

Ion Booster electro spray (IB-ESI) ionization source (Bruker Daltonics,

81

Wormer, The Netherlands) as described elsewhere (Carboni et al. 2016).

High purity nitrogen was used as collision gas (Avilo, Dirksland, The

Netherlands). Automatic mass calibration was performed for each sample

by injecting a 2 mM sodium acetate solution at the beginning of the

chromatographic run. Additional information regarding the analytical set up

and operational parameters are reported in the appendix C.2. In addition to

the pristine fullerenes structures C60 and C70, the samples were screened for

the presence of six methanofullerenes, such as [60]PCBM and [60]bisPCBM,

as reported in Carboni et al. (2016). The criteria used for the identification

of the analytes were: 1) the expected chromatographic retention time (±

0.2 min), 2) a mass accuracy threshold (≤ 5 ppm) and 3) an isotopic pattern

fit threshold (≤ 50 mSigma), where mSigma represents the goodness of fit

(the smaller the better) between the measured and theoretical isotopic

pattern in both MS and MS/MS modes (Carboni et al., 2016). The analyses

were performed with an injection volume of 10 µl after adding methanol

(25% in volume) to each sample prior to injection. External calibration

curves were obtained after analyzing standard solutions at concentrations

ranging from 500 ng/L to 1 ng/L and quantification was based on

chromatographic peak areas. The Bruker Compass 1.7 software was

employed in both the data collection and data processing. The method

allowed the determination of C60 and C70 fullerenes in the samples with

method limit of detection (MLOD) of 67 and 134 pg/kg for C60 and C70,

respectively, and limits of quantification (MLOQ) of 223 and 446 pg/kg for

C60 and C70, respectively. The matrix effect, i.e. the suppression or

enhancement of the signal due to the presence of co-extractants in the

samples, was investigated by the addition of a C60 and C70 standard to the

extracts at the concentration of 25 ng/l (appendix C.3). In order to study the

unknown C60-like fullerene, a larger amount (100 g) of the samples from the

incinerator area (location B) were extracted to increase its concentration.

These samples were additionally analyzed with a modified UHPLC-IB-ESI-

QTOF method (appendix C.2). For statistical analysis, a linear model was

used that considered the relationship between the concentration of the

samples and the distance from the possible source. Further analyses were

conducted where the sample location with regard to the possible source

(i.e. upwind or downwind) were taken into account.

82

5.3 Results

The high accuracy (< 5 ppm), resolution (80000 FWHM) and clarity of the

MS/MS spectra allowed the unambiguous determination of the fullerenes

in the soil samples. The UHPLC chromatograms of the soil extracts (fig. 5.1b)

resembled those obtained with standard solutions (Fig. 5.1a) with slightly

higher background noise at the beginning of the chromatograms. Fullerenes

were found to be present in almost half of the soil samples analyzed and C60

was the most detected specie (48% of the samples). C70 fullerene was

detected less frequently (7% of the samples) and mostly associated with the

presence of C60 whereas methanofullerenes structures (e.g. [60]PCBM])

were not detected in any of the samples under investigation.

An additional C60 peak was found in some of the samples (fig. 5.1c) that

showed mass, isotopic pattern and MS/MS spectra similar to those of the

C60 standard. This suggested the presence of other fullerenes species in the

extracts. Since the chemical structure has not been unequivocally

elucidated, we will refer to this compound as “unknown”. Specifically, the

unknown was found only in samples collected in the incinerator area and

only if C60 was also present. The unknown had an accurate mass

corresponding to C60 but eluted earlier in UHPLC (at a retention time of 3.1

min) suggesting the presence of a C60-like species (e.g. a functionalized

fullerene) that, although separated chromatographically, underwent in-

source fragmentation during the ionization process (i.e. the loss of the

external functionalization) with formation of pristine C60. As shown in figure

5.1c, the addition of C60 standard to the original sample resulted in the

enhancement of the C60 peak area (at retention time 4 min) but had no

effect to the additional peak (at retention time 3.1 min), confirming that the

peak at 3.1 min is due to the presence of another C60-like species in the

samples. Modification of the analytical method with regard to the

chromatography as well as the mass spectrometric method (i.e. by

decreasing the temperature and/or the capillary voltage) were not able to

prevent the fragmentation of the unknown and had the only effect of

reducing the ionization efficiency. Eventually, the extraction of a larger

amount of samples, i.e. the analysis of more concentrated extracts, did not

lead to the identification of the unknown and did not show any presence of

possible precursors.

83

Fig. 5.1. Chromatographic peak and mass spectra of A) a C60 standard at concentration 8 ng/L, B) a soil sample extract from the highway (location C) and C) a sample from the incinerator area (location B) presenting an additional C60 peak at retention time 3.1 min. The black line represents the signal of the sample spiked with standard.

84

Samples collected close to the coal power plant (location A) revealed the

presence of fullerenes in 53% of the extracts. As shown in figure 5.2, C60 was

the compound detected most often at concentrations ranging from 0.20 to

1.44 ng/kg whereas C70 was found at similar concentrations but only in two

of the samples. A larger occurrence was found downwind of the powerplant

(samples A_7-A_15) at distance up to 1.8 km from the chimney whereas no

fullerenes were detected upwind in proximity of the powerplant (samples

A_1-A_3). However, the highest concentrations of both C60 and C70 was

found close to a coal storage site (sample A_6).

In the samples collected in the incinerator area (Location B), C60 was

detected in the majority of the samples (69%) at concentrations ranging

from 0.15 to 3.14 ng/kg (Fig. 5.3). Although C70 was not found in any of the

samples under investigation, an additional peak representing a C60-like

structure (the unknown product discussed above) was detected in 44% of

the samples and was always accompanied by the presence of C60. In the

case of location B, the highest concentrations were found in proximity of

the incinerator (sample A_1-A_3), whereas lower concentrations were

found mostly downwind in the surrounding area. Similar to what was

observed in the power plant location (Fig. 5.2), samples collected

immediately upwind of the chimney did not show any presence of

fullerenes (samples B_7-B_8).

Fullerenes were found in 62% of the samples collected in proximity of the

highway in Amsterdam (Location C). C60 was the most abundant species also

in this case (in 62% of the samples at concentrations between MLOQ and

0.77 ng/kg), whereas C70 was found in 8% of the samples. In the soils

collected in the Amsterdam urban area (location D), C60 was the only

fullerene found in 18% of the samples at concentrations ranging from 0.43

to 0.87 ng/kg. In contrast, no fullerenes were detected in the samples

collected in the natural area in Castricum (Location E) whereas samples

collected close to the runway of Eindhoven Airport (Location F) showed the

presence of fullerenes in 33% of the soils at concentrations usually below

the MLOQ. However, the only quantifiable samples had the relatively high

concentrations of 2.14 and 2.17 ng/kg for C60 and C70, respectively.

85

Fig. 5.2. Concentrations of fullerenes in the coal power plant area (location A) in Amsterdam. Top: concentration of C60 (orange) and C70 (blue) in the samples. Bottom: sampling locations and range of concentrations of C60. The star represents the powerplant’s main chimney.

86

Fig. 5.3. Concentrations of fullerenes in the incinerator area (location B) in Duiven. Top: concentration of C60 (orange) in the samples, C70 was not detected in any sample. Bottom: sampling locations and range of concentrations of C60. The star represents the powerplant’s main chimney.

87

Fig. 5.4. Concentrations of fullerenes in the highway area (location C) in Amsterdam. Top: concentration of C60 (orange) and C70 (blue) in the samples. Bottom: sampling locations and range of concentrations of C60.

88

5.4 Discussion

Fullerenes were present in almost a half of the surface soils sampled in the

present study and C60 was the species detected most often (48% of the

samples) in comparison with C70 (7%). However, it must be noted that

method detection limits of C70 were higher compared to C60, which may

have hindered the detection of lower concentrations of C70 in the samples.

The average concentrations, excluding the samples at concentrations below

the MLOQ, were 0.7 and 0.5 ng/kg for C60 and C70, respectively. In general,

a higher occurrence of fullerenes was observed in intensively trafficked and

industrialized areas (e.g. power plant, incinerator and in proximity of the

highway) where fullerenes were found in 61% of the samples in average.

The incinerator was the most contaminated area with regard to both the

occurrence (69% of the samples) and concentration (1.2 ng/kg of C60 in

average) whereas lower concentrations (< 1 ng/kg in average) were found

in the other locations sampled. Along the airport’s runway, the highest

occurrence of C60 was found in coincidence with the touch-down spot, thus

suggesting that the fullerenes may result from the contact of tires with the

runway during landing. In particular, the fullerenes may be directly released

from the tires or produced during the landing due to the high temperatures.

In general, our results are consistent with previous observations and

modeled data (Gottshalk et al., 2009) and suggest that C60 is the most

abundant fullerene species found in the environment at concentrations in

the range of ng/kg in soil. In particular, Sanchis et al. recently reported

similar results, with regard to both concentration and occurrence (Sanchis

et al., 2015), when monitoring trafficked and industrialized areas and

suggested that the more frequent occurrence of C60, at relatively higher

concentrations, may be due to its higher stability in the environment in

comparison with other pristine and functionalized structures (Sanchis et al.,

2013). It must be noted that, to the best of our knowledge, fullerenes are

not intentionally produced close to the sampled areas and that no products

containing fullerenes are intensively used or disposed in those areas.

Furthermore, no natural sources (e.g. forest fires) are known to the authors

that occurred during the last years in the sampling locations and that may

account for the presence of fullerenes in the surrounding soils. Therefore,

89

the presence of the fullerenes is likely due to human activities and

particularly the result of combustion processes involving waste, coal and

other fossil fuels or, in the case of the power plant, associated with the

presence of coal stocks on the soil. Thus, although it is not possible to

exclude an engineered origin, the fullerenes found in the present work were

likely unintentionally produced i.e. not manufactured nanomaterials.

Although higher concentrations and occurrence were usually found

downwind and close to the expected sources (e.g. incinerator or power

plant chimney), statistical analysis did not highlight either a significant

correlation between the concentration and the distance or a role of the

wind in the distribution of the fullerenes (P>0.05). However, it must be

noticed that, in the locations sampled, an assessment of the fullerenes

origin is complicated by the presence of multiple sources. To the best of our

knowledge no analytical method currently available is able to discriminate

between the several sources of the fullerenes detected in the environment.

In this context the ratio between fullerene species could be helpful in

understanding the origins, as different sources (e.g. coal or fossil fuel

combustion) may produce different ratios. In the present work, the

[C70]/[C60] ratio in the samples was relatively constant in both the location

A and C, with ratios of 68.9 ± 3.6 and 59.4 ± 8.8 percent, respectively.

However, the limited number of samples containing C70 did not allow a

more comprehensive comparison. The C60-like unknown specie was only

found in samples collected in the incinerator area which may be due to

either specific combustion processes (e.g. waste combustion) or to changes

in the chemical composition of pristine C60 after release in the environment.

Although characterization of the chemical was not possible in the present

study, the presence of a specific C60 chromatographic peak, at a retention

time different than that of the pristine C60, suggests the presence of a

functionalized-C60 structure that underwent an in-source fragmentation

during the ionization process. Furthermore, the unknown product is

apparently more polar than C60 as it eluted earlier than C60 in reverse-phase

chromatography. Since the functionalized fullerenes (i.e. the

methanofullerene [60]PCBM) do not undergo in-source fragmentation

when analyzed with the same method (Carboni et al., 2016), the energy

associated with the bonding of the external group in the unknown

compound is probably lower than that generally observed in the

90

commercially available functionalized fullerene derivatives. These

observations suggest that the unknown product may be a functionalized

fullerene such as the oxides recently characterized by Tiwari et al. (2014)

during the ozonation of C60 aerosol.

Finally, although the recovery of pristine C60 during the analysis may be

useful for the identification of unknown species in environmental samples,

an alternative method should be tested for their characterization. In

particular, atmospheric pressure ionization methods such as APPI may

provide similar ionization efficiency for the fullerenes (Emke et al., 2015),

but preventing or minimizing the in-source fragmentation of the

compounds.

5.5 Conclusions

The present work demonstrated the presence of fullerenes in the terrestrial

environment and especially in locations close to highly industrialized and

trafficked areas in Netherlands. Similarly to what has been reported by

previous studies in different locations, C60 was the most abundant fullerene

and was present at higher concentrations in comparison with other species.

However, the presence of an unknown C60-like structure(s), which may be

linked to either the unintentional release of different fullerene species or

the modification of pristine fullerenes in the environment, highlights both

the complexity in the environmental study of these chemicals and the need

for novel analytical strategies. Although an assessment of the origin of the

fullerenes is not possible with the current methodologies, we found no

indications that the level of soil contamination with fullerenes is related to

the intentional production and / or use of engineered nanomaterials at the

sites. Our results rather indicate that fullerenes are unintentionally

released, enter the environment particularly by combustion processes and

were likely deposited on the soil from the atmosphere. These results will be

helpful in the understanding of fullerenes occurrence and behaviour in the

environment and provide useful data for both modelling studies and for the

planning of forthcoming monitoring campaigns.

91

Chapter 6

Incubation of solid state C60

fullerene under environmentally

relevant conditions

Submitted to Chemosphere as:

Carboni A., Helmus R., Parsons J. R., Kalbitz K. and de Voogt P. Incubation of

Solid State C60 fullerenes under environmentally relevant conditions.

92

Abstract

Carbon-based nanomaterials, such as C60 fullerenes are expected to

accumulate in soil due to direct release and deposition from the

atmosphere. Although little is known about their environmental fate, these

nanoparticles may be susceptible to photochemical and microbial

degradation. In the present work, C60 was incubated for a period of 28 days

and irradiated with UVA light. Three experiments were carried out where

the fullerenes were either spiked onto a glass surface or added to quartz

sand or sandy soil samples. At specific time intervals samples were extracted

and analysed by liquid chromatography coupled to UV or high resolution

mass spectrometric (HRMS) detection. The fullerenes were degraded in all

the treatments and the decay followed a pseudo-first-order rate law. In

absence of a solid matrix, the half-life (t1⁄2) of the C60 was 13.1 days, with an

overall degradation of 45.1% that was accompanied by the formation of

functionalized C60-like structures. Furthermore, mass spectrometric analysis

highlighted the presence of a large number of fulleroid products that were

not directly related to the irradiation and presented opened cage and

oxidised structures. When C60 was spiked into solid matrices the

degradation occurred at a faster rate (t1⁄2 of 4.5 and 0.8 days for quartz sand

and sandy soil, respectively). Minor but consisted losses were found in the

non-irradiated samples, presumably due to biotic or chemical processes in

these samples. The results of this study suggest that light-mediated

transformation of the fullerenes will occur in the environment but that an

accurate assessment of their fate is complicated by the large number of

products that may derive.

93

6.1 Introduction

Fullerenes, including C60, are a class of carbonaceous nanomaterials

displaying a closed-cage structure composed of five- and six-membered

rings of carbon (Kroto et al., 1985). Due to their innovative properties,

production and application of these chemicals is developing rapidly and

fullerene-based consumer products are already commercially available

(Hendren et al., 2011; Maynard et al., 2012). Increased production and

application will inevitably mean an increased release into the environment

and there is concern for the possible negative effects on humans and the

ecosystem. Although fullerenes can occur naturally (e.g. lightning strikes,

wildfires, e.g. Howard et al., 1991) or be unintentionally produced, (e.g. as

by-products of combustion or during oil refinery, e.g. Utsunomiya et al.,

2002; Tiwari et al., 2016), due to the large scale production, the potential

release of purposefully manufactured nanomaterials is expected to play a

major role in determining the occurrence of these novel contaminants in

the environment. In this context, C60 released in the environment is

expected to accumulate in soil more than in water and the atmosphere

(Gottschalk et al., 2009). In particular, fullerenes may enter the terrestrial

environment directly as a consequence of their use and disposal as well as

due to amendment of soil with bio-solids (Navarro et al., 2013) and

indirectly by deposition from the atmosphere (Laitinen et al., 2014; Tiwari

et al., 2016). Although little is known about their environmental fate,

fullerenes may undergo functionalization, polymerization, degradation and

mineralization reactions due to the interaction with ozone, biota and solar

radiation (Panina et al., 1997; Lee et al., 2009; Avanasi et al., 2013; Tiwari

et al., 2014). With regard to their photochemistry, fullerenes are well

known to be photosensitive and to absorb light in the UV range (Carboni et

al., 2013). In particular, interaction of the fullerenes with UV-A light, which

represent the main UV irradiation reaching the Earth surface, will likely play

a major role in determining their environmental fate (Hwang et al., 2010).

Previous studies have shown that C60 dissolved in organic solvent (e.g.

benzene) undergo photo-oxidation with consequent formation of epoxides,

oxides and more polar unidentified products (Taylor et al., 1991; Wood et

al., 1991; Cregan et al., 1992), whereas photo-polymerization was observed

94

under oxygen-limiting conditions (Sun et al., 1995). Oxidative pathways

were also observed when C60 was dissolved in water (i.e. in the form of

aqueous nano-aggregates, nC60) and irradiated by UV-A (Hwang et al.,

2010), UV-C (Lee et al., 2009) or sunlight (Hou et al., 2009). In particular,

studies carried out at environmentally relevant conditions reported half-

lives from 19 to 41 h and surface functionalization of C60 (oxygenation and

hydroxylation) as well as the formation of unidentified water soluble

intermediates. However, most of the research was focused on fullerenes

dissolved in solvents. To the best of our knowledge, no research was

conducted yet that irradiates C60 nanoparticles dispersed onto a surface or

into a solid matrix (e.g. soil), which may represent the main form in which

these nanoparticles are present in the environment. The characterization of

transformation pathways for fullerenes is challenging because of the large

number of possible products that they may create (e.g. polymerization,

functionalization and cage break-down) (Taylor et al., 1991; Hwang et al.,

2010) and possible strategies have been recently reviewed by Pycke et al.

(2012). In particular, spectroscopic detection is an effective tool for the

analysis of fullerenes, due to the strong absorptivity of these chemicals in

the UV range and the lack of selectivity that may help in the identification

of the totality of the species in a sample (Carboni et al., 2013). Mass

spectrometric analysis instead can provide structural information needed

for the identification of transformation products resulting from

fragmentation (Lee et al., 2009) and oxidation (Tiwari et al., 2014). In this

context, high resolution mass spectrometry (HRMS) was successfully

employed for the detection of fullerenes and functionalized fullerenes, also

in environmental matrices (Astefanei et al., 2014b). In the present work, the

fate of fullerenes at environmentally relevant conditions was studied by

incubating C60 for a period of 28 days. The fullerenes were spiked in soild

matrices and the effect of UVA light irradiation was investigated.

95

6.2 Materials and Methods

6.2.1 Reagents and chemicals

Toluene (analytical grade), methanol (ULC/MS grade) and acetonitrile (LC-

MS grade) were obtained from Biosolve B.V. (Valkenswaard, The

Netherlands). Fullerenes C60 (CAS: 99685-96-8) and C70 (CAS: 115383-22-7)

were purchased by Sigma-Aldrich (Steinheim, Germany). Stock solutions of

the individual fullerenes were prepared in toluene and placed on an orbital

shaker (Laboshake orbital shaker, Gerhardt, Königswinter, Germany) in the

dark overnight. Further solutions needed for the experiments were

obtained by dilution of the stock solutions, stored at 4°C in the dark and

sonicated for 2 min before use (Bransonic 12, Branson, Danbury CT, United

States). Quartz sand (silicon dioxide, SiO2) was purchased from Sigma-

Aldrich (Zwijndrecht, the Netherlands). The sandy soil consisted of a top-

soil (top 10 cm) collected in Oude Schulpweg, Castricum, the Netherlands

(52° 32’ 39.689”N, 4° 39’ 5.623”E). The soil was placed in an oven at 65 °C

for one week in order to remove traces of water and then sieved with a 1.68

mm mesh.

6.2.2 Sample preparation and incubation

The sample preparation took place into a fume hood, and the laboratory

windows were equipped with UV-filters. For the preparation of the samples,

100 µl of a C60 solution in toluene (10 mg/L) was spiked at the centre of the

glass Petri dishes. Then the dishes were covered with the quartz domes

(custom made) and the solvent let dry for 30 min. Quartz sand and sandy

soil samples were prepared as following: 2 ml of a C60 solution in toluene

(20 mg/L) were spiked into 100 g of sample that was then homogenized by

stirring and let to dry. Then 300 g were added while stirring in order to

obtain a final mass of 400 g of sample at the concentration of 100 µg/kg of

C60. Eventually, 10 g were placed in the glass dishes, resulting in a sample

thickness of circa 1.5 mm. The incubation took place in a 1 m2 area provided

with three UV lamps (UVP, Keswick, Australia) set at 365 nm and placed at

a height of 80 cm. This resulted in a uniform irradiation of UVA light (350-

96

390 nm range) whose maximum intensity was 9 µW/cm2/nm at circa 360

nm (fig. D.1 in appendix D). The light measurements in the incubator were

achieved with the aid of a USB2000+ spectrometer and the Spectrasuite

software (both Ocean Optics, Duiven, The Netherlands). Non-irradiated

samples were placed into the incubator but covered with aluminium foil in

order to prevent exposure to light. The incubator was covered with a 100%

obscurant curtain and the temperature was held constant at 25 ± 0.5 °C.

6.2.3 Extraction and analysis

The samples (n=3) from each treatment were extracted at the beginning of

the experiment and at specific time intervals of 1, 3, 7, 14 and 28 d as

following: the samples were taken out of the incubator and 50 µl of a C70

solution in toluene (internal standard, 10 mg/L) were added. After

equilibration time of 30 min, 5 ml of toluene was used for rinsing the petri

dishes and then transferred into a 10 ml vial. The procedure was repeated

twice. In the case of quartz sand and sandy soil, the samples were

homogenized by stirring after spiking of the internal standard and

underwent extraction with the protocol previously described by our group

(Carboni et al., 2013). The extracts were analysed with the HPLC-UV method

described in Carboni et al. (2013) with minor modifications. Briefly, the

method employed a pyrenylpropyl silica stationary phase (Buckyprep),

whereas the mobile phase consisted of toluene and acetonitrile. Elution

was achieved with an isocratic method employing a 80:20

toluene:acetonitrile mixture. Quantification was based on the

chromatogram peak areas as described in the appendix D.2. The UHPLC-

HRMS analyses were performed using an UHPLC system (Nexera, Shimadzu,

Den Bosch, The Netherlands) equipped with a binary pump, autosampler

and column oven. Retention of the chemicals was achieved with a core-shell

Kinetex 2.6 µm biphenyl 100 Å chromatographic column (Phenomenex,

Utrecht, the Netherlands) consisting of a biphenyl stationary phase whereas

methanol (A) and toluene (B) were used as mobile phase. A gentle elution

program was used to enhance separation of compounds with varying

polarity range: from 0 – 5 min B was kept at 0%, following a linear increase

to 75% at 21 min and a final holding step until 25 min. The flow rate was set

to 0.4 ml/min, whereas the column temperature was kept at 30°C. The MS

97

methodology was adopted from (Carboni et al., 2016). Mass detection was

carried out with a high resolution Quadrupole-Time of Flight mass

spectrometer (Q-TOF; maXis 4G equipped with HD collision cell, Bruker

Daltonics, Wormer, The Netherlands) coupled to the UHPLC system

described earlier. Compounds were ionized using an Ion Booster

electrospray ionization source (IB-ESI) operating in negative mode using the

following settings: capillary voltage 1000V, end plate offset -400V, charging

voltage 300V, dry heater 200°C, nebulizer gas 4.1 bar and dry gas 3 l/min.

Nitrogen was used for ionization and collision gas, and obtained from a N2

generator (Avilo, Dirksland, The Netherlands). Ion transfer settings were as

follows: funnel radio frequency (RF) 325 Vpp (voltage point to point) and

multipole RF 300 Vpp. Mass calibration was achieved as reported by

Carboni et al. (2016) with minor modifications and is described in the

appendix D.3. The collision cell RF, transfer time and prepulse storage time

were varied within four time segments throughout the analysis. The first

two segments were dedicate to low and high masses calibration,

respectively. Consequently, compounds eluting within the first ten minutes,

i.e. relative polar, were measured with settings optimized for low masses

(segment 3) whereas fulleroid and other very non-polar compounds eluting

after ten minutes were analysed with settings optimized for high masses

(segment 4). Mass spectra were recorded at 2 Hz with a range of 50 – 2000

m/z. Fragmentation data of the most intense peaks were automatically

acquired with Auto MS/MS mode using a maximum cycle time of 3 seconds.

Smart exclusion and active exclusion were enabled to limit acquisition of

continuous background and increase uniqueness of precursor selection,

respectively. More information regarding the MS settings for the segments

and Auto MS/MS are reported in the appendix D.4. An extensive list of

masses was compiled from background compounds found in solvent and

sample blanks , and set as exclusion masses (+/- 0.05 Da window) to

improve quality of mass spectra. Unknown transformation products (TPs)

in a sample were screened and identified according a five step procedure.

First, a list of chromatographic peaks of interest was compiled from manual

inspection of base peak chromatograms and by use of the software based

Auto MS/MS peak finder. Secondly, for each peak of interest, a list of

masses of interest was manually generated from averaged mass spectra.

For these masses, an extracted ion chromatogram (EIC) was generated to

98

verify its origin from the chromatographic peak. Thirdly, a candidate list of

chemical formulas was created based on accurate mass (≤ 5 ppm deviation)

and isotopic fit (≤ 50 mSigma) with SmartFormula. Fourthly, candidates

were removed if present in solvent or sample blanks. Finally, tentative

identification of candidates was performed from MS/MS fragmentation

data, if present. Mass spectral data were processed with DataAnalysis 4.3

(Bruker Daltonic, Wormerveer, The Netherlands).

6.3 Results and discussion

6.3.1 Incubation

Losses of C60 fullerenes occurred during irradiation with UV-A light in all the

experiments performed, as shown in fig. 6.1. Furthermore, similar to what

was observed by Hwang et al. (2010) when irradiating aqueous C60 with UV-

A light, the decay of the fullerenes followed a pseudo first order reaction

rate in all the irradiated treatments (r2 ≥ 0.85, Table 6.1).

Table 6.1. Degradation characteristics derived from photolysis experiments with C60 fullerene. The results of the UVA irradiated and non-irradiated incubations are reported on the left and right side of the table, respectively.

t½ : half-lives, k: pseudo-first-order reaction constant, r2: regression coefficient.

When the C60 was added onto glass dishes (“glass” in the further

manuscript), degradation of the fullerenes started after 3 d and a total loss

of 45.1% was observed after 28 d of incubation (fig. 6.1a). The negligible

99

losses in the non-irradiated glass samples indicate that, similar to what was

previously reported in water (Hwang et al., 2010; Hou et al., 2009) and

organic solvents (Taylor et al., 1991), the degradation of the C60 was due to

a photochemical process. These results suggest that fullerenes released in

the environment will interact with the ultraviolet portion of the solar

radiation and will be degraded as a result. However, it must be noted that

the half-life of 13.1 d in the glass incubation was much longer than those

previously reported for the dissolved C60, estimated in the range of 10-40 h

(Hou et al., 2009; Taylor et al., 1991), and could be explained with a lower

reactivity of the fullerenes in the solid state in comparison with the

dissolved structures or a role of the solvent in the process.

When C60 was spiked into solid matrices, the observed losses in the

irradiated samples were higher and reached 88.1% and 100% for quartz

sand and sandy soil, respectively (fig. 6.1b and 1c), at the end of the

incubation. These correspond to half-lives of 4.5 and 0.8 days, respectively

(Table 6.1), and suggest that other processes took place, in addition to the

irradiation, that enhanced the degradation in these samples. Especially in

the sandy soil samples, the matrix-enhanced loss may be explained with a

biotic degradation due to microorganisms present in the soil or by other

processes such as a matrix-related photosensitisation (e.g. due to organic

matter). Thus, these results indicate that once deposited onto soil, the fate

of fullerenes will be mostly determined by factors other than the

irradiation. This is supported by the fact that, in contrast to the glass

incubation, the degradation of the fullerenes started at the beginning of the

incubation and that consistent losses over time were found also in absence

of light (Table 6.1). In particular, also the decay of C60 in the non-irradiated

samples followed a pseudo first order reaction rate (r2 ≥ 0.80) with half-lives

of 14.3 and 6.7 days for quartz sand and sandy soil, respectively.

100

Fig. 6.1. Concentration of C60 fullerenes over time in (a) fullerenes dried on top of a glass surface (“glass”), (b) spiked into quartz sand and (c) spiked into sandy soil at the concentration of 100 µg/kg. The red line represents the sample irradiated with UV-A light whereas the blue line represents the samples incubated in the dark.

101

6.3.2 Transformation products

6.3.2.1 C60-like products

In contrast to the HPLC-UV chromatograms, that were dominated by the C60

and C70 peaks (fig. D.5 in appendix D), UHPLC-HRMS chromatograms

revealed the presence of additional peaks (fig. 6.2a). In particular, three

structures, namely “C60-I”, “C60-II” and “C60-III”, showed accurate masses

and isotopic cluster distributions equal to that of C60, but eluted later at

retention times 19.2, 20.0 and 22.5 min, respectively (fig 6.2a and 6.2c).

Fig. 6.2. UHPLC-HRMS analysis of glass incubation samples irradiated with UV-A light. (a) Chromatogram presenting the C60 peak at retention time 15 min and smaller C60-containing peaks (C60-1, C60-II and C60-III) eluting later. (b) Several unknowns peaks eluting before C60 (blue frame). (c) A zoom of the 18-24 min range (red frame), highlighting the presence of C60-containing structures.

102

These products could not be characterized in the present work and their

composition has not been elucidated yet. However, the detection of a C60

cage structure suggests the presence of surface functionalized fullerenes

undergoing in-source fragmentation during the ionization process.

Consequently, modification of the ionization method with regard to the

source temperature and capillary voltage (i.e. to promote ‘softer

ionization’) did not lead to the identification of the precursor masses and

resulted in a near or complete loss of the signal. The detection of C60On

traces together with the C60-like peaks (Fig. D.6 in appendix D), suggests that

these products may consist of oxidized products similar to those observed

by Tiwari et al., (2014) who recently reported the elution of C60 oxides (C60O,

C60O2 and C60O3) to occur after that of C60 with a methodology similar to that

of the present work (i.e. reverse phase chromatography with functionalized

silica stationary phase). Quantitation of these structures was hindered by

the lack of analytical standards. However, analysis of the peak areas showed

that, in the glass incubation samples, the decrease in concentration of C60

corresponded to the increase of the C60-like transformation products over

time (Fig. 6.3). In particular, the abundance of all the structures increased

from the baseline signal after 3 d of incubation, in correspondence with the

decrease of C60, and reached a maximum after 14 d. The decrease to lower

levels at the end of the incubation suggests the formation of intermediates

in the transformation of fullerenes irradiated with UV-A light. It must be

noted that, in quartz sand and sandy soil extracts, these products were

either not detected or did not show a different trend between irradiated

and non-irradiated treatments. This could be due to several reasons,

including 1) different processes (e.g. biotic transformation) that took place

in the solid matrices, 2) the extraction of these chemicals was hindered by

the sample matrix and / or 3) the products were degraded at a faster rate

and were already removed at the first sampling step.

103

Fig. 6.3. Abundance (peak area) over time of (a) C60-I, (b) C60-II and (c) C60-III products in the glass incubation. The blue and red lines represent the irradiated and non-irradiated samples, respectively.

104

6.3.2.2 Unknown products

Auto MSMS analysis (see section 6.2.3) highlighted the presence of several

structures that eluted immediately before C60, between 12.2 and 14.4 min

(fig. 6.2b), and will be referred as “unknowns” in the further manuscript. An

example of the Auto MS/MS analysis is provided in the appendix D.7. In

general, base peaks were identified as the most abundant traces at a certain

retention time and, within each base peak, several m/z traces were

automatically selected for MS/MS. A tentative identification of the

unknowns was based on the MS1 and MS2 accurate masses and isotopic

cluster distribution, but complicated by the low signals intensity and the co-

elution of more structures. Furthermore, in-source fragmentations may

have occurred similar to what was reported above (section 6.3.2.1). A

summary of the m/z values relative to the unknowns found in the glass

incubation is reported in table 6.2 with a tentative identification of their

chemical formula.

Analysis of the mass spectra highlighted an ionization mechanism (i.e. the

formation of a radical molecular ion [M]-•) and isotopic cluster distribution

similar to that of the fullerenes. Furthermore, the mass (as measured m/z)

was always larger than that of the C60, suggesting functionalization of the

carbon cage. These fulleroid structures displayed both carbon addition and

loss from the pristine C60 cage. The addition of carbon was always in the

form of (CH3O)n groups and has already been reported to occur as a

consequence of methanol adduction to the fullerenes (Kolkman et al.,

2013). However, since the methodology applied in the present work is less

prone to the formation of adducts in comparison with other ESI-MS

techniques (Carboni et al., 2016), the (CH3O) clusters suggests a

methoxylation of the fullerenes. This hypothesis is supported by the fact

that the number of (CH3O) additions were not associated with all the

transformation products detected. On the other hand, carbon loss occurred

with the formation of both C59 and C58 species. The loss of carbon, and

specifically C2 units, has already been observed in the mass spectrometric

analysis of functionalized C60 (e.g. fullerols; Chao et al., 2011) and indicates

the opening of the cage.

105

Table 6.2. Summary of the unknown products found in the glass incubation. The base peak indicates the most abundant structure detected in MS1 at a certain retention time (Rt). Within each base peak, some m/z traces were automatically selected (Auto MSMS) in MS1 and sent to the MS2.

Rt: retention time

106

Additional functionalization mostly consisted of hydrogen and oxygen that

could be located in hydroxylic, ketons and epoxides groups. We found no

evidence of clusterization of the fullerenes, such as the formation of C120O

observed by Taylor et al. (1998). These results indicate that the unknowns

found in the present work could be similar to those observed during UV and

sunlight irradiation of C60 dissolved in water and organic solvents (Taylor et

al., 1991; Hou et al., 2009; Lee et al., 2009). Similar to the C60-like structures

(see section 3.2.1), the unknowns were more abundant in absence of a

matrix. However, the analysis of the peak areas did not show a direct

correlation between irradiation of the samples and their concentrations.

Indeed, although not present in the spiking solutions and non-spiked

samples (fig. D8 in appendix D), some of the products were found to be

already present at the beginning of the incubation and may have been

formed during the sample preparation. Furthermore, some unknowns

increased in abundance in the non-irradiated treatment only (Fig. D.9 in

appendix D) indicating that some processes either did not occur in presence

of light or that the products were too short-lived to be appreciated at the

time intervals in the present study. Finally, it must be noted that no

evidence of transformation products was found at the beginning of the

chromatograms (between 0 and 12 min) where the more polar species are

expected to elute. This could be due to a limitation of the present method

that did not allow their detection or, similar to what reported by Taylor et

al. (1991), that too polar products may not have been extracted with

toluene. In general, alternative methodologies should be tested that may

overcome the limitations of this work. In particular, we recommend 1) the

use of higher concentrations that, although not mimicking environmental

conditions, would simplify the determination of the fulleroid species in the

samples, 2) a second step of extraction with a more polar solvent in order

to recover polar products that were likely excluded in this work and 3)

improve identification by the use of alternative MS ionization conditions

and/or interfaces to avoid in-source fragmentation of precursor ions.

In general, the results of the present work show that fullerenes will be

degraded in the environment and that the removal from the soil

compartment could be relatively fast. In this context, the data hereby

presented will be helpful for modelling and environmental studies in

general. In particular, we have shown that the assessment of fullerenes fate

107

and occurrence is complicated by the formation of a large number of

transformation products that will likely display different properties in

comparison with the original structures (e.g. enhanced mobility). Thus,

future research addressing these topics will need screening analysis such as

those hereby presented in order to identify the totality of the species

present in environmental samples as well as to further elucidate the

pathways of degradation in the environment.

6.4 Conclusions

The effect of UVA light irradiation on fullerenes incubated at the solid state

was investigated. When C60 was directly irradiated, i.e. in absence of a

matrix, the irradiation had a clear effect on the degradation. However,

when the C60 was spiked into a quartz sand or sandy soil samples, much

faster degradation occurred suggesting that, once deposited onto the soil,

their fate will more likely be determined by other factors such as the

interactions with soil microbiota, including bioturbation and

biotransformation. The results indicate that fullerenes could undergo an

oxidative pathway similar to that observed for C60 dissolved in water, and

can result in the break-down of the carbon cage. The present study

highlighted the complexity of these processes and the need for analytical

strategies for their understanding. For instance, less selective methodology,

such as UV detection, can be suitable for the quantitation of the totality of

fullerenes in the samples, whereas HRMS-based techniques are required for

the characterization of the many species that likely result from their

transformation. Eventually, the results hereby presented will be helpful in

the assessment of the environmental fate of fullerenes and in defining

future strategies for their study.

108

109

Chapter 7

Synthesis

110

Fullerenes are considered to be novel environmental contaminants and the

characterization of their occurrence and fate in the environment is

fundamental for the general assessment of their possible effects on humans

and the environment. At the start of the present work, several studies

claimed the lack of analytical methodologies able to fulfil these goals and in

one of the first critical reviews on the topic, Isaacson et al. (2009) suggested

that “Sensitive and mass-selective detection, such as that offered by mass

spectrometry when combined with optimized extraction procedures, offers

the greatest potential”. In this work, such methodologies were developed

to address two specific scientific issues, the analysis of fullerenes in

environmental samples and the study of the presence and fate of these

chemicals in soil.

7.1 Development of analytical methods for the analysis of fullerenes in the

terrestrial environment

The work achieved in this thesis contributed to overcome some limitations

in the soil analysis of fullerenes and particularly (I) the development of

routine extraction procedures (Chapter 2), (II) the analysis of mixtures of

pristine and functionalized structures (Chapter 3), (III) the determination of

these nanoparticles at environmentally relevant concentrations (Chapters

4 and 5) and (IV) the analysis of transformation products (Chapters 5 and

6). These are discussed in the following synthesis.

With regard to the extraction methodology, Chapters 2 and 3 show that the

combination of ultrasonication and shaking extraction is a robust and

reproducible technique to recover fullerenes from soil independent of the

particle size distribution (texture) of the soil under investigation.

Furthermore it has the advantages of being applicable to large batches of

samples with relatively low costs in terms of materials and energy. It was

observed that functionalized fullerenes (Chapters 2 and 3) and unknown

fulleroid structures (Chapters 4 and 5) can be recovered to a similar extent

of the pristine C60, making this procedure suitable for the study of both

engineered and naturally occurring nanoparticles as well as transformation

111

products that may result in the environment. A current drawback of this

technique is that it likely alters the conditions of the fullerenes in the soil,

i.e. may not be representative of their natural state (homo- or hetero-

aggregated).

Non-aqueous reverse phase (ultra) high performance liquid

chromatography can provide sub-optimal separation of several fullerenes

and functionalized fullerenes species (Chapters 3 and 4) and, in

combination with selective detection methods such as high resolution mass

spectrometry (Chapter 4), can provide the unambiguous determination of

the fullerenes, even in complex mixtures where co-extractants and similar

structures are present. In particular, the experiments reported in this thesis

highlighted the performance of novel stationary phases such as core-shell

biphenyl (Chapter 4) and pyrenylpropyl silica (Chapter 3). Although the

retention mechanism has not been completely elucidated yet, the main

retention is likely provided by pi-pi interactions between the aromatic rings

of both the fullerenes and the functional groups of the stationary phases.

This indicates that such materials can be applied for the analysis of most of

the fullerenes and related structures, but also that some functionalized

species such as the poly-hydroxylated fullerols may not be retained. A direct

comparison of the column’s performances is not possible due to the

differences in the overall analytical setup they were employed in. The

chromatographic study reported in Chapter 3 was optimized for the

separation of more structures, for which pyrenylpropyl silica is more

suitable than byphenyl by virtue of the larger surface available for the

interaction with the fullerenes. However, the higher retention is achieved

at the expense of longer analysis time and this is not always desirable,

especially in routine analysis. Thus, further methods development, aimed

to monitoring studies, employed the core-shell biphenyl stationary phase

that can provide much shorter analysis with complete resolution of the

pristine structures only (Chapter 4), which are the most abundant in the

environment (Chapter 5). Chapter 4 and 6 also show how methodologies

employing this stationary phase can be optimized with regard to the

separation of co-extractants and functionalized structures, respectively.

Eventually, the work in this thesis showed that, when coupled to high

performance liquid chromatography, both UV and high resolution mass

112

spectrometry detection can be successfully applied in the analysis of

fullerenes. However, for environmental studies these techniques must

necessarily address different goals. In particular, UV detection lacks the

specificity and sensitivity required for the analysis of the low concentrations

present in real soil samples. Nevertheless, since both fullerenes and

functionalized structures generally display similar absorptivity behaviour

(Chapter 3), UV detection can still be a valuable tool for the identification

of unknown species, i.e. products of transformation during incubations and

ecotoxicological studies, and in general in experiments where higher

concentrations are applied (Chapter 6). Mass spectrometry can fulfil the

requirements of specificity and sensitivity necessary for an accurate

determination in soil matrices and is therefore the detection method of

choice. In particular, most experiments carried out in this thesis rely on a

high resolution instrument, the quadruple time-of-flight (Q-TOF) mass

spectrometer. Chapters 4 and 5 show how Q-TOF-based methodologies can

be applied for the investigation of fullerenes in soil and sediment matrices

and in environmental surveys addressing these chemicals. Furthermore,

software-based data collection and analysis represent a valuable tool for

the search and identification of non-target species in the samples (Chapter

6).

Ionization of the fullerenes (singly-charged molecular ions) was obtained

with a heated electrospray ionization (H-ESI) interface, operating in

negative mode. This is a hard ionization technique in comparison with

standard ESI and particularly fits the analysis of fullerenes due to the

resistance of these chemicals to high temperatures. As shown in Chapter 4,

the Ion-Booster ESI (IB-ESI) interface applied in this study presents several

advantages in comparison with existing methodologies and specifically: (I)

a higher ionization efficiency and (II) a lower tendency to create adducts

than standard ESI, and (III) the production of an isotopic pattern distribution

better matching the theoretical pattern in comparison with other

techniques. This latter point is particularly important in the analysis of

fullerenes. Indeed, one of the peculiarities in their mass spectrometric

analysis is the difficulty (impossibility in this work) to obtain fragmentation

of the closed-cage structure and, although exohedralic species (e.g.

[60]PCBM) can undergo fragmentation of the functional group from the

cage, this limitation is especially relevant in the analysis of pristine species.

113

Thus, correct and highly resolved isotopic clusters in combination with the

accurate mass provided by the Q-TOF can compensate these drawbacks and

provide identification criteria needed for the unambiguous determination

(Chapters 4 and 5). Nonetheless, the lack of fragmentation allows analysis

at high collision energies with the advantage of a “background clean-up” in

MS2, i.e. the removal of co-extractants from the samples. As shown in the

illustration of MS1 and MS2 spectra in Chapter 4, this result in extremely

clean mass spectra and is especially useful during the analysis of complex

matrices. A further advantage of this approach is that in-source

fragmentation of functionalized fullerenes allows identification of

fullerenes that would be otherwise not detected by target analysis e.g. the

C60-like products found in environmental samples (Chapter 5) and

incubation studies (Chapter 6). Such “defunctionalization” strategy is

similar to that suggested by Pycke et al., (2012) and can represent a valid

tool for future analysis of heterogeneous mixtures as well as colloidal nC60

structures, were fullerenes congeners may represent the majority of the

species in the samples. In this context the main drawback of the IB-ESI Q-

TOF methodology hereby proposed is the impossibility to prevent the

defunctionalization, since any attempt to prevent the fragmentation (i.e.

with “softer” analysis) was inconclusive. Thus, further studies should

address such limitations with the optimization of the current methods or its

combination with alternative techniques. Eventually, although optimized

for the study of soil samples, the methodologies developed in this study

should in principle be applicable to the analysis of fullerenes extracted from

other media (e.g. biological matrices).

Eventually, it must be noted that the current methodologies/strategies

present several limitations that future studies should address. One is

represented by the lack of proper internal standards for quantitative

purposes. Although standard addition methods (Chapters 4 and 5) and the

use of other fullerenes as internal standard (Chapter 6) can allow a proper

quantitation in certain experimental conditions, there is a need for high

purity, isotopically labelled, materials that are currently unavailable on the

market at reasonable prices. A second limitation, especially in the analysis

of complex matrices, is the lack of optimized clean-up procedures that can

improve the quality of the extracts and allow a better identification of the

fullerenes. In conclusion, further enhancements of the current strategies

114

for the characterization of fullerenes in soil, and generally in environmental

matrices, will necessarily involve the combination, and possibly

hyphenation, of more analytical techniques. Indeed, characterization of

fullerenes’ behaviour in the environment necessarily relies on the study of

the interaction with natural components that will determine transport,

accumulation and the general fate of these nanoparticles. Methods such as

those hereby proposed provide qualitative and quantitative data about the

presence of fullerenes but cannot completely describe these interaction.

Thus, imaging techniques (e.g. transmission electron microscophy, Goel et

al., 2004) and novel methodologies such as the recent coupling of field flow

fractionation (FFF) with HRMS (Herrero et al., 2014) could be employed in

order to characterize these processes with regard to properties such as the

composition, size, size distribution and morphology.

7.2 Occurrence and fate in the environment

Some of the main research questions that formed the basis of this project

were: “To which extent are fullerenes present in the soil compartment?”

and “what is the contribution of engineered nanomaterials to their overall

concentration?”. The answers were mostly hindered by the lack of

monitoring studies that allow estimation and modelling of the

environmental concentrations on the basis of empirical data.

The environmental survey reported in Chapter 5 shows that fullerenes are

widely present in the environment but at relatively low concentrations. In

particular, pristine structures occur in the soil in the part per billion range

and, although a precise assessment of their presence will need further

work, similar studies carried out in the last two years support these

observations. These findings are also in line with the amounts predicted in

soils due to use and production of engineered nanoparticles (Gotthschalk

et al., 2009). However, although the presence of manufactured materials

cannot be excluded a priori, the occurrence of fullerenes in the Dutch soils

investigated in this thesis was likely due to incidental sources, i.e. due to

combustion processes of carbonaceous materials such as coal and fuel.

Thus, the overall occurrence in the terrestrial environment may nowadays

be of anthropogenic origin but related to incidental sources more than the

115

production and application of ENMs. This hypothesis is also supported by a

recent modelling study (Tiwari et al., 2016), which suggests that the global

emission of incidental C60 may be several orders of magnitude higher than

that of manufactured C60. However, no methodologies are available yet that

can clarify the origin of the nanoparticles detected in the environment and,

although some species could be linked to a source (e.g. [60]PCBM in

photovoltaics application), this is not valid for pristine structures (e.g. C60,

C70) that can result from both natural and anthropogenic activities as well

as resulting from transformations of precursors and other species in the

environment. In Chapter 4 it is proposed that an assessment of the ratio at

which different species occur in the environment may be helpful in

clarifying their origin, but this hypothesis is hindered by the lack of similar

environmental surveys and the fact that most studies are focused on the

determination of C60 only. Thus, the best strategy for the unequivocal

identification of fullerenes’ origins may be the characterization of source-

related (or process-related) products, such as the unknown C60-like

structure presented in Chapter 5 that, although not characterized yet, may

represent a marker for nanoparticles production during incinerators

activity.

The incubation studies in Chapter 6 show that, upon release in the

environment, C60 will interact with the ultraviolet portion of the sunlight

irradiation and that it will be degraded as a consequence. However, this

process will take place at a relatively slow rate and will likely affect only the

fullerenes that are released in the atmosphere and those that deposit on

the top-soil (i.e. those that are directly exposed to the light). Once mixed

into soil, other biotic and abiotic phenomena will likely determine their fate

to a larger extent. In particular, biotic factors could determine faster decay

rates similar to those observed for C60 incubated into sandy soil in this study.

The study in Chapter 6 also suggests that, upon release in the environment,

fullerenes in the “solid” state will be degraded by abiotic factors and that

the transformation pathways will be similar to those already observed

during ozonation and irradiation of the colloidal structures. However, it is

not possible to clarify yet whether biotic factors will determine similar

pathways and if they will proceed until mineralization of the fullerenes. The

figure below shows a putative pathway of degradation of C60 in the

environment.

116

Fig. 7.1. Possible degradation pathway of C60 upon release in the environment. Light and oxygen related reactions on the closed cage structure (1) can lead to the functionalization of the fullerenes (2) with formation of oxidized species (3). Destabilization of the fullerenes can determine carbon removal and opening of the cage (4) that could in turn trigger the fragmentation of the cage (5) with the production of transformation products or mineralization (6).

In general, reactions on the cage, enhanced by interactions with light and

oxygen species, are expected to involve the carbon(s) in the junctions

between two hexagonal rings (Diederich, 1997). This can lead to the

oxidation of the fullerenes with consequent production of oxidized

functionalized structures that can in turn determine a destabilization of the

fullerenes and their progressive fragmentation. Such hypothesis is

supported by the detection reported in Chapter 6 of fulleroid traces that

display carbon loss and several degrees of oxidation, hydroxylation and

methoxylation. These species are more polar than the pristine fullerenes

and will likely exhibit enhanced transport in soil. In order to characterize the

life-cycle of fullerenes in the environment, further studies will need to focus

on such oxidative pathway and the interaction of the transformation

products with environmental components such as atmosphere particulates

and organic matter.

117

7.3 Concluding remark

The studies reported in this thesis provide tools and pioneering data about

the occurrence and fate of fullerenes in soil. Hopefully, these will be helpful

for future studies and will enhance the assessment, regulation and the

general understanding of these nanoparticles in the environment. A further

ambition of this work is that of being helpful for environmental studies

focussed on other carbon-based nanomaterials, such as graphene and

carbon nanotubes, whose assessment is at least as difficult and whose

production and application is emerging as well.

Nonetheless, future perspectives about fullerenes occurrence and fate are

very uncertain. To date, the majority of these nanoparticles in the

environment is likely due to incidental sources (both natural and

anthropogenic), but the production of engineered nanomaterials will play a

major role in determining both the amount and the number of species that

will enter the environment. This is particularly relevant for the colloidal

nC60, whose behaviour and fate are expected to differ from the “solid”

structures and that could easily find industrial application in the near future

due to the impressive properties that they display and the relative ease and

low costs of production. In addition, while C60 by-product release may

decrease in the future, due to current and future policies aimed to limit the

global emissions and the use of coal and fossil fuels (e.g. Kyoto protocol),

engineered nanomaterials will likely find increasing applications and uses.

In this context is at least interesting to note that sustainable energy

technologies, such as photovoltaics, could strongly rely on the application

of engineered fullerenes. Eventually, an increasing contribution of the

manufactured fullerenes to the global load could present new scenarios

where their presence is enhanced in a wider variety of ecosystems. For

instance, while incidental species are necessarily dispersed onto soil by the

atmosphere, resulting in a more homogeneous distribution in the

environment, manufactured nanomaterials will likely present hotspots of

accumulation close to production, use and dumping sites. In general, future

studies focussed on the characterization of fullerenes in the environment

will necessarily need to characterize (I) such hotspots of accumulation, (II)

the origin of fullerenes found in the environment, with particular regard to

118

the ENMs and (III) to identify the totality of the fulleroid species that may

result from nanotechnology as well as from natural processes.

119

Appendices

120

Appendix A:

Supportive information chapter 3 Table A.1. The fullerenes included in the study.

121

Fig. A.2. HPLC-UV chromatogram of the fullerenes. In blue detection at 332 nm, in red detection at 305 nm.

122

Appendix B:

Supportive Information Chapter 4

Fig. B.1. Temperature gradient of the IB-ESI interface during the analysis. Notice that, although the Y axis indicates APCI heater, the interface employed was a heated ESI.

Mass calibration

Table B.2. An example of the mass calibration obtained from sodium acetate cluster.

Reference m/z Resulting m/z Intensity Error [ppm]

387.0261 387.0261 245506 -0.033

469.0292 469.0293 598888 0.147

551.0323 551.0322 1142349 -0.241

633.0354 633.0354 1651113 0.104

715.0384 715.0386 1805570 0.182

797.0415 797.0413 920056 -0.28

879.0446 879.0447 494061 0.136

961.0477 961.0477 331092 0

1043.0507 1043.0507 155736 -0.022

1125.0538 1125.0538 70091 0.005

Standard deviation of

the calibration curve:

0.279

1 2 3 4 5 6 7 Time [min]

325

350

375

400

425

450

[°C]

AP

CI H

eate

r

123

An example of the SAM results in the present work.

Figure B.3. Chromatograms of the standard addition method (SAM) in matrix 5. Left: C60 at the concentrations of 25 (yellow), 50 (purple), 100 (orange), 250 (brown) and 500 ng/L (red). Right: C70 at the concentrations of 25 (grey), 50 (black), 100 (violet), 250 (green) and 500 ng/L (blue).

Fig. B.4. C60 response versus concentration calibrations obtained during the standard addition method experiments using method B.

124

Analysis of several fullerene derivatives with the method developed in the

present study.

Fig. B.5. Chromatograms of C60, C70 and six methanofullerene structures at the concentration of 500 ng/l analyzed with the method B.

125

Table B.6. Standard addition method results in the present study.

126

Fig. B.7. The response of C60 spiked at 500 ng/L into matrix 2 with toluene:methanol ratio of A) 50:50 (v:v), B) 40:60 (v:v) and C) 35:65 (v:v).

Chromatographic separation obtained applying the method B.

Fig. B.8. Chromatographic separation of C60 (Rt 4 min) and C70 (Rt 6 min) obtained with the method B. The orange and red lines indicate the MS1 and MS2 respectively for C60 whereas the green line and blue line are relative to MS1 and MS2 signal respectively for C70.

127

Appendix C:

Supportive Information Chapter 5

Table C.1. List of the samples included in the environmental survey.

128

129

130

131

132

C.2 UHPLC-IB-ESI QTOF analysis

The chromatographic method employed a core-shell biphenyl stationary

phase and a mobile phase consisting of methanol (eluent A) and toluene

(eluent B). The eluent was set to flow at 600 μl/min and programmed to

start with 40 sec at 100% eluent A (focusing step), a linear gradient up to

35% eluent B in 20 sec, an isocratic step of 4 min at 35% eluent B followed

by a linear gradient of 1.5 min to reach 60% B, a linear gradient of 10 sec

was in order to reach 100% eluent B which was maintained for 2.5 min. The

chromatograms were divided into four segments: segment 1 (from 0 min to

0.1 min) to assure the correct position of the loop and loading the mass

calibration solution during the LC equilibration, segment 2 (from 0.1 min to

0.4 min) dedicated to the mass calibration, segment 3 (from 0.4 min to 4.4

min) for the SRM analysis of C60 (m/z 720.0005) and segment 4 (from 4.4

min to 9.6 min) for the SRM analysis of C70 (m/z 840.0005). The optimized

working conditions for the IB-ESI operating in negative mode were: capillary

voltage 1000V, end plate offset -400 V, charging voltage 300 V, nebulizer

gas 4.1 bar, dry gas 3.0 l/min and dry heater 200 °C. The present settings

allowed a mass resolving power up to 80000 (0.03 m/z FWHM). An

additional UHPLC-IB-ESI-QTOF method was employed for the analysis of the

samples collected in the incinerator area (location B) that was specifically

aimed to the characterization of the Unknown C60-like fullerene found in

the samples. The method consisted of a longer analysis (25 min) with the

following program: the first minute was a focusing step with 100% eluent

A, then a gradient up to 21 min in order to reach 75% eluent B which was

maintained for 4 min. The analysis were performed with a column

temperature of 30 °C and with injection volume of 10 µl. The

chromatograms were split in three segments, segment 1 (from 0 min to 0.1

min) to assure the correct position of the loop, segment 2 (0.1-0.4 min) for

mass calibration with 2 mM sodium acetate solution and segment 3 (0.4-25

min) for the MS detection of the fullerenes. The IB-ESI parameters were the

same as reported above with the exception of 1) the method was run at

different temperatures and specifically 275, 300, 350, and 475 °C and 2)

both Auto MSMS mode and broad band collision induced dissociation mode

(bbCID) were tested. Furthermore, several capillary voltages (i.e. 800 V vs

1000 V) were tested.

133

C.3. Matrix effect in the samples

The matrix effect (ME) in the soil extracts was calculated as following:

𝑀𝐸 =𝐶𝑠𝑝𝑖𝑘𝑒

∗ − 𝐶𝑠𝑎𝑚𝑝𝑙𝑒∗

𝐶𝑠𝑝𝑖𝑘𝑒 (1)

where C*spike is the concentration observed after the addition of the

fullerenes standard. C*sample is the concentration observed in the extracts

and Cspike is the concentration added to the samples. The concentration in

the extracts (Csample) was calculated as:

𝐶𝑠𝑎𝑚𝑝𝑙𝑒 = 𝐶𝑠𝑎𝑚𝑝𝑙𝑒

∗

𝑀𝐸 (2)

134

Appendix D:

Supportive Information Chapter 6

Fig. D.1. The spectrum of the lamps used for the irradiation in the present study.

135

D.2 Quantitation of the C60 in the samples

A calibration line was obtained by running levels with fixed C70

concentration (50 µg/L) and increasing C60 concentration (from 4 to 128

µg/L). The resulting line can be described as:

(1) 𝑌 = 𝑎𝑋 + 𝑏

where a and b are the slope and intercept of the line, respectively; Y is the

ratio of C60 and C70 chromatographic peak areas and X is the ratio between

C60 and C70 concentration (conc) in the calibration levels. Accordingly, the

equation (1) can be rearranged as:

(2) 𝐶60 𝐴𝑟𝑒𝑎

𝐶70 𝐴𝑟𝑒𝑎= 𝑎

𝐶60 𝑐𝑜𝑛𝑐

𝐶70 𝑐𝑜𝑛𝑐+ 𝑏

The concentration of C60 in the samples was obtained as following

(3) 𝐶60 𝑐𝑜𝑛𝑐 =(

𝐶60 𝐴𝑟𝑒𝑎

𝐶70 𝐴𝑟𝑒𝑎−𝑏)

𝑎

136

D.3. Mass calibration

In general, mass calibration was achieved as reported in Carboni et al., 2016

with minor modifications. In order to allow the analysis of both the mass

calibration solution and the analytes, at the start of each analysis the source

heater temperature was 325°C for mass calibration and quickly ramped to

450°C afterwards during analysis of chromatographed compounds. Internal

mass calibration was automatically performed by loop injection of a 2 mM

sodium acetate solution dissolved in 1:1 ultrapure water-isopropanol. The

total amount (20 μl) was discharged into the MS source in two subsequent

steps with roughly the same volume. This resulted in two separate ‘mass

calibrant peaks’ (figure S3), where the former was measured with MS

methodology optimized for a low mass range (circa 100 - 715 m/z, fig. S4),

and the latter with settings optimized for a high mass range (circa 450 - 1050

m/z, fig. S5). Depending on the m/z range of interest, mass data were

recalibrated either with the first or second mass calibrant peak, allowing a

wide mass range to be monitored during one injection.

Fig. D.3.1. Chromatogram relative to the mass calibration solution. The peak at Rt 0.12 min was used for low masses range whereas the one at 0.4 min was used for the high masses range.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Time [min]

0

1

2

3

5x10

Intens.

20150306-silica28 3-1_1-37_01_3158.d: EIC 537.0206±0.05 -All MS

137

Fig. D.3.2. MS spectra relative to the low-masses (m/z) calibration solution.

Fig. D.3.3. MS spectra relative to the high-masses (m/z) calibration solution.

141.0169

223.0200

305.0230

387.0262

469.0291

551.0323

633.0353

715.0384

20150306-solvent-8_1-1_01_3237.d: -MS, 0.1-0.1min #10-12

0

1

2

3

4

5

5x10

Intens.

100 200 300 400 500 600 700 800 900 m/z

551.0382

633.0394

715.0401

797.0424

879.0447

961.0480

1043.0528

20150306-solvent-8_1-1_01_3237.d: -MS, 0.3-0.3min #35-36

0

1

2

3

5x10

Intens.

300 400 500 600 700 800 900 1000 1100 m/z

138

D.4. MS and Auto MS/MS settings

Table D.8.1. Q-TOF spectrometer parameters during the analysis. The chromatograms were split into four segments, the first two dedicated to mass calibration and the latter two to the analysis of the fullerenes and their transformation products.

Table D.8.2. Q-TOF spectrometer parameters in the Auto MS/MS analysis

139

Fig. D.5. An example of the chromatograms obtained during HPLC-UV analysis. C60 and C70 are eluting at 7.5 and 12.5 min respectively. No additional peaks were present in the sample.

Fig. D.6. Mass spectrum showing the presence of C60O and C60O3 traces in correspondence of the C60-II peak at retention time 20.0 min. (14 days sample of the glass incubation).

140

Fig. D.7. An example of the AutoMSMS analysis in the present work. (A) Auto MSMS detection of m/z 831.0251 (purple line) and 784.0115 (blue line) in the chromatograms, the grey line represents the base peak chromatogram obtained from the m/z 831.0261 trace. Below the MS2 spectra relative to the m/z 831.0251 (in purple, B) and m/z 784.0115 (in blue, C).

141

Fig. D.8. Chromatograms of the spiking solution (top) and a non-spiked glass sample extract (bottom) were the extracted ion chromatograms (EIC) of C60 and all the unknowns are highlighted.

Fig. D.9. An example of an unknown structure that showed increasing concentration in the non-irradiated samples (red line) but not in the irradiated ones (blue line). (m/z 860.0772, retention time 14.4 min, sandy soil incubation).

142

143

Fullerenes nanoparticles in soil:

Analysis, occurrence and fate

Summary

Fullerenes are carbon-based nanomaterials that can occur in the

environment due to both natural events and human production. Once

released, little is known about the fate of these chemicals and the

assessment is complicated by the large number of species that may occur

and the low concentrations at which they are present. Thus, the aim of this

work was to study the occurrence and fate of these chemicals in the soil

compartment, which is expected to act as a sink. The research consisted of

two parts. First, analytical methods were developed for the analysis of

fullerenes in soil samples, then these methodologies were applied in

environmental survey and fate studies.

The developed methodologies made use of a combination of ultra-

sonication and shaking extraction techniques. This resulted in good

recoveries for both the fullerenes and the functionalized structures and

allowed the processing of large batches of samples in relatively short time

and low energy costs. High performance liquid chromatography (HPLC) was

extensively investigated for the separation of mixtures of fullerenes and

was optimized for the analysis of complex soil matrices. Two stationary

phases, namely pyrenyl-propyl silica and core-shell biphenyl were tested

that can grant good retention of the fullerenes and sub-optimal separation

of similar structures. Eventually the methodologies were optimized with

regard to the detection techniques. Although the fullerenes displayed

strong absorbance in the UV range, mass spectrometric detection provided

144

the selectivity and sensitivity characteristics needed for the analysis of more

species and at environmentally relevant concentrations. Thus, most of the

analyses in this study relied on high resolution mass spectrometry (HRMS)

coupled to a heated-electrospray ionization (H-ESI) interface that allowed

unambiguous determination of the fullerenes at concentrations below the

parts per billion range.

The analysis of soil samples collected in The Netherlands showed that

fullerenes are extensively occurring in the environment and that C60 is the

most present and abundant specie. Furthermore, their origin is likely

anthropogenic and originates from emissions of traffic and combustion

processes (by-product), whereas the contribution of engineered

nanomaterials (ENMs) is expected to be comparatively low. Subsequently,

incubation studies showed that, once released in the environment,

fullerenes will interact with the solar radiation and will be degraded as a

result. However, once deposited onto soil, their fate will likely be

determined by other factors such as the action of microorganisms. The

degradation of the fullerenes will follow an oxidative pathway with the

possible formation of a large number of transformation products that will

presumably lead to the fragmentation of the closed-cage structure.

The research work in this thesis provided valuable tools for the analysis of

fullerenes in environmental samples and pioneering observations about

their occurrence and fate in the environment.

145

Samenvatting

Fullerenen zijn uit koolstof opgebouwde nanomaterialen die door zowel

natuurlijke processen als kunstmatige productie in het milieu voor kunnen

komen. Over het lot van deze stoffen in het milieu is nog weinig bekend en

de studie hiervan is gecompliceerd gezien de vele chemische vormen en

veelal lage concentraties waarin deze stoffen aanwezig kunnen zijn.

Verwacht wordt dat fullerenen accumuleren in de bodem en het doel van

deze studie was daarom ook om de aanwezigheid en het lot van deze

stoffen in de bodem te onderzoeken. Het onderzoek was tweeledig:

allereerst werd een analytische methode ontwikkeld om fullerenen in

bodem te meten, die daarna werd toegepast om de aanwezigheid en het

lot in het milieu te monitoren.

De ontwikkelde methode voor monsteropwerking bestaat uit een

combinatie van extractie met ultrasoon en uitschudden. Dit resulteerde in

bijna volledige extractie voor zowel fullerenen met en zonder functionele

groep en de methode is geschikt voor de verwerking van hoge aantallen

monsters binnen een relatief korte tijd en met lage (energie)kosten. High

Performance Liquid Chromatography (HPLC) werd uitgebreid getest en

toegepast voor de chromatografische scheiding van fullerenen en de

methode werd geoptimaliseerd voor de analyse van deze stoffen in

complexe bodem matrices. De stationaire fases pyrenyl-propyl silica en

core-shell biphenyl werden getest en dit resulteerde in beide gevallen in

volledige scheiding van fullerenen en gedeeltelijke scheiding van

vergelijkbare componenten. Uiteindelijk werden de detectiemethoden

geoptimaliseerd. Hoewel fullerenen sterke absorptie in het UV bereik

vertonen, werd alleen met massaspectrometrische detectie de selectiviteit

en gevoeligheid behaald welke nodig zijn voor de analyse van deze en

vergelijkbare componenten bij in het milieu voorkomende concentraties.

Om die reden zijn de meeste analyseresultaten in deze studie verkregen

met hoge resolutie massa spectrometrie (HRMS) gekoppeld met een

heated-electrospray ionizatie (H-ESI) interface welke eenduidige bepaling

146

van fullerenen met concentraties beneden het delen per miljard (ppb)

niveau toelaten.

Uit de analyse van in Nederland verzamelde bodemmonsters bleek dat

fullerenen op grote schaal voorkomen in het milieu, waarbij C60 het meeste

voorkomt en in de hoogste concentraties. Het is zeer waarschijnlijk dat deze

stoffen van antropogene herkomst zijn. Aangenomen wordt dat de

oorsprong voornamelijk ligt bij uitlaatgassen (bijvoorbeeld van auto’s en

vrachtwagens) of als bijproduct van andere verbrandingsprocessen

(bijvoorbeeld verbrandingsovens) en de in de bodem aangetroffen

fullerenen in mindere mate afkomstig zijn van geproduceerde

nanomaterialen. Door de uitgevoerde incubatie-experimenten werd

aangetoond dat fullerenen in het milieu worden afgebroken onder invloed

van zonlicht. Wanneer fullerenen in de bodem terecht komen zal de afbraak

door zonlicht echter in mindere mate optreden en zullen andere

afbraakprocessen, bijvoorbeeld door micro-organismen een grotere rol

spelen. De afbraak van fullerenen vindt waarschijnlijk plaats door een

oxidatieproces, waarbij mogelijk een groot aantal omzettingsproducten

gevormd worden als gevolg van het openbreken van de C60 structuur.

De onderzoeksresultaten die in dit proefschrift worden gepresenteerd

vormen een waardevol hulpmiddel voor de analyse van fullerenen in

bodem en bevatten unieke en innovatieve waarnemingen over het

voorkomen en het lot van deze stoffen in het milieu.

147

References

Astefanei A., Núñez O., Galceran M. T. 2014a. In: S.B. Ellis (Ed.), Fullerenes,

Chemistry, Natural Sources and Technological Applications, Nova Science

Publishers, New York, pp. 35-63.

Astefanei A., Núñez O., Galceran M. T. 2014b. Analysis of C60-fullerene

derivatives and pristine fullerenes in environmental samples by ultrahigh

performance liquid chromatography–atmospheric pressure

photoionization-mass spectrometry, Journal Chromatography A 1365, 61-

71. -63.

Avanasi R., Jackson W. A., Sherwin B., Mudge J. F., Anderson T. A. 2014. C60

Fullerene Soil Sorption, Biodegradation, and Plant Uptake. Environmental

Science and Technology, 48, 2792−2797. dx.doi.org/10.1021/es405306w.

Baena J. R., Gallego M., Valca´rcel M. 2002. Fullerenes in the analytical

sciences. Trends in Analytical Chemistry 21, 187-198.

Baierl B., Fabel B., Gabos P., Pancheri L., Lugli P., ScarpaG. 2010. Efficient

indium-tin-oxide (ITO) free top-absorbing organic photodetector with

highly transparent polymer top electrode. Organic Electronics 11, 1199–

1206.

Bakry R, Vallant R. M., Najam-ul-Haq M., Rainer M., Szabo Z., Huck C. W.,

Bonn G. K. 2007. Medicinal applications of fullerenes. International Journal

of Nanomedicine 2:4, 639-649.

Becker L., Bada J. L., Winans R. E., Hunt J. E., Bunch T. E. and French B.M.

1994. Fullerenes in the 1,85-billion-year-old Sudbury impact structure.

Science, 265: 642–645.

148

Benn T. M., Westerhoff P., Herckes P. 2011. Detection of fullerenes (C60 and

C70) in commercial cosmetics. Environmental Pollution. 159 (5), 1334-1342.

Bhatt I. and Triphathi B. N. 2011. Interaction of engineered nanoparticles

with various components of the environment and possible strategies for

their risk assessment. Chemosphere 82 (2011) 308–317

Bouchard D., Ma X. 2008. Extraction and high-performance liquid

chromatographic analysis of C60, C70, and [6,6]-phenyl C61-butyric acid

methyl ester in synthetic and natural waters. Journal Chromatography A,

1203, 153-159.

Boxall, A. B., Tiede, K., Chaudhry, Q. 2007. Engineered nanomaterials in

soils and water: How do they behave and could they pose a risk to human

health? Nanomedicine 2, (6), 919‐927.

Buhl M. and Hirsch A. 2001. Spherical aromaticity of fullerenes. Chemical

Reviews. 101, 1153-1183.

Burangulov N., Moravsky A.P., Kulikova Y.V., Loutfy R.O., Dyachuk G.I. 2005.

Cosmetic composition containing fullerene clusters. U.S. Patent Application

20050136079.

Buseck P.R., Tsipursky S. J., Hettich R. 1992. Fullerenes from the geological

environment. Science 257, 215-217.

Cantrill S. 2011. Endohedral fullerenes: Water behind walls. Nature

Chemistry 3, 753. doi:10.1038/nchem.1168

149

Carboni A., Emke E., Parsons J. R., Kalbitz K., de Voogt P. 2013. An analytical

method for determination of fullerenes and fullerene derivatives in soil with

high performance liquid chromatography and UV detection. Analitica

Chimica Acta 807, 159-165.

Carboni A., Helmus R., Parsons J. R., Kalbitz K., de Voogt P. 2016. A method

for the determination of fullerenes in soil and sediment matrices using

ultra-high performance liquid chromatography coupled with heated

electrospray quadrupole time of flight mass spectrometry. Journal of

Chromatography A. 1433, 123–130.

Chai, Y.,Guo T., Jin C., Haufler R. E., Chibante L. P., Felipe; Fure J., Wang L.,

Alford J.M., Smalley R.E. 1991. Fullerenes with metals inside. Journal of

Physical Chemistry 95 (20), 7564–7568.

Chao T.-C., Song G., Hansmeier N., Westerhoff P., Herckes P., Halden R. U.

2011. Characterization and Liquid Chromatography-MS/MS Based

Quantification of Hydroxylated Fullerenes. Analytical Chemistry 83, 1777–

1783. dx.doi.org/10.1021/ac1031379

Christian P., Von der Kammer F., Baalousha M., Hofmann Th. 2008.

Nanoparticles: structure, properties, preparation and behaviour in

environmental media. Ecotoxicology 17:326–343

Creegan, K. M., Robbins, J. L., Robbins, W. K., Millar, J. M., Sherwood, R. D.,

Tindall, P. J., Cox, D. M. 1992. Synthesis and characterization ofC60O, the

first fullerene epoxide. Journal American Chemical Society 114, 1103–1105.

Daly T. K., Buseck P. R., Williams P., Lewis C. F. 1993. Fullerenes from a

fulgurite. Science 259, 1599–1601.

150

Dang M. T., Hirsch L., Wantz G. 2011. P3HT:PCBM, Best Seller in Polymer

Photovoltaic Research. Advanced Materials 23, 3597-3602.

Deguchi S., Alargova G. R., and Tsujii K. 2001. Stable Dispersions of

Fullerenes, C60 and C70, in Water. Preparation and Characterization.

Langmuir 17, 6013-6017.

Deye J., Shiveley A. N., Oehrle S. A., Walters K. A.. 2008. Separation of

Substituted Fullerenes using Normal-Phase HPLC-MS. Journal

Chromatography A 1181, 159–161.

Diederich F. 1997. Covalent fullerene chemistry. Pure and applied chemistry

69:3, 395-400.

Ebbesen T. W., Lezec H. J., Hiura H., Bennett J. W., Ghaemi H. F., Thio T.

1996. Electrical conductivity of individual carbon nanotubes. Nature, 382,

54‐56.

Elsila J.E., de Leon N.P., Plows L., Buseck P.R., and Zare R.N. 2005 Extracts of

impact breccia samples from Sudbury, Gardnos, and Ries impact craters and

the effects of aggregation on C60 detection. Geochimica Cosmochimica Acta

69, 2891–2899.

Emke E., Sanchís J., Farré M., Bäuerlein P. S., de Voogt P. 2015.

Determination of several fullerenes in sewage water by LC HR-MS using

atmospheric pressure photoionisation. Environmntal Science Nano 2, 167-

176. doi: 10.1039/C4EN00133H.

151

European Commission, Commission Directive 200/90/EC. 2009. laying

down technical specifications for chemical analysis and monitoring of water

status, Official Journal of the European Union L201, 36-38.

European Commission, Commission recommendation of 18 October 2011

on the definition of nanomaterial (2011/696/EU). Official Journal of the

European Union 2011, 54, 38‐40.

Farre M., Perez S., Gajda-Schrantz K., Osorio V., Kantiani L., Ginebreda A.,

Barcelo D. 2010. First determination of C60 and C70 fullerenes and N-

methylfulleropyrrolidine C60 on the suspended material of wastewater

effluents by liquid chromatography hybrid quadrupole linear ion trap

tandem mass spectrometry, Journal of Hydrology 383, 44-51.

Gago-Ferrero P., Schymansky E. L., Bletsou A. A., Aalizadeh R., Hollender J.,

Thomaidis N. S. 2015. Extended Suspects and Non-Target Strategies to

Characterize Emerging Polar Organic Contaminants in Raw Watewater with

LC-HRMS/MS, Environmental Science and Technology 49, 12333-12341.

Goel A., Howard J. B., J. B. Vander Sande. 2004. Size analysis of single

fullerene molecules by electron microscopy. Carbon 42, 1907-1915.

Gottschalk F., Sonderer T., Scholz R. W. 2009. Modeled Environmental

concentrations of engineered nanomaterials (TiO2, ZnO, Ag, CNT,

Fullerenes) for different regions, Environmental Science and Technology 43,

9216–9222.

Gottschalk F., Soonderer T., Scholz R. W., Nowack B. 2010. Possibilities and

limitations of modelling environmental exposure to engineered

152

nanomaterials by probabilistic material flow analysis. Environmental

Toxicology and Chemistry, Vol. 29, No. 5, pp. 1036–1048.

Guldi D.M., Martin N. 2002. Fullerenes: From Synthesis to Optoelectronic

Properties, Kluwer Academic Publishers, Norwell, MA.

Haftka J. J.-H., Bäuerlein P. S., Emke E., Lammertse N., Belokhovstova D.,

Hilvering B., de Voogt P., ter Laak T. L. 2015. Colloidal stability of

(functionalised) fullerenes in the presence of dissolved organic carbon and

electrolytes. Environmental Science Nano 2, 280-287. DOI:

10.1039/C4EN00206G.

Hassellov M., Readman J. W., Ranville J .F., Tiede K. 2008. Nanoparticle

analysis and characterization methodologies in environmental risk

assessment of engineered nanoparticles. 2008. Ecotoxicology 17, 344–361.

DOI 10.1007/s10646-008-0225-x

Hendren, C. O., Mesnard, X., Dröge, J., Wiesner, M. R. 2011. Estimating

Production Data for Five Engineered Nanomaterials as a Basis for Exposure

Assessment. Environmental Science and Technology 45, (7), 2562–2569.

Henglein A. 1993. Physicochemical Properties of Small Metal Particles in

Solution – Microelectrode Reactions, Chemisorption, Composite Metal

Particles, and the Atom-To-Metal Transition. Journal of Physical Chemistry

97, 5457–5471.

Herrero P., Bauerlein P. S., Emke E., Pocurull E., de Voogt P. 2014.

Asymmetrical flow field-flow fractionation hyphenated to Orbitrap high

resolution mass spectrometry for the determination of (functionalised)

aqueous fullerene aggregates. Journal of Chromatography A 1356, 277-282.

153

Heymann D., Dressler B.O., Knell I., Thiemens M.H., Buseck P.R., Dunbar

R.B., Mucciarone D. 1999. Origin of carbonaceous matter, fullerenes, and

elemental sulfur in rocks of the Whitewater Group, Sudbury impact

structure, Ontario, Canada. In Large Meteorite Impacts and Planetary

Evolution II, Special Paper, vol. 339, Dressler, B. O. and Sharption, V. L.

(eds.), 345–360. Boulder, CO: Geological Society of America.

Heymann D., Chibante L.P.F., Brooks R.R., Wolbach W.S., Smalley R.E. 1994.

Fullerenes in the cretaceous-tertiary boundary-layer. Science 265, 645–647.

Hirsch A. 2005. The Chemistry of Fullerenes, Wiley-VCH, Weinheim,

Germany.

Hou W.-C., Jafvert C. 2009. Photochemical transformation of aqueous C60

clusters in sunlight. Environmental Science and Technology 43, 362-367.

Howard J.B., Mc Kinnon J.T., Makarovsky Y., Lafleur A.L., Johnson M.E. 1991.

Fullerenes C60 and C70 in flames, Nature 352, 139 – 141.

Hummelen J.C., Knight B.W., Lepeq F., Wudl F., Yao J., Wilkins C.L. 1995.

Preparation and Characterization of Fulleroid and Methanofullerene

Derivatives, Journal of Organic Chemistry 60, 532-538.

Huppertz L. M., Kneisel S., Auwarter V., Kempf J. 2014. A comprehensive

library-based, automated screening procedure for 46 synthetic

cannabinoids in serum employing liquid chromatography-quadrupole ion

trap mass spectrometry with high-temperature electrospray ionizaton,

Journal of Mass Spectrometry 49, 117-127.

154

Hwang Y. S., Li Q. 2010. Characterizing photochemical transformation of

aqueous nC60 under environmentally relevant conditions. Environmental

Science and Technology 44, 3008-3013.

Isaacson C. W., Kleber M., Field J.A. 2009. Quantitative Analysis of Fullerene

Nanomaterials in Environmental Systems: A Critical Review. Environmental

Science and Technology 43 (17), 6463-6474.

Jafvert C. T., Kulkarni P. P. 2008. Buckminsterfullerene’s (C60) octanol-water

partition coefiicient (Kow) and aqueous solubility. Environmental Science

and Technology 42, 5945-5950.

Jehlicka J., Frank O. , Hamplova´ V., Pokorna´ Z., Juha L., Bohacek Z.,

Weishauptova Z. 2005. Low extraction recovery of fullerene from

carbonaceous geological materials spiked with C-60. Carbon 43, 1909–

1917.

Kempf J., Traber J., Auwarter V., Huppertz L. M. 2014. ‘Psychotropics caught

in a trap’ – Adopting a screening approach to specific needs, Forensic

Science International 243, 84-89.

Kim Y., Cook S., Tuladhar S. M., Choulis S. A., Nelson J., Durrant J. R., Bradley

D. D. C., Giles M., McCulloch I., Ha C., Ree M. 2006. A strong regioregularity

effect in self-organizing conjugated polymer films and high-efficiency

polythiophene: fullerene solar cells. Nature Materials 5, 197–203.

Klaine S. J., Alvarez P. J. J., Batley, G. E., Fernande T. F., Handy R. D., Lyon D.

Y., Mahendra, S., McLaughlin M. J., Lead J. R. 2008. Nanomaterials in the

environment: Behavior, Fate, Bioavailability, and Effects. Environmental

Toxicology and Chemistry 27, (9), 1825‐1851.

155

Kolkman A., Emke E., Bauerlein P. S., Carboni A., Tran D. T., ter Laak T. L.,

van Wezel A. P., de Voogt P. 2013. Analysis of (functionalized) fullerenes in

water samples by liquid chromatography coupled to high-resolution mass

spectrometry, Analytical Chemistry 85, 5867–5874.

Kroto, H. W., Heath, J. R., O’Brien, S. C., Curl, R. F. and Smalley, R. E. 1985.

C60: Buckminsterfullerene. Nature, 318, pp.162-163.

Kroto H. W., Taylor R., Walton D.R.M. 1994.The structure and reactivity of

C60. Pure & Applied Chemistry 66, 2091-2094.

Laitinen T., Petaja T., Backman J., Hartonen K., Junninen H., Ruiz-Jimenez J.,

Worsnop D., Kulmala M., Riekkola M.-L. 2014. Carbon clusters in 50 nm

urban air aerosol particles quantified by laser desorption-ionization aerosol

mass spectrometer. International journal of mass spectrometry 358, 17-24.

Lee J., Cho M., Fortner J. D.,Hughes J. B., Kim J.-H. 2009. Transformation of

Aggregated C60 in the Aqueous Phase by UV Irradiation. Environmental

Science and Technology 43, 4878-4883.

Lenes M., Wetzelaer G. A. H., Kooistra F. B. , Veenstra S. C. , Hummelen J.

C., Blom P. W. M. 2008. Fullerenes bisadducts for enhanced open-circuit

voltages and efficiencies in polymer solar cells. Advanced Materials 20,

2116-2119.

Li C., Yipab H., Jen A. K. Y. Jen. 2012. Functional fullerenes for organic

photovoltaics. Journal of Materials Chemistry 22, 416.

156

Lin S.-K., Lin S., Shiu L., Chien K., Luh T., Lin T. 1995. Fluorescence of fullerene

derivatives at room temperature. Journal of Physical Chemistry 99, 105 –

111.

Lomas S. 2015. Identifying the Increased Scope of Core-Shell Technology for

HPLC and UHPLC Chromatographers, Chromatography Today (Feb/Mar

2015) 43-44.

Lux Research. 2004. The Nanotech Report 2004. New York: Lux Research.

Marchesan S., Da Ros T., Prato M. 2005. Isolation and characterization of

nine tris-adducts of N-methylfulleropyrrolidine derivatives. Journal of

Organic Chemistry 70, 4706–4713.

Mauter M. S. and Elimelech M., 2008. Environmental Applications of

Carbon-Based Nanomaterials. Environmental Science and Technology 42

(16), 5843–5859. DOI: 10.1021/es8006904

Maynard A. D. 2006. Nanotechnology: A Research Strategy for Addressing

Risk. WoodrowWilson International Center for Scholars,Washington, DC.

Maynard A. and Michelson E. 2012. Woodrow Wilson International Centre

for Scholars, The project on Emerging Nanotechnologies. Consumer

Products Inventory of Nanotechnology Products.

http://www.nanotechproject.org/inventories/consumer (5‐12‐2012)

Moussa F., Trivin F., Ce´olin R., Hadchouel M., Sizaret P.Y., Greugny V., Fabre

C., Rassat A., Szwarc H. Early effects of C60 admininstration in swiss mice: a

157

preliminary account for in vivo C60 toxicity. 1997. Fullerenes Nanotubes and

Carbon Nanostructures 21–29.

Mueller N. C., Nowack B. 2008. Exposure modeling of engineered

nanoparticles in the environment, Environmental Science and Technology

42, 4447–4453.

Murayama H., Tomonoh S., Alford J.M., Karpuk E., 2004. Fullerene

production in tons and more: from science to industry. Fullerenes

Nanotubes and Carbon Nanostructures 12, 1-9.

Murr L. E., Bang J. J., Esquivel E. V., Guerrero P. A., Lopez D. A., 2004. Carbon

nanotubes, nanocrystal forms, and complex nanoparticle aggregates in

common fuel-gas combustion sources and the ambient air. Journal of

Nanoparticle Research 6: 241–251.

Navarro D.A., Kookana R.S., Kirby J.K., Martin S.M., Shareef A., Du J., Mc

Laughlin M.J. 2013. Behaviour of fullerenes (C60) in the terrestrial

environment: potential release from biosolids-amended soils, Journal of

Hazardous Materials 262, 496-503.

Nowack B., Bucheli T. D., 2007. Occurrence, behaviour and effects of

nanoparticles in the environment. Environmental Pollution 150, (1), 5‐22.

Núñez O., Gallart-Ayala H., Martins C. P., Moyano E., Galceran M. T. 2012.

Atmospheric Pressure Photoionization Mass Spectrometry of Fullerenes,

Analytical Chemistry 84, 5316-5326.

158

Osawa, E., 2002. Perspectives of Fullerene Nanotechnology. Kluwer

Academic Publisher, Dordrecht.

Owen, R., Handy R.. 2007. Viewpoint: Formulating the Problems for

Environmental Risk Assessment of Nanomaterials. Environmental Science

and Technology 41, (16), 5582‐5588.

Pakaninen K. 2013. Carbon nanoparticles in aquatic environments: Fate of

fullerenes (C60) in freshwaters and their effects on organisms. Publication of

the University of Eastern Finland.

Pal T., Sau T. K., Jana N. R. 2010. Reversible Formation and Dissolution of

Silver

Nanoparticles in Aqueous Surfactant Media. Langmuir, 13, (6), 1481‐1485.

Pan B., Xing B. 2012. Application and Implication of manufactured

nanoparticles in soils: a review. European journal of soil science 63, 437-

456.

Panina L. K., Kurochkin V. E., Bogomolova E. V., Evstrapov A. A., Spitsyna N.

G. 1997. Biotransfromation of fullerenes. Bocklady Biological Sciences

357,530-532.

Parthasarathy G., Bhandari N., Vairamani M., Kunwar A.C., 2008. High-

pressure phase of natural fullerene C60 in iridium-rich Cretaceous-Tertiary

boundary layers of Deccan intertrappean deposits, Anjar, Kutch, India.

Geochimica Cosmochimica Acta 72, 978-987.

159

Peralta-Videa J. R., Zhao L., Lopez-Moreno M. L., de la Rosa G., Hong J,

Gardea-Torresdey J. L. 2011. Nanomaterials and the environment: A review

for the biennium 2008–2010 Journal of Hazardous Materials 186 1–15

Perez R. A., Albero B., Miguel E., Tadeo J. L., Sanchez-Brunete C. 2013. A

Rapid Procedure for the Determination of C60 and C70 Fullerenes in Soil and

Sediments by Ultrasound-assisted Extraction and HPLC-UV. Analytical

Science 29, 533-538.

Piccinno F., Gottschalk F., Seeger S., Nowack B. 2012. Industrial production

quantities and uses of ten engineered nanomaterials in Europe and the

world. Journal Nanoparticles Research 14, 1109

Popescu L. M., van’t Hof P., Sieval A. B., Jonkman H. T., Hummelen C.. 2006.

Thienyl analog of 1-(3-methoxycarbonyl)propyl-1-phenyl-[6,6]-

methanofullerene for bulk heterojunction photovoltaic devices in

combination with polythiophenes. Applied Physics Letters 89, 213507.

Powell B. R., Bloink R. L., Eickel C. C. 1988. Preparation of Cerium Dioxide

Powders for

Catalyst Supports. Journal American Ceramic Society, 71 (2), C‐104‐C‐106.

Pycke B. F. G., Chao T.-Z., Herckes P., Westerhoff P., Halden R. U. 2012.

Beyond nC60: strategies for identification of transformation products of

fullerene oxidation in aquatic and biological samples. Analytical

Bioanalytical Chemistry, DOI 10.1007/s00216-012-6090-8.

Reed M. A. 1993. Quantum Dots Scientific American. 268, (1), 118‐123.

160

Roco, M. C., Bainbridge, W. S. 2005. Societal implications of nanoscience

and nanotechnology: Maximizing human benefit. Journal of nanoparticles

Research 7(1):1-13.

RS & RAE, Royal Society and Royal Academy of Engineering. 2004.

“Nanoscience and nanotechnologies: opportunities and uncertainties”.

Ruoff R. S., Tse D. S., Malhotra R., Lorents D. C. 1993. Solubility of C60 in a

variety of solvents. Journal Physical Chemistry 97, 3379-3383.

Saito Y., Ohta H., Jinno K. 2004. Chromatographic separation of fullerenes,

Analytical Chemistry 76, 266A-272A.

Sanchis J., Berrojalbiz N., Caballero G., Dachs J., Farre M., Barcelo D. 2012.

Occurrence of Aerosol-Bound Fullerenes in the Mediterranean Sea

Atmosphere, Environmental Science and Technology 46, 1335-1343.

Sanchis J., Bozovic D., Al-Harbi N. A., Silva L. F., Farre M., Barcelo D. 2013.

Quantitative trace analysis of fullerenes in river sediment from Spain and

soils from Saudi Arabia, Analytical Bioanalytical Chemistry 405, 5915-5923.

Sanchis J., Oliviera L.F., Leão F.B., Farre M., Barcelo D. 2015. Liquid

chromatography–atmospheric pressure photoionization–Orbitrap analysis

of fullerene aggregates on surface soils and river sediments from Santa

Catarina (Brazil), Science Total Environment 505, 172-179.

Shareef A., Li G., Kookana R. S. 2010. Quantitative determination of

fullerene (C-60) in soils by high performance liquid chromatography and

161

accelerated solvent extraction technique. Environmental Chemistry 7, 292-

297.

Song L., Wellman A.D., Huifang Y., Adcock J. 2007. Electron capture

atmospheric pressure photoionization mass spectrometry: analysis of

fullerenes, perfluorinated compounds, and pentafluorobenzyl derivatives,

Rapid communications mass spectrometry 21, 1343-1351.

Sun, Y. P., Ma, B., Bunker, C. E., Liu, B. 1995. All-cabon polymers

(polyfullerenes) from photochemical reactions of fullerene clusters in

room-temperature solvent mixtures. Journal American Chemical Society

117, 12705–12711.

Tagmatarchis N. and Shinohara H. 2001. Fullerenes in medicinal chemistry

and their biological applications. Mini-Reviews in Medicinal Chemistry 1 (4),

339–348(MA).

Taylor, R., Parsons, J. P., Avent, A. G., Rannard, S. P., Dennis, T. J., Hare, J. P.,

Kroto, H. W., Walton, D. R. 1991. Degradation of C60 by light. Nature 351,

277.

Taylor R., Barrow M. P., Drewello T. 1998. C60 degrades to C120O. Chemical

Communications, 2497-2498. DOI: 10.1039/A806726K

Tiwari J.A., Morris J.R., Vejerano E.P., Hochella M.F., Marr L.C., 2014.

Oxidation of C60 aerosol by atmospherically relevant levels of O3.

Environmental Science and Technology 48, 2706-2714.

162

Tiwari J.A., Ashraf-Khorassani M., Marr L. C. 2016. C60 fullerenes from

combustion of common fuels. Science of the Total Environment 547, 254–

260

Tiwari S. P., Namdas E. B., Rao V. R., Fichou D., Mhaisalkar S. G. 2007.

Solution-Processed n-Type Organic Field-Effect Transistors With High on

/off Current Ratios Based on Fullerene Derivatives. IEEE Electron Device

Letters 28, 880-883.

Utsunomiya S., Jensen K. A., Keeler G. J., Ewing R. C., 2002. Uraninite and

fullerenes in atmospheric particulates. Environmental Science and

Technology 36 (23), 4943-4947.

Van der Ploeg M. J. C., Baveco J. M., van der Hout A., Bakker R., Rietjens I.

M. C. M., van den brink N. W. 2011. Effects of C60 nanoparticle exposure on

earthworms (Lumbricus rubellus) and implications for population dynamics.

Environmental Pollution 159, 198-203.

van Wesemael J.C. 1955. De bepaling van het calciumcarbonaatgehalte van

gronden, Chem Weekblaad 51, 35-36.

van Wezel A. P., Morinière V., Emke E., ter Laak T., Hogenboom A. C. 2011.

Quantifying summed fullerene nC60 and related transformation products in

water using LC LTQ Orbitrap MS and application to environmental samples,

Environment International 37, 1063–1067.

Vítek P., Jehlička J., Frank O., Hamplová V., Pokorná Z., Juha L., Boháček L.

2009. Optimizing Conditions for Ultrasound Extraction of Fullerenes from

Coal Matrices. fullerenes nanotubes carbon nanomaterials 17, 109-122.

163

Wang C., Shang C., Westerhoff P. 2010. Quantification of fullerene

aggregate nC60 in wastewater by high-performance liquid chromatography

with UV-vis spectroscopic and mass spectrometric detection. Chemosphere

80, 334–339.

Wang J., Cai Q., Fang Y., Anderson T.A., Cobb G.P. 2011. Determination of

fullerenes (C60) in artificial sediments by liquid chromatography. Talanta 87,

35–39.

Wienk M. M., Kroon J. M., Verhees W. J. H., Knol J., Hummelen J. C., van Hal

P. A., Janssen R. A. 2003. Efficient methano[70]fullerene/MDMO-PPV bulk

heterojunction photovoltaic cells. Angewandte Chemie International

Edition, 42, 3371-3375.

Wijnhoven S. W. P., Dekkers S., Kooi M., Jongeneel W. P., Jong W. H. 2010.

Nanomaterials in consumer products. Update of products on the European

market in: National Institute for Public Health and the Environment: 2010.

Wood, J. M., Kahr, B., Hoke, S. H., Dejarme, L., Cooks, R. G., Ben-Amotz, D.

1991. Oxygen and methylene adducts of C60 and C70. Journal American

Chemical Society 113, 5907–5908.

WRB, World Reference Base for Soil Resources, FAO, 2006.

Xia X. R., Monteiro-Riviere N. A., Je R. 2006. Trace analysis of fullerenes in

biological samples by simplified liquid–liquid extraction and high-

performance liquid chromatography. Journal Chromatography A, 1129,

216–222.

164

Yadav B.C. and Kumar R. 2008. Structure, properties and applications of

fullerenes International Journal of Nanotechnology and Applications ISSN

0973-631X Volume 2, Number 1, pp. 15–24

Zhang K., Xu C. H., Wang C. Z., Chan C. T., Ho K. M. 1992.Systematic study

of structures and stabilities of fullerenes. Physical Review B 46, 11, 7333-

7336.

Zhao Y., Fang Y., Jian Y. 2006. Fluorescence study of fullerene in organic

solvents at room temperature. Spectrochimica Acta A 64, 564–567.

165

List of papers used in this thesis

I. Carboni A., Emke E., Parsons J. R., Kalbitz K., de Voogt P. 2013. An analytical method for determination of fullerenes and fullerene derivatives in soil with high performance liquid chromatography and UV detection. Analitica Chimica Acta 807, 159-165. Sample collection: A. Carboni Standard preparation: A. Carboni, E. Emke LC-UV/FLU analysis: A. Carboni Writing: A. Carboni Supervision and reviewing: J.R. Parsons, K. Kalbitz, P. de Voogt

II. Carboni A., Helmus R., Parsons J. R., Kalbitz K., de Voogt P. 2016. A

method for the determination of fullerenes in soil and sediment matrices using ultra-high performance liquid chromatography coupled with heated electrospray quadrupole time of flight mass spectrometry. Journal of Chromatography A, 1433, 123–130. Sample collection: A. Carboni Standard preparation: A. Carboni LC-MS analysis: A. Carboni, R. Helmus Writing: A. Carboni Supervision and reviewing: J.R. Parsons, K. Kalbitz, P. de Voogt

III. Carboni A., Helmus R., Emke E., van den Brink N., Parsons J.R., Kalbitz K., de Voogt P. Analysis of fullerenes in soils samples collected in The Netherlands (Environmental Pollution, in revision) Sample collection: A. Carboni, N. van den Brink, E. Emke Standard preparation: A. Carboni, E. Emke LC-MS analysis: A. Carboni, R. Helmus Writing: A. Carboni Supervision and reviewing: J.R. Parsons, N. van den Brink, K. Kalbitz, P. de Voogt

IV. Carboni A., Helmus R., Parsons J. R., Kalbitz K., de Voogt P. Incubation of solid state C60 fullerene under environmentally relevant conditions (submitted to Chemosphere) Experiments: A. Carboni LC-MS analysis: A. Carboni, R. Helmus Writing: A. Carboni Supervision and reviewing: J.R. Parsons, K. Kalbitz, P. de Voogt

166

Curiosities

Fullerenes were named in homage to Sir Richard Buckminster Fuller (1895

– 1983), the inventor of the geodesic domes which resemble these

molecules. For this C60 is commonly named Buckminsterfullerene or even

Buckyball. Sir Fuller was likely among the most interesting personalities of

the last century and one of the first concerned about sustainability and

human survival under the existing socio-economic system (yet he was

optimistic about the future of humans). In addition to being an architect,

systems theorist, designer and inventor (28 patents) Sir Fuller was also an

excellent writer (he published more than 30 books) and author of some

amazing quotes like the one at the beginning of this book. My absolute

favourite is:

“When I am working on a problem, I never think about beauty but when I

have finished, if the solution is not beautiful, I know it is wrong.”

R. Buckminster Fuller

The size ratio between fullerenes (C60) and a soccer ball is similar to that

between a soccer ball and the planet Earth.

Fullerenes are the most symmetric molecules in the world. There are 120

symmetry operations, like rotations around an axis or reflections in a plane,

which map the molecule onto itself. This makes C60 the molecule with the

largest number of symmetry operations.

167

Like many other interesting things, the discovery of fullerenes happened by

chance. It seems that the first 3d structure was obtained by the authors

while playing with paper hexagons and pentagons, basically an origami.

The discovery of fullerenes (in 1985), a new allotrope of carbon out of the

blue, was quite an event the scientific community and the discoverers will

receive a Nobel Prize for it (in 1996). However, at that time the possible

applications and consequences for humans were still unknown and

argument of debate. Shortly after the discovery a member of the British

House of Lords remarkably commented “It (C60) does nothing in particular

and does it very well”.

Considering that it was published on Nature, was cited some thousands

times, it won a Nobel prize and basically started a new entire field of

research, the first publication about fullerenes was pretty amazing. The title

is short and sweet (‘C60:Buckminsterfullerene’), the paper runs to less than

two pages, there is a photograph of a soccer ball but no supplementary

information, and in places the text is written in an informal style that is hard

to imagine appearing in a journal today.

Even though C60 is relatively soft under normal conditions it can be

compressed at 320 000 atmospheric pressure to create a substance so hard

it can dent a diamond (it could be the hardest material on Earth).

When a molecule of C60 is attached to twelve molecules of nitrous oxide,

the resulting structure can explode in a controlled reaction. Surprisingly, the

main application of this “Buckybomb” could be medical.

According to the rules for making icosahedron, an infinite number of

(always larger) fullerenes can exist. The smallest is a C20 (actually the

smallest is a C28…C20 is unstable). In addition, you also get “Buckyonions”,

168

i.e. multi-layered round carbon particles where fullerenes are one into the

other like an onion peel.

The solubility of fullerenes in certain solvents is weird and maxed at room

temperature. It’s something like putting sugar in your hot tea and see it

dissolving while the tea cools down.

There’s a rock (actually a mineraloid) relatively rich in fullerenes. It’s called

Shungite, has a 98 weight percent of carbon, it’s very black, relatively rare

and mostly found in Russia.

Fullerenes are also weird to work with in the lab. When dissolved into

solvents (which is something I did on a daily basis while working at the

experiments in this thesis) different molecules gives different colours in

different solvents. For instance, C60 is violet in toluene and yellow in water.

C70 instead is dark red in toluene and pink into water.

About nanoparticles in general, nanotechnology is the (recent) ability to

manipulate matter at the nano size. However, ancient populations have

used nanoparticles since thousands years ago, e.g. introducing them into

utensils in order to produce a glittering effect.

Nanoparticles sounds small but they are not that small. For instance, they

are much bigger that most of the other stuff you breath in the traffic

(oxygen, methane, NOXs, benzene, PAHs). However, for being particles,

they are absolutely small.

Nanoparticle can actually be pretty big. This is due to the current

terminology that defines nanoparticles to be so if they are very small (less

than 100 nm) in at least one dimension.

This means that a nano-wire could be very thin but also very very long. For

the same reason, a nano-sheet could be 1 nm thick but have a surface large

enough to wrap the planet. However, this is not likely to happen.

169

Acknowledgements

First, my deepest gratitude for my supervisors and promotors, Pim de

Voogt, Karsten Kalbitz and John R. Parsons for supporting my research and

trusting me and my ideas.

I gratefully aknowledge the Dutch government and the NanoNextNL project

which supported and funded my doctoral work.

I am thankful and grateful to all my colleagues that participated, inspired

and helped this project. Most especially Chiara Cerli, Joke Westerveld, Rick

Helmus Leo Hoitinga and Peter Serne at the UvA and Erik Emke, Annemarie

Kolkman and Patrick Bauerlein at the KWR.

Many thanks to my colleagues and friends Vittorio, Manuel, Christian and

Sebastian whom I can always rely on and who have been travel companions.

Many thanks also to the NanoNextNL and NanoSENSE colleagues, it was

wonderful to work with you.

The deepest thanks are to my family and loved ones, my partner Cinzia and

my son Leonardo.

170

Notes