The first order Raman spectrum of isotope labelled nitrogen-doped reduced graphene oxide

labelled nitrogen-doped reduced graphene oxide

Tobias Dahlberg

Tobias Dahlberg February 19, 2016

reduced graphene oxide using Raman spectroscopy. Specifically, the project set

out to investigate if the Raman active nitrogen-related vibrational modes of

graphene can be identified via isotope labelling. Previous studies have used

Raman spectroscopy to characterise nitrogen doped graphene, but none has

employed the method of isotope labelling to do so. The study was conducted by

producing undoped, nitrogen-doped and nitrogen-15-doped reduced graphene

oxide and comparing the differences in the first-order Raman spectrum of the

samples. Results of this study are inconclusive. However, some indications

linking the I band, a band previously speculated to be nitrogen or sp ³ carbon

related, to nitrogen functionalities are found. Also, a hypothetical Raman band

denoted I* possibly related to sp ³ hybridised carbon is introduced in the same

spectral area as I. This indication of a separation of the I band into two bands,

each dependent on one of these factors could bring clarity to this poorly under-

stood spectral area. As the results of this study are highly speculative, further

research is needed to confirm them and the work presented here serves as a

preliminary investigation.

Sammanfattning

Amnet som behandlas i denna avhandling ¨ ¨ ar studien av kv¨ avefunktionaliteter i

kv¨ avedopat reducerat grafenoxid med hj¨ alp av Ramanspektroskopi. Specifikt s˚ a

var m˚ alet med projektet att unders¨ oka om Ramanaktiva kv¨ averelaterade vib-

rationsmoder i grafen kunde identifieras via isotopm¨ arkning. I tidigare studier

har Ramanspektroskopi anv¨ ants f¨ or att karakterisera kv¨ avedopat grafen, men

isotopm¨ arkning har aldrig tll¨ ampats f¨ or detta ¨ andam˚ al. Studien genomf¨ ordes

genom framst¨ allning av odopat, kv¨ ave-dopat och kv¨ ave-15-dopat reducerat gra-

fenoxid vilkas Ramanspektra sedan j¨ amf¨ ordes. Resultaten av denna studie kan

inte ses som avg¨ orande d˚ a d˚ a datam¨ angden var f¨ or liten. Men vissa indikatio-

ner som f¨ orbinder I-bandet, ett band som tidigare spekluerats som relaterat till

kv¨ ave eller sp ³ hybridiserat kol, till kv¨ avefunktionaliteter hittades. Dessutom

introducerades ett hypotetisk Raman band, I*, i samma spektrala region som

I. Detta band anses m¨ ojligen vara relaterade till sp ³ hybridiserat kol. Denna

indikation av en separation av I bandet till tv˚ a band, vardera beroende av en

av dessa faktorer kan bringa klarhet till detta tidigare d˚ aligt f¨ orst˚ adda spektra-

la omr˚ ade. Eftersom resultaten av denna studie ¨ ar h¨ ogst spekulativa, beh¨ ovs

det ytterligare studier f¨ or att bekr¨ afta dem och det arbete som presenteras h¨ ar

fungerar enbart som en f¨ orstudie for eventuellt fortsatt arbete.

Contents

1 Introduction 3

2 Theory 5

2.1 Graphene . . . . 5

2.1.1 Electronic Properties . . . . 6

2.1.2 Vibrational Properties . . . . 6

2.1.3 Nitrogen Doping . . . . 8

2.1.4 Reduced Graphene Oxide . . . . 9

2.2 Characterization Methods . . . . 10

2.2.1 XPS . . . . 10

2.2.2 Raman Spectroscopy . . . . 12

2.2.2.1 Raman Spectra of Graphene . . . . 13

2.2.2.2 Isotope Labelling . . . . 15

3 Experimental 17 3.1 Synthesis . . . . 17

3.1.1 Synthesis of GO . . . . 17

3.1.2 Synthesis of N-rGO, N15-rGO and rGO . . . . 17

3.2 Characterization . . . . 18

3.2.1 XPS . . . . 18

3.2.2 Raman Spectroscopy . . . . 18

4 Results and Discussion 19 4.1 XPS . . . . 19

4.2 Raman Analysis . . . . 20

4.2.1 D-Band . . . . 20

4.2.2 I-Band . . . . 22

4.2.3 I*-Band . . . . 24

4.2.4 D”-Band . . . . 26

4.2.5 G-Band . . . . 28

4.2.6 D’-Band . . . . 29

5 Summary and Outlook 31

Chapter 1

Introduction

One of the most talked about and hyped groups of materials of the last decade are the sp ² nanocarbons which contains well-known names such as graphene, carbon nanotubes, fullerenes and graphite. The arguably most famous mem- ber of this group is graphene, which is a one-dimensional sheet comprised of carbon atoms arranged in hexagonal rings. Graphene, which was first isolated in 2004 by Novoselov et al. [21], is the youngest member of the sp ² nanocar- bon family. Although being the youngest member of the family graphene is sometimes referred to as the mother of all sp ² nanocarbons. The reason be- hind this moniker is that all other members of this family can be seen as being constructed using graphene as a basic building block. For example, carbon nan- otubes, being nanometer scale (diameter wise) tubes of hexagonal carbon rings, can simply be seen as rolled up sheets of graphene. During the last decade, the research community has seen a figurative explosion of graphene-related articles being published [14]. What has drawn researchers to this area are the extraordi- nary properties of this material. For example, graphene exhibits both excellent chemical, mechanical and electrical properties, making it attractive for many different potential applications including; electronics [13], hydrogen storage [16]

and fuel cells [17]. However, for graphene to be properly useful, its properties need to be carefully tuned to the particular application. One way of modify- ing both the chemical and electronic qualities of graphene is through doping, the intentional inclusion of impurities. Being next to carbon in the periodic table nitrogen is one of the most suitable dopant atoms for carbon structures due to their similarities in size and electronic characteristics [18]. The resulting properties of nitrogen-doped carbon depend on the exact nature and amount of nitrogen inclusions present. As of yet, the only precise method of analysing the structure of the nitrogen functionalities present in doped sp ² nanocarbon has been through the use of X-Ray Photoelectron Spectroscopy (XPS) which is a powerful but difficult and time-consuming method [19].

In this thesis, we try to expand this tool-kit also to include Raman spec-

troscopy. Raman spectroscopy is a fast, non-destructive and easy to use method,

and it is often used to characterise nanocarbons as they have a strong Raman

activity [15]. Because of this, the Raman spectra of graphene has been in- tensely researched and studies have shown it to hold a great deal of structural information [29, 30, 31, 32]. One method of singling out the effects of cer- tain functionalities on the Raman spectra is through isotope labelling. In this method instead of using, for example, the common earth abundant nitrogen-14, the heavier nitrogen-15 is used. As Raman spectroscopy probes the vibrational properties of a material, this mass change modifies the vibrations related to that element and they can thus, easily be identified and characterised. Isotope la- belling and Raman spectroscopy have been used before to study the vibrational modes of sp ² carbon [33, 34, 12]. The results of one of these studies were that the Raman modes related to the isotope labelled element were down-shifted in frequency. The frequency down-shift was found to follow the same behaviour as a classical harmonic oscillator [12]. This study observed these results in carbon- 13 labelled carbon nanotubes and similar effects should be observed while using the heavier nitrogen-15. This method has not been employed before to inves- tigate the nitrogen inclusion in sp ² carbon structures. We expect to be able to distinguish the nitrogen related Raman vibrations in a more straightforward manner by employing isotope-labelling during the synthesis process.

The study was conducted by systhesising undoped, nitrogen-doped and nitrogen- 15-doped reduced graphene oxide, a form of sp ² carbon similar to graphene with the main difference being a higher degree of defectiveness, and then comparing their respective Raman spectra. XPS was used to characterise the elemental and chemical compositions of the materials, to act as a support for the observations made in Raman spectroscopy. Raman spectra are often measured as single spot measurements. This type of measurement provides spectral information that is very spatially limited, making it unideal for some projects. Instead, the Raman spectra in this study were measured as overview maps. These maps contained around 1000 spectra spatially separated over large areas for each sample. This method provides a greater amount of data points allowing for stronger statis- tical analysis while simultaneously ensuring that genuine representations of the materials bulk properties are measured.

This thesis studied nitrogen inclusion in reduced graphene oxide by isotope

labelling and Raman spectroscopy. With the aim to expand the understanding

of vibrational modes related to nitrogen included in sp ² carbon.

Chapter 2

Theory

2.1 Graphene

The structure of the graphene lattice can be seen in Figure 2.1a. As we can see, the structure is comprised of a hexagonal network of carbon atoms with two unique atomic sites A and B. The atoms are separated by the C-C bond distance a _c−c = 0.142nm. In this figure, we can also see the unit cell (dotted region) and the unit vectors of the cell. The unit vectors a 1 and a 2 are described by Equation (2.1) [2, section 2.2.1]. This basic geometry also serves as the fun- damental building blocks of other sp ² nanocarbons, such as carbon nanotubes, leading to them having similar properties [2, section 1.1].

Figure 2.1: The real space unit cell of graphene showing unit vectors a 1 and a 2

and atomic sites A and B.[1]

a ₁ =

 √ 3 2 a, a

2



, a ₂ =

 √ 3 2 a, − a

2



(2.1)

2.1.1 Electronic Properties

In Figure 2.2 the electron dispersion relation, the relationship between momen- tum and energy of electrons, for the 1BZ (first Brillouin zone) of graphene is shown, the high symmetry points K,K’,Γ and M are marked. One important feature to note here is the degeneracy of the valence and conduction bands, the lower and upper surfaces shown in Figure 2.2, at the K and K ⁰ high symmetry points. With the Fermi level of graphene also located at these points, this makes graphene a zero band gap semiconductor with properties similar to metals. If we zoom in at the K point, another interesting feature is revealed about the elec- tronic properties of graphene. Near these points, the dispersion relationship is linearly dependent on the wave vector. This linear dependency is something usu- ally seen in massless relativistic particles, and it gives the electrons in graphene similar properties. From this, electrons and holes in graphene are effectively massless, resulting in a material with remarkable transport properties. These points are known as the Dirac points. [2, section 2.2]

Figure 2.2: The electron dispersion relationship of the 1BZ of graphene with high symmetry points Γ, K, K ⁰ and M marked. Inset shows dispersion relationship over directions KΓ, ΓM and M K. [2, p. 30]

2.1.2 Vibrational Properties

The phonon dispersion relation is shown in Figure 2.3. There we can see that graphene has six phonon branches corresponding to different lattice vibrations.

Lattice vibrations or molecular vibrations are the oscillations of a material due

to the molecular bonds present bending or stretching, essentially acting as a

complex system of harmonic oscillators. The abbreviations in the figure stand

for iLO (in-plane longitudinal optical), iTO (in-plane transversal optical), oTO

(out-of-plane transversal optical), iLA (in-plane longitudinal acoustic), iTA (in-

plane transversal acoustic) and oTA (out-of-plane transversal acoustic). These

abbreviations explain how the atoms in the vibrations moves relative to the lat-

tice. If we take a closer look at, for example, the iLO phonons the corresponding

Tobias Dahlberg February 19, 2016

lattice vibration is shown in Figure 2.4a. As we can see, iLO corresponds to atoms moving parallel to the graphene plane in the longitudinal direction (trans- lates to up and down in the figure) with each atom moving out of phase to its neighbours, which is the definition of an optical vibrational mode (with acoustic modes refers to neighbours being in phase). Some other examples of vibrational modes can also be seen in Figure 2.4 where vibrations near Γ Figure 2.4(a) and K Figure 2.4(b) are shown. [2, section 3]

Figure 2.3: The phonon dispersion relationship of graphene over directions KΓ,

ΓM and M K. [2, p. 53]

Figure 2.4: Vibrations in graphene and their related phonons near a) Γ b) K.

[4]

2.1.3 Nitrogen Doping

As stated before, nitrogen doping is a commonly used method for tuning the

properties of graphene. For example, the inclusion of nitrogen in graphene in-

troduces chemically active sites into this otherwise highly inert structure. These

active sites give catalytic properties to graphene as they can participate in chem-

ical reactions, such as the breaking of O-O bonds which is useful for fuel cell ap-

plications [10]. Nitrogen doping also alters the electronic properties of graphene,

due to nitrogen having one more valence electron than carbon. This introduc-

tion of donor atoms then shifts the Fermi level of the graphene, changing the

ordinarily conductive material into a semiconductor. The exact nature of the in-

duced changes depends on the total number and types of nitrogen functionalities

present in the sample. Usually four types of nitrogen inclusions are considered in

Nitrogen-doped rGO (N-rGO) and these are called N _pyridinic , N _pyrrolic , Quater-

nary center Nitrogen (N _Q

) and N _Q

. The N _pyridinic arises when nitrogen

shares one p-electron to the π-system next to a vacancy, it can be seen in Fig-

ure 2.5 marked in blue. N pyrrolic which originates from nitrogen atoms that

share two p-electrons with the π system of graphene, resulting in five member

rings or six member rings bonded to hydrogen next to a vacancy, this is shown

in Figure 2.5 marked in green. N Q

, also known as graphitic nitrogen, re-

sults from sp ² coordinated nitrogen that forms a direct in-plane substitution of

a carbon atom. Another variation of graphitic nitrogen is when nitrogen forms

a graphitic substitution near an edge or vacancy and it is known as N Q

.

These inclusions are shown in Figure 2.5 marked in black and red, respectively.

Tobias Dahlberg February 19, 2016

Figure 2.5: Illustration of different nitrogen inclusions in graphene. N pyrrolic

(green), N pyridinic (blue), N-Q center (black) and N-Q valley (red). [8]

2.1.4 Reduced Graphene Oxide

Reduced graphene oxide (rGO) are reduced and exfoliated flakes of graphite

oxide (GO); this process is shown in Figure 2.6. The GO precursor used is

often synthesised using a modified Hummer’s method where graphite is mixed

with sulfuric acid, potassium permanganate and sodium nitrate [27]. These

flakes are similar to graphene in structure, with the primary difference being the

degree of defectiveness. As rGO is subjected to heavy oxidisation, the resulting

structures are massively defective and corrugated even after reduction. Despite

being defective, this material retains some of the properties of graphene while

also being much easier to synthesise with a high yield. It is worth emphasising

that the possibility to upscale this method is one of its most promising features

as it makes it suitable for industrial use. Also, the high degree of defectiveness

of rGO also makes it easier to functionalise than pristine graphene, another

attractive property. These two properties factor in to make rGO a promising

material for future applications. [20]

Figure 2.6: Illustration of the transformation of graphite to reduced graphene oxide.[22]

2.2 Characterization Methods

2.2.1 XPS

XPS is a spectroscopic technique that is used to measure the elemental and chemical composition of a material. The basic principle of this technique is to measure the number and energy of photoelectrons emitted by a sample dur- ing x-ray irradiation. As each element and each chemical state have electrons with unique binding energies, the energy of emitted photoelectrons works as a fingerprint indicating the composition of the material. Counting the number of emitted electrons and correlating the count to their binding energy gives a measure of the relative amount of that chemical species present. XPS is limited to detecting surface properties as only the electrons emitted close to the surface can easily make it to the detector. Electrons emitted further into the sample have a significant chance of being recaptured or scattered by the material.

The relevant bands for the analysis of rGO and N-rGO are the N1s, O1s

and C1s bands shown in Figure 2.7. From these spectral regions the nitrogen,

oxygen and carbon content and their chemical states can be estimated.

Tobias Dahlberg February 19, 2016

Figure 2.7: Example XPS spectra of N-rGO showcasing the relevant bands O1s, N1s and C1s used to calculate oxygen, nitrogen and carbon content respectively.

The N1s band used for the analysis of nitrogen inclusion appear at a binding energy of 400 eV. This band is composed of a varying number of sub-peaks.

These sub-peaks correspond to the different types of nitrogen functionalities present in the sample as shown in Figure 2.8a. The first relevant band appear around 398 eV and it is assigned to N pyridinic . Next a band appear around 399 eV and it is assigned to N pyrrolic . Then a band at 401 eV appear which originates from N-Q _center . Lastly, a band at 402 eV is observed and it is assigned to N- Q _valley .[8]

Appearing around 285 eV the C1s band is related to the types of carbon

found in a structure. A high-resolution spectrum of this band is shown in

Figure 2.8b where the sub-bands relevant to this study is highlighted. The first

important sub-band appears at 284.5 eV and is related to carbon bonded to

carbon in an sp ² configuration. At 285.5 eV the next relevant sub-band appears

(a) (b)

Figure 2.8: A high resolution XPS spectra of a) N1s and b) C1s band used to characterise the amount and types of carbon and nitrogen present.

which is also related to carbon-carbon bonds but this time, to sp ³ coordinated carbon, which is the same type of carbon found in diamond. [9]

2.2.2 Raman Spectroscopy

Raman spectroscopy is a spectroscopic technique that is used to analyse the

molecular vibrational modes of a material. This method has its strengths in

that it is non-destructive, fast and requires little sample preparation. Relying

on the Raman effect, this method measures the inelastic scattering of photons

in a material to create a vibrational spectrum characteristic of that material. In

Figure 2.9, schematic energy level diagrams depicting the scattering processes

are shown. Firstly we have the elastic Rayleigh scattering that is not relevant to

Raman spectroscopy as it does not change the photon energy. In this process,

a photon excites the molecule to a virtual state with energy denoted as hv 0 , a

short-lived excited state not related to any eigenfunction of the system. From

this virtual state, the molecule then relaxes back to the initial state, emitting a

photon. Then we have the first of the Raman scattering processes, Stokes scat-

tering. In this process, the photon again excites the system to a virtual state

from the ground state E ₀ . The molecule then emits a phonon of energy hv _m

becoming vibrationally excited before radiatively relaxing to E ₀ − hv _m . This

phonon interaction causes a difference in energy of hv _m between the absorbed

and emitted photon. This energy difference is the measured quantity in Raman

spectroscopy, known as the Raman shift. Lastly, we have the anti-stokes scat-

Tobias Dahlberg February 19, 2016

Figure 2.9: Illustration of Rayleigh and Raman scattering processes for photons.

[23]

tering. In this process, the system, instead of generating a phonon, interacts with an already excited phonon increasing the energy of the resulting photon.

Raman processes are not limited to these examples and can contain multiple phonon and defect related scattering events. The number of scattering events taking place in Raman processes indicates its order. As those listed in Figure 2.9 contain only one, they are considered first order processes. If the excited virtual state coincides with a real electronic state, the process is called resonant and the likelihood of the transitions occurring is greatly increased. [2, section 4]

The shape of Raman bands is often described using Gaussian or Lorentzian functions. When deconvoluting Raman spectra containing many bands the Voigt functions is often used instead. This function is a convolution of the Gaussian and Lorentzian functions and, therefore, it can describe both shapes simultaneously. [28]

2.2.2.1 Raman Spectra of Graphene

Raman spectroscopy is a commonly used technique to characterise both elec-

tronic and structural properties of graphene. To give a more detailed description

of the Raman processes in graphene, they are usually not illustrated as can be

seen in Figure 2.9. Instead, they are commonly shown at their respective lo-

cation in the electron band structure, as can be seen in Figure 2.10. In this

section the Raman active bands present in graphene and their properties will

be discussed. In Figure 2.11 an example of the first order Raman spectrum of graphene is shown. There we can see the measured data (blue dots in the figure) and its deconvolution; an estimation of the bands making up the spectrum). In the following sections, a brief overview of the I, D, D”, G and D’ bands are presented. Commonly the spectral area 1000-1200 cm ⁻¹ is only assigned to one band, the I band. It should be noted that the regions 1000-1200 cm ⁻¹ and 1350- 1550 cm ⁻¹ are poorly researched and there is no real consensus on their exact nature. The I* band presented in the figure will be discussed in Section 4.2.3, as it is unique to this study.

Figure 2.10: Illustration of Raman processes of different bands where the cones show the shape of the electron dispersion relationship close to the Dirac points of graphene. a) G band, b) D band, c) D’ band. [2, p. 55]

I Band

The I band appears in the 1100-1200 cm ⁻¹ region, and little is known about this feature. It has been reported that this band originates in C-C, C=C or sp ² -sp ³ carbon bonds [3]. Alternatively, the band is seen as related to functionalised graphene (such as nitrogen-doped graphene) or heavily disordered carbon [10].

G Band

Named so after the fact that it occurs in all graphene allotropes and the existence of this band is a clear indicator of the presence of sp ² carbon. [2, p. 161] The G band is a first order Raman band where a virtually excited electron inelastically scatters with an iTO or iLO phonons near the Γ point; this process can be seen in Figure 2.10a [4]. The G band is usually located around 1585 cm ⁻¹ .

D Band

The D band is named after its close relation to defects in graphenes sp ² structure.

Tobias Dahlberg February 19, 2016

Figure 2.11: An example of the first order Raman spectrum of graphene (blue) showcasing a deconvolution of its sub-bands.

As we can see in Figure 2.10b, this is a second order process where a virtually excited electron near the K point scatters elastically from a defect to the K ⁰ point where it is inelastically scattered by an iTO phonon back to the initial state near the K point. The D band is usually observed in the spectral region around 1300 cm ⁻¹ and it intensifies and broadens with increased defectiveness.

Due to this relationship, the relative intensity of the D band is often used to estimate the crystallinity of sp ² nanocarbons. [2, p. 215]

D” Band

The D” band is a broad peak observed in the region 1400-1550 cm ⁻¹ and is another poorly understood band. This spectral feature is believed to be defect related [11].

D’ Band

The D’ band is another defect related Raman band [3], it usually appears around 1615 cm ⁻¹ , the underlying mechanics can be seen in Figure 2.10c.

2.2.2.2 Isotope Labelling

Isotope labelling of a structure is to exchange some elements with heavier iso-

topes. Doing this induces changes in the vibrational properties of the material,

leaving the electronic characteristics unchanged. Molecular vibrations can, in

a simplified manner, be seen as harmonic oscillators whose frequency ω is de-

scribed by Equation (2.2)

ω = r k

m , (2.2)

where m is the mass of the oscillator and k is the spring constant (corresponds to the bond strength in a molecule). This means that, if we assume the spring constant to remain unchanged during isotope labelling, the change in frequency after a mass change can be expressed as Equation (2.3)

ω 2 = ω

r m

m(1 − x) + m 2 x (2.3)

where m 2 , ω 2 the isotope labelled mass and frequency and x is ration of atoms

in the structure which have been isotope exchanged (x = 1 all atoms are iso-

topes, x = 0 no atoms are isotopes). As we can see from the equation, the

oscillator frequency down-shifts with increased mass. The relationship shown in

Equation (2.3) have been shown to accurately describe the down-shift of Raman

bands in carbon-13 isotope labelled carbon nanotubes [12].

Chapter 3

Experimental

3.1 Synthesis

3.1.1 Synthesis of GO

The GO was synthesised using a modified Hummer’s method. This method has more recently become known as Tours method and involves oxidation in a concentrated sulfuric and phosphoric acid mixture together with potassium permanganate. This method is safer and faster, yet yields similar products as the conventional Hummers method. The starting graphite used was Alfa aesar -100mesh natural flake graphite, briquetting grade.[25]

3.1.2 Synthesis of N-rGO, N15-rGO and rGO

rGO was chosen for this study because of it being easily functionalised and

synthesised. The three samples N-rGO, Nitrogen-15 doped rGO (N15-rGO)

and rGO were prepared using a method of simultaneous thermal reduction and

exfoliation similar to [26]. For the nitrogen-doped and isotope labelled samples,

100 mg of ammonium nitrate (Sigma-Aldrich > 99%) and 100 mg nitrogen-15

enriched ammonium nitrate (Sigma-Aldrich 98% N15) was mixed with 100 mg

of GO by stirring for 30 min in 10 ml of ethanol 99.5%. rGO was prepared using

the same method omitting the addition of ammonium nitrate. The mixture was

then dried while stirred at 60 ^◦ C on a hotplate. After drying, the sample was

calcinated in an oven at 350 ^◦ C for 1 h and then washed with ethanol and with

milli-Q water with repeated centrifugation. Finally, the samples were dried in

90 ^◦ C until thoroughly dry (< 2 h).

3.2 Characterization

3.2.1 XPS

The measurements were carried out using a Kratos Axis Ultra DLD electron spectrometer using a monochromatic Al K _α source operating at 120 W .

3.2.2 Raman Spectroscopy

Three samples of rGO, N-rGO and N15-rGO, all in powder form, were flattened

onto a glass slide to ease the analysis. The samples were analysed using a

Renishaw inVia confocal Raman microscope with an excitation wavelength of

633nm. An overview map (an image of where each pixel contains a Raman

spectrum) of more than 1000 spectra were recorded for each of N15-rGO, N-

rGO and rGO respectively. A high amount of spatially separated measurement

points was acquired to ensure that the data obtained were representative of the

materials bulk characteristics. The acquired spectra were noise filtered using a

multivariate noise filter and baseline corrected using an asymmetric least square

fit. The resulting processed spectra were batch de-convoluted in MATLAB ^TM

via a custom script. The script used 14 randomised starting guesses for each

spectrum to minimise any bias introduced via the choice of starting parameters

and all bands were fitted with Voigt functions. During the deconvolution, an

additional Raman band was added to the region 1100-1300 cm ⁻¹ as the data

indicated its existence. This hypothetical band will in the subsequent sections

be referred to as the I* band. Statistical analysis was conducted on the data

resulting from the deconvolution. The statistical analysis was used to estimate

the mean differences between the three samples. A one-way ANOVA (Analysis

Of Variance) test at a 5% significance level was used to achieve this.

Chapter 4

Results and Discussion

4.1 XPS

Table 4.1 and Table 4.2 shows a compilation of the relevant information ob- tained from the XPS measurements. There we can see the measured atomic content of the different nitrogen inclusion and also the total nitrogen and oxy- gen content of the three samples. The most important result shown here is that the two nitrogen doped samples, N-rGO and N15-rGO, were similar in total and specific nitrogen inclusion content. We can also observe that the N15-rGO sample contained a lower amount of sp ² carbon (peak at 284.5 eV) compared to N-rGO. While both samples had almost identical sp ³ carbon (peak at 285.5 eV) content. These results indicate that these samples are suitable for Raman comparison, as the main differences are those introduced via isotope labelling.

Table 4.1: The results of the XPS analysis of the carbon content of the three samples.

Sample C _sp

[at%] C _sp

[at%]

N-rGO 43.14 18.36

N15-rGO 38.85 18.45

rGO 48.54 15.3

Table 4.2: The results of the XPS study of the nitrogen content of the three samples. The atomic percentages of the different nitrogen inclusions and the total nitrogen content are shown.

Sample N _pyridinic [at%]

N _pyrrolic [at%]

N _Q

[at%]

N _Q

[at%]

N _total [at%]

N-rGO 3.66 4.38 0.84 0.51 9.31

N15-rGO 3.93 4.61 0.89 0.61 10.04

4.2 Raman Analysis

In this section, the results of the Raman analysis is presented. All peak heights mentioned in this section are normalized with respect to the G band

4.2.1 D-Band

The results of the measurements of the D band parameters of the three samples are shown in Figure 4.1 and the results of the statistical analysis are shown in Table 4.3. Firstly we can note that the mean center position of the D-band in N-rGO and N15-rGO is down-shifted relative to rGO by 8.2 cm ⁻¹ and 7.5 cm ⁻¹ , for N-rGO and N15-rGO respectively. The results also show that there is a statistically significant but weak up-shift of 0.7 cm ⁻¹ in the mean center position of N15-rGO, compared to N-rGO. This result indicates that this band does not directly originate in phonons related to nitrogen inclusions. Instead, this can be explained by the increased defectiveness introduced by the nitrogen inclusion in the structure. Knowing that the D band has been observed as closely linked with defective sp ² structures this result is expected.

From the analysis of the peak height, all of which have been normalized with respect to the G-band intensity, we can see that both N-rGO and N15- rGO shows an intensity increase of 85% and 56% (relative rGO), respectively.

Also, it is apparent that the D band in the N15-rGO is significantly lower than that of N-rGO, being decreased by 28.3%. From this, it can be reasoned that the decline in intensity can be linked to the lower sp ² content in N15-rGO, as revealed by the XPS analysis. Reports studying nanodiamond powders have also related a reduction of D band intensity to increased sp ³ content [7, p. 104].

This reason may also be an explanation of this behaviour as the N15-rGO sample contains a higher sp ³ to sp ² ratio than the N-rGO.

Widthwise this band seems to broaden with nitrogen doping as a width increase (relative rGO) of −10.1 cm ⁻¹ and −5.3 cm ⁻¹ is observed for N-rGO and N15-rGO. This behaviour is expected as the D bandwidth has been reported to increase with increased defectiveness due to changes in the lifetime of related phonons.

To summarise these results the D-band intensifies, broadens and down-shifts

with nitrogen doping while being almost unaffected by isotope labelling. These

findings indicate that the band is related to the degree of defectiveness present

in the sample and not necessarily nitrogen inclusion directly. As the D band is

well studied and has been confirmed to originate in defect activated vibrational

modes of sp ² carbon this behaviour is theoretically expected.

Tobias Dahlberg February 19, 2016

Figure 4.1: D band center position, height relative to G band and width with mean values marked by a star and the 95% confidence interval marked by solid lines. The colors correspond to rGO (red), N-rGO (blue) and N15-rGO (black).

Table 4.3: Results of ANOVA on D band parameters for rGO, N-rGO and N15- rGO. Results should be read as the difference between samples means i.e. rGO vs N-rGO is equivalent to mean(rGO)-mean(N-rGO).

Sample Mean center shift [cm ⁻¹ ] p value

rGO vs N-rGO 8.2 ± 0.1 1 ∗ 10 ⁻⁹

rGO vs N15-rGO 7.5 ± 0.1 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO −0.7 ± 0.1 1 ∗ 10 ⁻⁹

Sample Mean height shift [-] p value

rGO vs N-rGO −0.85 ± 0.01 1 ∗ 10 ⁻⁹

rGO vs N15-rGO −0.56 ± 0.01 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 0.283 ± 0.009 1 ∗ 10 ⁻⁹

Sample Mean width shift [cm ⁻¹ ] p value

rGO vs N-rGO −10.1 ± 0.2 1 ∗ 10 ⁻⁹

rGO vs N15-rGO −5.3 ± 0.2 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 4.8 ± 0.2 1 ∗ 10 ⁻⁹

4.2.2 I-Band

The most important result shown here can be seen in Figure 4.2 and Table 4.4. It is the significant (p-value << 0.05) down-shift of 11.7 cm ⁻¹ of the I band center in N15-rGO relative N-rGO. As this change is both statistically significant and larger than 10 cm ⁻¹ it is not unreasonable to correlate this down-shift to the isotopic labelling. Knowing that isotope labelling a chemical species with heavier isotopes should down-shift the related Raman bands, this correlation is further supported. As mentioned before, the down-shift upon isotope labelling can be described by Equation (2.3). By using this equation, and setting m ₁ = 14, m ₂ = 15 and x = 0.99 the expected down-shift can be estimated to 40 cm ⁻¹ ; four times the measured value. This discrepancy between theory and experimental results indicates that while this equation describes the behaviour of carbon-13 labelled sp ² carbon well, it does not accurately represent the more complicated case of nitrogen-15 labelled carbon. As a nitrogen doped structures are more defective and nitrogen inclusions form more complex and varied structures than carbon-13 inclusions, this might serve as an explanation for this deviation. It can also be noted that the mean I band center position up-shifts by 1.2 cm ⁻¹ in N-rGO compared to rGO. However, as this change is relatively weak, it cannot be correlated to nitrogen doping with any certainty.

If we move on to how the mean height of the peak differs between the samples we can see that it is increased by 18.5% in N-rGO and 9.2% in N15-rGO, compared to rGO. As this increase is strongly pronounced for both N-rGO and N15-rGO, it can be related to nitrogen doping. Interestingly, we can notice that the mean height of the peak in N15-rGO is lower than that in N-rGO. This result can again be correlated to the fact that the N15-rGO, as revealed by the XPS study, contained less sp ² carbon than the other samples.

Another interesting feature is that the width decreases by 5.7 cm ⁻¹ and by 15.0 cm ⁻¹ relative rGO, in N-rGO and N15-rGO respectively. This observation shows that the band narrows with isotope labelling that again could indicate a link to nitrogen functionalities.

To sum up these results they indicate that the I band is firmly related to

nitrogen doping in graphene as we can see that it both intensifies and narrows

after nitrogen doping. This connection is further supported by the significant

down-shift of the band after isotope labelling.

Tobias Dahlberg February 19, 2016

Figure 4.2: I band center position, height relative to G band and width with mean values marked by a star and the 95% confidence interval marked by solid lines. The colors correspond to rGO (red), N-rGO (blue) and N15-rGO (black).

Table 4.4: Results of ANOVA on I band parameters for rGO, N-rGO and N15- rGO. Results should be read as the difference between samples means i.e. rGO vs N-rGO is equivalent to mean(rGO)-mean(N-rGO).

Comparison Mean center shift [cm ⁻¹ ] p value

rGO vs N-rGO −1.2 ± 0.5 8 ∗ 10 ⁻⁹

rGO vs N15-rGO 10.5 ± 0.5 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 11.7 ± 0.5 1 ∗ 10 ⁻⁹

Comparison Mean height shift [-] p value

rGO vs N-rGO −0.18.5 ± 0.002 1 ∗ 10 ⁻⁹

rGO vs N15-rGO −0.092 ± 0.002 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 0.093 ± 0.002 1 ∗ 10 ⁻⁹

Comparison Mean width shift [cm ⁻¹ ] p value

rGO vs N-rGO 5.7 ± 0.5 1 ∗ 10 ⁻⁹

rGO vs N15-rGO 15.0 ± 0.5 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 9.2 ± 0.5 1 ∗ 10 ⁻⁹

4.2.3 I*-Band

The results of the measured hypothetical I* band parameters of rGO, N-rGO and N15-rGO is shown in Figure 4.3 and Table 4.5. As we can see from these results, the mean center position of the I* band is significantly down-shifted by 6.9 cm ⁻¹ and 7.1 cm ⁻¹ (relative to rGO) for N-rGO and N15-rGO respectively.

Whereas, only a 0.2 cm ⁻¹ mean difference is seen between N-rGO and N15- rGO. As this change is small and barely significant, having a p-value (strength of evidence, should be below 0.05 for statistical significance) of 0.0248, this band cannot be said to be affected by isotope labelling. From this, it can be speculated that as this band down-shifts as a result of nitrogen doping but is not affected by the isotope labelling, the band is probably defect related.

Figure 4.3: I* band center position, height relative to G band and width with mean values marked by a star and the 95% confidence interval marked by solid lines. The colors correspond to rGO (red), N-rGO (blue) and N15-rGO (black).

In the statistical analysis, we also see a significant increase in peak height

relative rGO of 25.2% for N-rGO and 25.5% for N15-rGO. As the p-value of the

test between the mean heights of N-rGO and N15-rGO is 0.1395, no significant

difference in mean height can be established between the two. Following the

same reasoning as in Section 4.2.1 the height of all band linked to sp ² carbon

should be decreased in the N15-rGO sample. However, as this is not the case

for this band, it can be speculated to be associated with sp ³ carbon. Drawing

Tobias Dahlberg February 19, 2016

this conclusion is not unreasonable as this spectral region has been reported as connected to sp ³ carbon in previous studies [6]. As we saw in the XPS analysis, the sp ³ carbon content in both N-rGO and N15-rGO are at similar levels. This result would mean that a Raman band related to this species would likely have a comparable intensity in both samples, which further supports this conclusion.

Table 4.5: Results of ANOVA on I* band parameters for rGO, N-rGO and N15- rGO. Results should be read as the difference between samples means i.e. rGO vs N-rGO is equivalent to mean(rGO)-mean(N-rGO).

Sample Mean center shift [cm ⁻¹ ] p value

rGO vs N-rGO 6.9 ± 0.2 1 ∗ 10 ⁻⁹

rGO vs N15-rGO 7.1 ± 0.2 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 0.2 ± 0.2 0.0248

Sample Mean height shift [-] p value

rGO vs N-rGO −0.252 ± 0.004 1 ∗ 10 ⁻⁹

rGO vs N15-rGO −0.255 ± 0.004 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO −0.003 ± 0.004 0.1395

Sample Mean width shift [cm ⁻¹ ] p value

rGO vs N-rGO 2.5 ± 0.4 1 ∗ 10 ⁻⁹

rGO vs N15-rGO −4.4 ± 0.4 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO −7.0 ± 0.3 1 ∗ 10 ⁻⁹

4.2.4 D”-Band

The results relevant to the D” band are presented in Figure 4.4 and Table 4.6.

We can see that this band displays a similar behaviour to that observed in the D band. The mean height is strongly increased by nitrogen doping being shifted relative rGO by 31.7% in N-rGO and 24.8% in N15-rGO. Again the band intensity is lowered in N15-rGO compared to N-rGO. Similarly to the D

Figure 4.4: D” band center position, height relative to G band and width with mean values marked by a star and the 95% confidence interval marked by solid lines. The colors correspond to rGO (red), N-rGO (blue) and N15-rGO (black).

band, the mean center position shows a dependence on nitrogen doping but not strongly on isotope labelling, being down-shifted relative rGO by 8.2 cm ⁻¹ in N- rGO and 10.3 cm ⁻¹ in N15-rGO. The difference between N-rGO and N15-rGO here is stronger than that observed for the D band but still relatively weak. As stated before, changes of this magnitude cannot be assigned significance owing to the inherently inhomogeneous nature of reduced graphene oxide.

Interestingly it can be noted that these results indicate that the width of this band is not different between the N-rGO and N15-rGO. This result is not consistent with the behaviour exhibited by the D-band and the reason behind this is unclear.

The results presented here seems to indicate that this band is defect related,

as it shows a behaviour similar to the D band. This observation is in line with

Tobias Dahlberg February 19, 2016

the results of previous studies.

Table 4.6: Results of ANOVA on D” band parameters for rGO, N-rGO and N15-rGO. Results should be read as the difference between samples means i.e.

rGO vs N-rGO is equivalent to mean(rGO)-mean(N-rGO).

Sample Mean center shift [cm ⁻¹ ] p value

rGO vs N-rGO 8.2 ± 0.5 1 ∗ 10 ⁻⁹

rGO vs N15-rGO 10.3 ± 0.5 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 2.1 ± 0.4 1 ∗ 10 ⁻⁹

Sample Mean height shift [-] p value

rGO vs N-rGO −0.317 ± 0.005 1 ∗ 10 ⁻⁹

rGO vs N15-rGO −0.248 ± 0.005 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 0.07 ± 0.004 1 ∗ 10 ⁻⁹

Sample Mean width shift [cm ⁻¹ ] p value

rGO vs N-rGO −8.0 ± 0.3 1 ∗ 10 ⁻⁹

rGO vs N15-rGO −7.7 ± 0.3 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 0.3 ± 0.3 0.1008

4.2.5 G-Band

As we can see from both Figure 4.5 and the statistical analysis in Table 4.7 the G band is weakly affected by both nitrogen doping and isotope labelling. All changes are on the order of 1 cm ⁻¹ making these results hard to use to draw any conclusions.

Figure 4.5: G band center position and width with mean values marked by a

star and the 95% confidence interval marked by solid lines. The height has been

omitted as it has been normalized to 1 for all samples. The colors correspond

to rGO (red), N-rGO (blue) and N15-rGO (black).

Tobias Dahlberg February 19, 2016

Table 4.7: Results of ANOVA on G band parameters for rGO, N-rGO and N15- rGO. Results should be read as the difference between samples means i.e. rGO vs N-rGO is equivalent to mean(rGO)-mean(N-rGO).

Sample Mean center shift [cm ⁻¹ ] p value

rGO vs N-rGO 2.7 ± 0.2 1 ∗ 10 ⁻⁹

rGO vs N15-rGO 3.7 ± 0.2 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 1.0 ± 0.1 1 ∗ 10 ⁻⁹

Sample Mean width shift [cm ⁻¹ ] p value

rGO vs N-rGO −6.2 ± 0.3 1 ∗ 10 ⁻⁹

rGO vs N15-rGO −7.5 ± 0.3 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO −1.3 ± 0.3 1 ∗ 10 ⁻⁹

4.2.6 D’-Band

In Figure 4.6 and Table 4.8 we see that this band exhibits similar behaviour to the D band, intensifying, broadening and down-shifting with nitrogen doping.

Also, we can see that the down-shift of this band in the N15-rGO relative N- rGO is quite small, being less than 5 cm ⁻¹ which makes it hard to draw any conclusions, as stated before. This result is in line with the expected behaviour of this band as it has been linked to defects. Again the trend of a lowered intensity of the band in N15-rGO can be observed.

Table 4.8: Results of ANOVA on D’ band parameters for rGO, N-rGO and N15-rGO. Results should be read as the difference between samples means i.e.

rGO vs N-rGO is equivalent to mean(rGO)-mean(N-rGO).

Sample Mean center shift [cm ⁻¹ ] p value

rGO vs N-rGO −0.9 ± 0.2 1 ∗ 10 ⁻⁹

rGO vs N15-rGO 3.6 ± 0.2 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 4.5 ± 0.2 1 ∗ 10 ⁻⁹

Sample Mean height shift [-] p value

rGO vs N-rGO −0.340 ± 0.009 1 ∗ 10 ⁻⁹

rGO vs N15-rGO −0.264 ± 0.009 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 0.075 ± 0.009 1 ∗ 10 ⁻⁹

Sample Mean width shift [cm ⁻¹ ] p value

rGO vs N-rGO −8.8 ± 0.2 1 ∗ 10 ⁻⁹

rGO vs N15-rGO −8.2 ± 0.2 1 ∗ 10 ⁻⁹

N-rGO vs N15-rGO 0.6 ± 0.2 1 ∗ 10 ⁻⁹

Figure 4.6: D’ band center position, height relative to G band and width with

mean values marked by a star and the 95% confidence interval marked by solid

lines. The colors correspond to rGO (red), N-rGO (blue) and N15-rGO (black).

Chapter 5

Summary and Outlook

This study set out to investigate if isotope labelling in conjunction with Raman spectroscopy could be used to characterise nitrogen functionalities in graphene.

Raman spectroscopy is a fast, easy and non-destructive method that probes materials molecular vibrational modes. These qualities make it an attractive method for characterisation. A strong understanding of the effects that nitrogen inclusions have on the Raman spectrum of graphene could potentially enable the characterisation of these functionalities through this method alone.

In this thesis, rGO doped with both nitrogen (N-rGO) and nitrogen-15 (N15- rGO) was synthesised, and their Raman spectra were analysed. rGO was cho- sen as it is easier to both functionalise and synthesise compared to ordinary graphene. From the study, potential evidence linking the I band to function- alised graphene was found, as it responded strongly to isotope labelling. Also, tentatively a new band denoted I*, possibly related to sp3 carbon, was identi- fied. Before this, the spectral region 1000-1300 cm ⁻¹ has only been thought to contain the I band and it has been speculated to be connected to functionalised graphene or sp2-sp3 carbon bonds. As this study indicates a separation of the I band into two bands, each dependent on one of these factors, it could bring clar- ity to this poorly understood area. If these results can be replicated, and backed up by more measurements, they could lead to a better future understanding of the Raman spectrum of graphene.

The results of this thesis are speculative and further studies would be needed to confirm them. Firstly, more data from more samples from each category (un- doped, N-doped and N15-doped) would be necessary to gather, as one sample from each class only gives an idea of the variations present in those specific sam- ples and not the variations between the different sample categories. Gathering more data would result in the possibility to use stronger statistical methods and also, most importantly, the ability to accurately judge which observations can be deemed significant.

Besides this, a study of the Raman spectra for N-rGO doped with different

levels of nitrogen could be used to identify clearly the nitrogen dependency of

the Raman bands. This process would likewise be interesting to conduct for

varying degrees of isotope content, which would serve as further support of the observation made in this thesis.

Further, the Raman spectrum of graphene is not only limited to the spectral features discussed in this thesis. Thus additional information could be found by studying, for example, the higher order Raman bands in the region 2300-3300 cm ⁻¹ .

To gain insight into the impact of specific nitrogen functionalities on the Ra- man spectrum, and possibly identify their vibrational modes. An experiment where samples containing varying amounts of particular types of nitrogen inclu- sion could be synthesised. Such an experiment could, for example, be carried out via heat treatment of N-rGO which would transform less stable nitrogen functionalities in more stable ones [8]. Also, another possibility would be to tweak synthesis parameters to promote different types of nitrogen inclusions, as suggested by Indrawirawan et al.[26].

To conclude, we can see that the results of this thesis are intriguing but

vague. However, they clearly indicate what to expect in similar experiments

and they provide a sound foundation for further studies.

Bibliography

[1] M. Huang, Studies of mechanically deformed single walled carbon nan- otubes and graphene by optical spectroscopy, Columbia University, (2009) [2] A. Jorio, M. Dresselhaus, R. Saito and G. Dresselhaus, Raman Spec-

troscopy in Graphene Related Systems, Wiley-VCH, Germany, (2011) . [3] A. Sadezky, H. Muckenhuber b, H. Grothe, R. Niessner and U. P¨ oschl,

Carbon 43, 1731, (2005)

[4] R. Beams, L.G. Cancado and L. Novotny, Journal of Physics: Condensed Matter J. Phys.: Condens. Matter 27, (2015)

[5] L.M. Malard, M.A. Pimenta, G. Dresselhaus, M.S. Dresselhaus, Physics Reports 473,51, (2009)

[6] Z. Zhu, F. Cheng and J. Chen, J. Mater. Chem. A, 1, 9484, (2013) [7] O.A. Williams, P. O’Brien, R. Nuzzo, Nanodiamond, Royal Society of

Chemistry, (2014).

[8] T. Sharifi, G. Hu, X. Jia and T. Wagberg, ACS Nano, 10, 8904,(2012).

[9] Z. Fan, W. Kai, J. Yan, T. Wei, L. Zhi, J. Feng, Y. Ren, L. Song and F.

Wei, ACS Nano, 5, 191, (2011).

[10] T. Sharifi, F. Nitze, H.R. Barzegar, C. Tai, M. Mazurkiewicz, A. Malolep- szy, L. Stobinski and T. Wagberg, Carbon, 50, 3535, (2012).

[11] S. Claramunt, A. Varea, D. Lopez-Diaz, M. Velazquez fixa texten, A. Cor- net and A. Cirera, Carbon, . Phys. Chem. C 119, 10123, (2015)

[12] S.D. Costa, C. Fantini, A. Righi, A. Bachmatiuk, M.H. Rummeli, R. Saito and M.A. Pimenta , 49, 4719, (2011)

[13] J. Lee, T. Ha, H. Li, K.N. Parrish, M. Holt, A. Dodabalapur, R.S. Ruoff and D. Akinwande, ACS Nano, 7, 7744, (2013)

[14] K.S. Novoselov, V.I. Falko, L. Colombo, P.R. Gellert, M.G. Schwab and K.

Kim, Nature ,490, 192, (2012)

[15] M.A. Pimenta, G. Dresselhaus, M.S. Dresselhaus, L.G. Cancado, A. Jorio and R. Saito, Phys.Chem.Chem.Phys., 9, 1276, (2007)

[16] S. Zhu, T. Li, ACS Nano, 8, 2864, (2014)

[17] S. Hu, M. Lozada-Hidalgo, F.C. Wang, A. Mishchenko, F. Schedin, R.R.

Nair, E.W. Hill, D.W. Boukhvalov, M.I. Katsnelson, R.A.W. Dryfe, I.V.

Grigorieva, H.A. Wu and A.K. Geim, Nature, 516, 227, (2014) [18] H. Wang, T. Maiyalagan, and X. Wang, ACS Catal, 2, 781, (2012) [19] K.A. Wepasnick, B.A. Smith, J.L. Bitter and D.H. Fairbrother, Anal

Bioanal Chem, 396, 1003, (2010)

[20] F. Perrozzi, S. Prezioso and L. Ottaviano, J. Phys.: Condens. Matter, 27, 21, (2015)

[21] K.S. Novoselov, A.K. Geim, S.V. Morozov, D. Jiang, Y. Zhang, S.V.

Dubonos, I.V.Grigorieva, A.A. Firsov, Science, 306, 666,(2004)

[22] rGO synthesis, illustration, viewed 4 January 2016, ”https:

//www.utu.fi/en/units/sci/units/chemistry/research/mcca/

PublishingImages/GO%20rGO.jpg”.

[23] Raman scattering processes, illustration, viewed 4 January 2016,”http:

//bwtek.com/wp-content/uploads/2012/06/R-F-04.jpg”

[24] S. undu, T.C. Nagaiah, W. Xia, Y. Wang, S.V. Dommele, J.H. Bitter, M.

Santa, G. Grundmeier, M. Bron and W. Schuhmann, J. Phys. Chem. C, 113, 14302, (2009)

[25] D.C. Marcano, D.V. Kosynkin, J.M. Berlin, A. Sinitskii, Z. Sun, A. Sle- sarev, L.B. Alemany, W. Lu, and J.M. Tour, ACS Nano, 4, 4806, (2010) [26] S. Indrawirawan, H Sun, X. Duana and S. Wang, J. Mater. Chem. A, 3,

3432, (2015)

[27] W.S. Hummers, R.E. Offeman, 80 (6), 1339, (1958) [28] M. Bradley, Thermo Fisher Scientific Inc., (2007)

[29] T.M.G Mohiuddin, A. Lombardo, R.R. Nair, A. Bonetti, G. Savini, R. Jalil, N. Bonini, D.M. Basko, C. Galiotis, N. Marzari, K.S. Novoselov, A.K. Geim and A.C. Ferrari, Phys. Rev. B, 79, 205433, (2009)

[30] A. Jorio, A.G. Souza Filho, G. Dresselhaus, M.S. Dresselhaus, A.K. Swan, M.S. nl, B. Goldberg, M.A. Pimenta, J.H. Hafner, C.M. Lieber and R.

Saito,Phys. Rev. B, 65, 155412, (2002)

[31] A. Das, S. Pisana, B. Chakraborty, S. Piscanec, S.K. Saha, U.V. Waghmare, K.S. Novoselov, H.R. Krishnamurthy, A.K. Geim, A.C. Ferrari and A.K.

Sood, Nat. Nanotech., 3, 210, (2008)

Tobias Dahlberg February 19, 2016

[32] M.M. Lucchese, F. Stavale, E.H. Ferreira, C. Vilane, M.V.O. Moutinho, R.B. Capaz, C.A. Achete, A. Jorio, Carbon, 48, 1592, (2010)

[33] X. Li, W. Cai, L. Colombo and R.S. Ruoff, Nano Letters, 9, 4268, (2009) [34] F. Simon, C. Kramberger, R. Pfeiffer, H. Kuzmany, V. Zlyomi, J. Krti,

P.M. Singer and H. Alloul, Physical Review Letters, 95, 017401, (2005)