Defect study in monolayer graphene: A comparative analysis of CVD and

exfoliated graphene using raman spectroscopy

Erika L. Buthersa,b,∗, Braulio S. Archanjob , Marcus V. O. Moutinhoa

a Federal University of Rio de Janeiro, Duque de Caxias, Rio de Janeiro, Brazil
b Materials Division, INMETRO, Duque de Caxias, Rio de Janeiro, Brazil

Abstract
This study presents a comparative analysis of defect densities in monolayer graphene synthesized via chemical vapor
deposition (CVD) and mechanical exfoliation, using Raman spectroscopy as the primary analytical tool. Defects were
induced in both types of graphene through helium-ion beam lithography to enable controlled and precise comparisons.
Raman spectroscopy, focusing on the D, G, and 2D bands, was used to quantify the defect density, with the intensity
ratio ID/IG serving as a key metric. Results indicated that CVD graphene exhibited a significantly higher initial
defect density compared to exfoliated graphene, attributed to its grain boundaries and impurities. Post-processing
thermal annealing was applied to CVD samples, resulting in a notable reduction in defect levels, as evidenced by the
decreased D-band intensity. After annealing, CVD graphene’s defect density approached that of exfoliated graphene,
underscoring the effectiveness of thermal treatment in improving its structural integrity. These findings highlight the
potential of CVD graphene for large-scale applications, provided that post-synthesis treatments, such as annealing,
are utilized to enhance its quality, making it more comparable to the defect tolerance and performance of exfoliated
graphene.
Keywords: Graphene, Helium Ion Beam, Nanolithography, 2D material

1. Introduction produced by peeling layers of highly oriented pyrolytic

graphite (HOPG), is known for its low defect concentra-
Graphene, a single layer of carbon atoms arranged tion and high electronic quality. However, the process
in a two-dimensional honeycomb lattice, has garnered is not scalable, which limits its use in large-scale appli-
significant attention because of its exceptional mechan- cations. In contrast, CVD graphene, which involves the
ical, thermal, and electronic properties. Since its dis- chemical deposition of carbon atoms onto a substrate,
covery, the potential of the material has been hailed in a is highly scalable and suitable for industrial production.
range of applications, from flexible electronics to high- However, this method introduces a higher density of de-
performance transistors and biosensors. However, the fects, such as grain boundaries and impurities, which
quality of graphene, particularly its defect density, plays degrade its performance.
a crucial role in determining its suitability for these ad- Raman spectroscopy has proven to be an invaluable
vanced technologies. Defects in graphene, such as va- tool for the nondestructive characterization of defects
cancies, grain boundaries, and impurities, can signifi- in graphene. By analyzing the intensity ratios of char-
cantly affect its conductivity, strength, and optical prop- acteristic peaks, such as the D band (defects), G band
erties, making the study of defect formation and miti- (graphitic sp² carbon), and 2D band (stacking and thick-
gation critical for the development of high-performance ness), researchers can gain insight into the defect struc-
graphene-based devices [1, 2]. ture and distribution within the graphene lattice. The
There are two primary methods for producing mono- ID /IG ratio, in particular, serves as a reliable indicator
layer graphene: mechanical exfoliation [3] and chemi- of defect concentration, allowing precise comparisons
cal vapor deposition (CVD). Exfoliated graphene, often between different graphene samples. [4, 5].
One promising approach for reducing defects, partic-
∗ Correspondingauthor ularly in CVD graphene, is thermal annealing. Heating
Email address: buthers@[Link] (Erika L. Buthers) graphene at high temperatures can lead to the healing
Preprint submitted to Carbon December 4, 2024
of vacancies, migration of impurities, and reorganiza- The other sample was a graphene sheet with a total di-
tion of grain boundaries, thus improving its structural mension of 1 × 1 cm on a 300 nm S i/S iO2 waffer (Fig.
integrity. Earlier works have shown that thermal an- ??), which was produced and provided by the Nano-
nealing reduces the D-band intensity in CVD graphene, materials Laboratory of the Physics Department of the
indicating a reduction in defect concentration [6]. Ther- Federal University of Minas Gerais (UFMG) using the
mal treatments not only repair intrinsic defects but also CVD technique.
help remove residual metal catalysts and other impuri-
ties from the CVD process.
In this study, we focus on comparing the defect den-
sities in monolayer graphene synthesized through CVD
and mechanical exfoliation. To facilitate a controlled
analysis, we introduced defects into both samples using
helium-ion beam lithography, allowing precise defect
formation. Additionally, we explore the use of thermal
annealing as a post-processing step for CVD graphene,
investigating its effectiveness in reducing defect density
and improving the material’s properties. By compar-
ing the defect structures before and after annealing, and
juxtaposing the results with exfoliated graphene, this re-
search aims to provide insight into improving the qual-
Figure 2: Optical image of a CVD graphene sheet.
ity of CVD graphene for scalable, high-performance ap-
plications.
2.2. Sample Preparation
Rectangular defect patterns were produced using the
2. Experiment section
Zeiss Orion NanoFab triple beam scanning microscope
2.1. Materials at the National Institute of Metrology, Quality and Tech-
nology (INMETRO). Helium and neon ion beams were
In this work, two samples were used. One of mono- used to create defects in a controlled manner in the sam-
layer graphene on a 300 nm S i/S iO2 waffer (Fig. ??), ples [7–9].
produced by the mechanical exfoliation of graphite For the exfoliated samples, square patterns spaced by
flakes supplied by Nacional de Grafite. This procedure 1 µm were made, varying the dose and exposure time
involved the prior heating of the substrate to a tempera- in the sample. The squares have dimensions of 2 × 2
ture of 100 °C, followed by the gentle application of the µm. and were produced by a helium beam with a cur-
graphite-bearing part of the Scotch tape onto the heated rent of 0.5 pA with random point defects with a beam
substrate. After maintaining a constant temperature for displacement of 10 nm in the x and y directions.
one minute, the system was cooled naturally to room In the CVD sample, other sets of square patterns with
temperature before carefully removing the tape. the same dimensions as the previous experiment were
made with the helium and neon ion beam in order to
evaluate the influence of other parameters during the
nanolithography procedure and to identify possible ef-
fects occurring in the graphene produced by chemical
deposition that differ from the graphene produced by
mechanical exfoliation. The doses, current, dwell time,
number of times the beam passes through a given point
(number of repetitions) and the spacing between defects
were varied.

2.3. Characterization
Optical characterization was performed using an
Olympus BX51M microscope, through which a map-
Figure 1: Optical image of a graphene sheet. ping of the substrate was performed to facilitate the
2
localization of the sample. Raman spectra were ob-
tained by a Witec microscope with a 532 nm wavelength
laser. Scanning microscopy images were obtained by
the Orion NanoFab with a size of 1024 by 1024 pixels
and a dwell time of 0.5 µs.

2.4. Simulation Parameters

Theoretical simulations of the electronic structure of

defective graphene monolayers were performed within
the tight-binding approximation, where the interaction
between neighboring carbons is parameterized by re-
sults obtained by first-principles calculations within
Density Functional Theory (DFT). A model that takes Figure 3: Raman spectra of the executed patterns on the CVD sample
into account the interaction up to fifth neighbors [10] with the He+ beam with a dosage of 1.04 ×1015 ions/cm2 , 100 nm
step in each direction and a current of 16 pA. Dwell time increasing
and periodic boundary conditions were used to write the upwards on the figure.
Hamiltonian of the system.
The code used was developed by the group to pro- Increasing the point dose may be favoring more heat-
vide information on some of the electronic properties of ing by increasing the exposure time in a smaller area
graphene with or without defects, such as capacitance, while using a higher current. Meanwhile, the possible
absorbance and density of states for supercells with one redeposition of material on the sample surface may be
or two layers of graphene. In this work, only absorbance occurring at a very low rate, not influencing the Raman
was evaluated, leaving the other properties for a later, spectrum result. Therefore, taking the results obtained
more in-depth study. with exfoliated graphene, we can see that more precise
Despite the large number of applications for this defects can be obtained by using a shorter exposure time
code, it presents some intrinsic problems of the theo- and a higher number of repetitions. However, other fac-
retical model used, such as the generation of random tors [11] may be jointly influencing this variation in both
defects in the structure in ways that are impossible to the intensity and width of the bands, and therefore, more
happen in practical experiments, as will be seen in the tests are necessary to find a relationship between the
results section. In addition, there was the computational contribution of each factor on the graphene bands.
limitation itself that restricts the use of the code to su-
percells of size greater than 12 × 12.

3. Results and discussion

3.1. Raman Spectroscopy

3.2. Dwell Time x Repetition

We evaluated the influence that the dose applied with

a helium ion beam at a single point under different con-
ditions can generate during the lithography process on
the CVD graphene bands. This was evaluated based on
the dose applied point-wise, maintaining the same total
dose applied to each bombarded pattern for each set of Figure 4: Repetition number.
tests while varying the number of repetitions and the ex-
posure time. The 2D band in Figure 3 did not show sig- Using different dosages we can compare figure 4 with
nificant changes, while the width of the D and G bands figure 5 and see the difference better. While maintain-
decreased with the decrease in the point dose. ing a fixed dwell time and changing only the number of
3
repetitions, the 2D band intensity noticeably increased
with the D band decreasing, which shows a reduction in
the number of defects caused. We can also see the 2D’
and D+D’ bands using lower doses which weren’t vis-
ible when fixing the number of repetitions and varying
the dwell time.

Figure 7:

Bombardments Before annealing.

Figure 5: Dwell time.

3.3. Annealing Treatment

Before annealing.
Figure 8:

Bombardments After annealing.

Figure 6:

After annealing. Figure 9:

4
3.4. Theoretical Simulations

In order to further investigate the effects on the phys-

ical properties of graphene as a function of the concen-
tration of defects in the crystal lattice, theoretical cal-
culations were performed using the tight-binding ap-
proximation for different systems. The simulated sys-
tems have different sizes and numbers of defects (vacan-
cies), representing different concentrations of defects
per cm2 , similar to those obtained experimentally by he-
lium beam. Through these calculations, it is possible to Figure 11: Supercells of the executed patterns referring to the absorp-
obtain the energy bands and the probability of electronic tion spectra shown in the figure 10. No defects.
transitions, resulting in the absorption spectrum. [12]

Figure 12: Supercells of the executed patterns referring to the absorp-

tion spectra shown in the figure 10. 2,86 ×1014 vacancies/cm2

Figure 10: Absorption spectra of graphene simulated with different

lattice dimensions and amount of resulting defects at different doses.

It is noticeable that the presence of defects in the

sheet alters the absorption spectrum, generating both a
shift in the position of the peaks of greatest intensity, as
well as a variation in the intensity and quantity of peaks
throughout the spectrum, as can be seen in Figure 10.
The supercells corresponding to the absorption spec- Figure 13: Supercells of the executed patterns referring to the absorp-
tra obtained can be seen in figures 11 to 17. It can be tion spectra shown in the figure 10. 3,58 ×1014 vacancies/cm2
seen that even with small variations in the dose, there
is a large difference between one spectrum and another. Conversely, it is also possible to estimate laser energy
Therefore, it is possible to easily adjust the defect pat- values that provide an increase in electronic absorption
tern to obtain absorption peaks in specific energy re- and, consequently, in the Raman signal, for certain de-
gions without damaging the material excessively. fect concentrations. It is important to emphasize that, in
5
this model, the defects are described as the simple re-
moval of a carbon atom from the crystal lattice, without
taking into account distortions due to atomic rearrange-
ment or other structural damage caused by ion bombard-
ment.

Thus, these simulation results should be close to the

He+ -beam-bombarded system, since these are atoms
with a small nucleus and low mass, compared to other
noble gas beams. Furthermore, in our damaged super-
cell simulation, the effects on the electronic structure
Figure 15: Supercells of the executed patterns referring to the absorp-
caused by 1 defect in a (n, m) cell are different from the tion spectra shown in the figure 10. 4,77 ×1014 vacancies/cm2
effects caused by 4 defects in a (2n, 2m) cell, for ex-
ample, since, in the latter case, the defects may not be
(and should not be) homogeneously distributed through-
out the structure, unlike in the first case. case, even if in
both cases the concentration of defects is the same.

This fact can be observed in the absorption spectra

contained in figure 10 obtained for cells with the same
concentration of 4.77 ×101 4 vacancies/cm2 , but with
completely different defect patterns, as shown in figures
14 and 15, respectively. These results also indicate that
not only the degree of local defect concentration alters
the absorption levels of the material, but the ordering of
these defects plays an important role in the electronic Figure 16: Supercells of the executed patterns referring to the absorp-
absorption of light. tion spectra shown in the figure 10. 9,54 ×1014 vacancies/cm2

Figure 17: Supercells of the executed patterns referring to the absorp-

tion spectra shown in the figure 10. 1,70 ×1015 vacancies/cm2

For example, it is possible to observe that the defects

generated in the structures represented in figures 4.13e,
Figure 14: Supercells of the executed patterns referring to the absorp- 4.13f, 4.14d and 4.14 give rise to linear patterns that re-
tion spectra shown in the figure 10. 4,77 ×1014 vacancies/cm2 sult in graphene nanoribbons with different thicknesses.
6
This causes the blocking of electronic transitions to low the neon beam there was a shift to smaller ones, reach-
energies that do not depend only on the defect concen- ing a maximum of around 2 × 1014 ions/cm2 . We no-
tration, but also on the thickness and separation between ticed that the exfoliated material has better crystallinity,
the resulting graphene nanoribbons, which is a direct possibly better heat dissipation and conductivity due to
consequence of our theoretical model where we take the absence of defects that CVD graphene intrinsically
into account the interaction between carbons up to and presented when observing the difference in band widths
fifth neighbors. The absorption spectrum for a concen- and intensities between the samples. Furthermore, due
tration of 4.77×1014 vacancies/cm2 contained in Figure to time, carbon redeposition of the material itself may
4.12b also shows this blocking of low-energy electronic occur during bombardment in CVD graphene, and this
transitions, but in this case the effect is due to the re- hypothesis is the subject of more in-depth future tests.
duced size of the cell, which contains one vacancy for Theoretical simulations of the electronic structure of
every 7 carbon atoms. single layers of defective graphene with different defect
These results pave the way for a more in-depth anal- concentrations were also performed within the tight-
ysis of how defects in geometric patterns, which can be binding approximation. These simulations allowed us to
generated and easily controlled by nanolithography, in- explore in detail the optical properties of the defect pat-
fluence the electronic structure and, consequently, the terns, highlighting the importance of their quantity and
optical absorption of graphene. Another interesting distribution in determining the properties of graphene.
point is in Figure 4.14c, where there is an isolated atom Even small variations in the dose of defects applied
in the middle of a triangular defect pattern. This struc- to the material present large variations in the absorbance
ture would not be possible to reproduce in a real envi- results, making this property easily modulated to exceed
ronment, showing one of the flaws in this code. doses of around 2 × 1015 ions/cm2 , which produce max-
imum ID /IG . Therefore, these simulations allow the ap-
4. Conclusions proximate selection for the most appropriate dose to ob-
tain the desired property before applying the defects to
Here, we investigated helium and neon ion beam the material.
nanolithography techniques in order to understand some
The thermal treatment both using a simple hot plate
of the essential parameters for their possible application
and a stove with argon atmosphere showed the best re-
in graphene-based nanodevices. Some of the optical
sults at 200 ºC. With a decrease on the intensity and
properties of monolayer graphene were studied through
width of the D band together with a rise in the intensity
theoretical simulations; and using Raman spectroscopy,
of the 2D band.
some of the influences of point defects produced by ion
beams on the bands of monolayer graphene were evalu- The results highlight the relevance of defect engineer-
ated. ing in graphene and ion beam nanolithography tech-
After the initial characterization of the samples via niques in the fabrication of nanostructured devices.
Raman spectroscopy to verify the graphene mono- From these, other works will be explored, such as the
layer, bombardments at different dosages with helium practical effect of simulated defect patterns and the pro-
ions were performed on the mechanically exfoliated duction of a nanobiodevice [13, 14] based on these de-
graphene sample and helium and neon on the CVD fect patterns. It is also expected to study in more depth
graphene sample, which produced different concentra- the effect of these defect patterns with TERS, collecting
tions of defects in the crystal lattice of this material. better resolution Near Field spectra [15]; exploring new
During the development of the project, it was possible geometric patterns.
to verify the viability of helium beam lithography as a
powerful tool for producing high-resolution and high-
precision defect patterns in graphene. Declaration of generative AI and AI-assisted tech-
By analyzing the response of graphene as a function nologies in the writing process
of the applied doses, we were able to observe the be-
havior of the bands intrinsic to the material as certain
parameters, such as ion dose, heating and bombardment During the preparation of this work the author(s) used
time, were varied in order to understand the behavior Chat-GPT in order to improve the readability and lan-
of graphene as a substrate for the development of nan- guage. After using this tool/service, the author(s) re-
odevices. The maximum ID /IG ratio was found around viewed and edited the content as needed and take(s) full
2 × 1015 ions/cm2 for the helium beam, whereas with responsibility for the content of the publication.
7
Acknowledgements [11] M. Huth, F. Porrati, C. Schwalb, M. Winhold, R. Sachser,
M. Dukic, J. Adams, G. Fantner, Focused electron beam induced
deposition: A perspective, Beilstein Journal of Nanotechnology
This work was supported by Nacional de Grafite, 3 (2012) 597–619. doi:10.3762/bjnano.3.70.
which sent us material to make our exfoliated samples, URL [Link]
Federal University of Minas Gerais, which sent us the [12] K. F. Mak, M. Y. Sfeir, Y. Wu, C. H. Lui, J. A. Mis-
CVD graphene samples, FAPERJ and CAPES for the ewich, T. F. Heinz, Measurement of the optical conduc-
tivity of graphene, Phys. Rev. Lett. 101 (2008) 196405.
financial support. doi:10.1103/PhysRevLett.101.196405.
URL [Link]
PhysRevLett.101.196405
References [13] S. Yu, T. Chen, Q. Zhang, M. Zhou, X. Zhu, Application of
dna nanodevices for biosensing, Analyst 145 (2020) 3481–3489.
[1] X. Li, W. Cai, J. An, S. Kim, J. Nah, D. Yang, R. Piner, doi:10.1039/D0AN00159G.
A. Velamakanni, I. Jung, E. Tutuc, S. K. Banerjee, URL [Link]
L. Colombo, R. S. Ruoff, Large-area synthesis of high- [14] J. Fried, J. Swett, X. Bian, J. Mol, Challenges in fabricat-
quality and uniform graphene films on copper foils, Science ing graphene nanodevices for electronic dna sequencing, MRS
324 (5932) (2009) 1312–1314. arXiv:[Link] Communications 8 (2018) 1–9. doi:10.1557/mrc.2018.
[Link]/doi/pdf/10.1126/science.1171245, 187.
doi:10.1126/science.1171245. [15] X. Jiang, W. Cai, W. Luo, Y. Xiang, N. Zhang, M. Ren,
URL [Link] X. Zhang, J. Xu, Near-field imaging of graphene triangles
science.1171245 patterned by helium ion lithography, Nanotechnology 29 (38)
[2] A. Reina, X. Jia, J. Ho, D. Nezich, H. Son, V. Bulovic, M. S. (2018) 385205. doi:10.1088/1361-6528/aad0b4.
Dresselhaus, J. Kong, Large area, few-layer graphene films on URL [Link]
arbitrary substrates by chemical vapor deposition, Nano Letters
9 (1) (2009) 30–35, pMID: 19046078. arXiv:[Link]
org/10.1021/nl801827v, doi:10.1021/nl801827v.
URL [Link]
[3] A. Geim, K. Novoselov, The rise of graphene, Nature materials
6 (2007) 183–91. doi:10.1038/nmat1849.
[4] M. Lucchese, F. Stavale, E. M. Ferreira, C. Vilani, M. Moutinho,
R. B. Capaz, C. Achete, A. Jorio, Quantifying ion-induced de-
fects and raman relaxation length in graphene, Carbon 48 (5)
(2010) 1592–1597. doi:[Link]
carbon.2009.12.057.
[5] A. Ferrari, D. Basko, Raman spectroscopy as a versatile tool for
studying the properties of graphene, Nature nanotechnology 8
(2013) 235–46. doi:10.1038/nnano.2013.46.
[6] K. Jia, Y. Su, Y. Chen, J. Luo, J. Yang, P. Lv, Z. Zhang, H. Zhu,
C. Zhao, T. Ye, Effects of defects and thermal treatment on
the properties of graphene, Vacuum 116 (2015) 90–95. doi:
[Link]
URL [Link]
article/pii/S0042207X15000949
[7] S. Boden, Z. Moktadir, D. Bagnall, H. Mizuta, H. Rutt, Focused
helium ion beam milling and deposition, Microelectronic Engi-
neering 88 (8) (2011) 2452–2455, proceedings of the 36th Inter-
national Conference on Micro- and Nano-Engineering (MNE).
doi:[Link]
[8] Y. Naitou, T. Iijima, S. Ogawa, Direct nano-patterning of
graphene with helium ion beams, Applied Physics Letters
106 (3) (2015) 033103. doi:10.1063/1.4906415.
[9] A. N. Abbas, G. Liu, B. Liu, L. Zhang, H. Liu, D. Ohlberg,
W. Wu, C. Zhou, Patterning, characterization, and chemical
sensing applications of graphene nanoribbon arrays down to 5
nm using helium ion beam lithography, ACS Nano 8 (2) (2014)
1538–1546, pMID: 24467172. arXiv:[Link]
10.1021/nn405759v, doi:10.1021/nn405759v.
URL [Link]
[10] P. Venezuela, M. Lazzeri, F. Mauri, Theory of double-resonant
raman spectra in graphene: Intensity and line shape of defect-
induced and two-phonon bands, Phys. Rev. B 84 (2011) 035433.
doi:10.1103/PhysRevB.84.035433.
URL [Link]
84.035433