Diamond & Related Materials 24 (2012) 179–183

Contents lists available at SciVerse ScienceDirect

Diamond & Related Materials

journal homepage: www.elsevier.com/locate/diamond

A novel high efficiency method for the synthesis of graphite encapsulated metal
(GEM) nanoparticles☆
Chih-Cheng Chiu, Jen-Chieh Lo, Mao-Hua Teng ⁎
Department of Geosciences, National Taiwan University, Taipei 106, Taiwan

a r t i c l e i n f o a b s t r a c t

Article history: Graphite encapsulated metal (GEM) nanoparticle is a composite material with a core-shell structure; it can
Received 20 June 2011 retain its novel properties even under severe environments. In recent years, many researchers have dedicated
Received in revised form 2 January 2012 their efforts to improving the synthesizing efficiency of GEM by the arc-discharge method; although some
Accepted 10 January 2012
progress has been made, the results remain unsatisfactory. To solve the problem, and under the guidance
Available online 14 January 2012
of the two-step mechanism model, it is clear that the carbon-to-metal ratio within the coalescence region
Keywords:
of the arc-discharge must be effectively controlled. Thus, an injection of liquid ethanol was used as the carbon
Nanoparticle source to replace the synthetic diamond powder; the results showed that the encapsulation efficiency of fer-
Spontaneous combustion romagnetic GEM was dramatically increased, i.e., 70–80% for Ni-GEM and Co-GEM, and for Fe-GEM it was
Ethanol quadrupled to about 50%. Furthermore, high encapsulation efficiency also prevents the common problem
Arc-discharge of spontaneous combustion of the as-made Fe-GEM powders. Some evidence as well as observations, indi-
cates that the ethanol jet provides a large quantity of carbon in the arc-discharge area in a relatively short
time. Although the detailed mechanism of this phenomenon is not well-understood, it shows that the pro-
duction rate could be increased by a factor of ten or higher than it was previously. In this study, a simple
and novel liquid ethanol injecting method has been developed; experiments have also demonstrated that
this is an effective method for the synthesis of ferromagnetic GEM nanoparticles.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction
Graphite encapsulated metal (GEM) nanoparticles have attracted
much attention since they were first discovered by Tomita et al. [1]
and Ruoff et al. [2] in 1993 using the Krätschmer-Huffman arc dis-
charge method [3]. The core of GEM is nanocrystalline metal, while
the outer shell is constituted of several layers of graphite/graphene.
With this special core-shell structure, GEM can survive even under
severe conditions/environments, such as oxidation or strong acidic
corrosion. Consequently, GEM has a wide variety of prospective appli-
cations in many fields, such as magnetic data storage, sorbents of
heavy metal ions [4], drug delivery agents and hydrogen storage [5].
Recently, some special behaviors of GEM materials have been studied
[6–9], such as magnetic packing, as well as their dispersive ability in
different dispersants. Although GEM has shown promising applica-
tion potential, the poor production rate and unsatisfactory encapsula-
tion efficiency hinder the study of this novel material.
In recent years, extensive efforts have been made to improve the
poor encapsulation efficiency of GEM nanoparticles. Among these stud-
ies, the modified tungsten arc-discharge method developed by Dravid
☆ Presented at NDNC 2011, the 5th International Conference on New Diamond and
Nano Carbons, Suzhou, China.
⁎ Corresponding author. Tel.: + 886 2 3366 6022.
E-mail address: mhteng@ntu.edu.tw (M.-H. Teng).
et al. [10] sheds light on the synthesis of GEM materials, whereby the
amount of carbon debris of the as-made product can be significantly
reduced. Nevertheless, the encapsulation efficiency remains unsatis-
factory. Note that since the arc-discharge method is an inert-gas-
condensation process [11], the coalescence should be the dominant
growth mechanism for GEM nanoparticles. Based on the two-step
mechanism model of Elliott et al. [12], it is the evaporated carbon-to-
metal ratio that determines the encapsulation efficiency of GEM mate-
rials. Thus, the key to successful encapsulation is to control the evapo-
rated carbon-to-metal ratio, in particular within the thin coalescence
region that closely surrounds the arc-discharge plasma area.
However, this is not an easy task for the old setup by which the
evaporated carbon vapor could not be sufficiently concentrated
until the solid carbon source material could first dissolve and mix
with the molten metal. On the other hand, injecting gaseous or liquid
carbon sources could be an effective alternative. Preliminary experi-
ments prove that the gas source is not a good choice due to its low
carbon density; rather, a liquid carbon source could be a more satis-
factory choice. In order to effectively deliver the liquid carbon source
(ethanol is the choice of this study) to the arc-discharge area, an obli-
que feedthrough inlet was designed to spray any fixed amount of eth-
anol during each arcing experiment. In this study, we will
demonstrate that a simple injection of liquid ethanol into the arc-
discharge area could dramatically increase the encapsulation efficien-
cy of the synthesis of GEM nanoparticles.

0925-9635/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.diamond.2012.01.015
180 C.-C. Chiu et al. / Diamond & Related Materials 24 (2012) 179–183

2. Experimental

To reduce the amount of carbonaceous debris and optimize the

encapsulation efficiency of the as-made products, the modified tung-
sten arc-discharge device was used to synthesize GEM nanoparticles.
In addition, a newly designed liquid ethanol inlet is attached to the
vacuum chamber, as shown in Fig. 1. The metal blocks (Fe, Co, Ni,
99.999%, Merck Ltd., Taiwan) are placed in a graphite crucible as the
anode, and the cathode is a tungsten rod. After the chamber was
purged twice with pure helium, the arcing experiment started to
run at 200 Torr helium, 30 V, and 100 A. Ethanol was chosen as the
liquid carbon source because it has a higher carbon density than
methanol does, and a lower evaporation temperature than
propanol's.
During the arc-discharge experiment, 6 ml of ethanol (99.8%,
Ferak Berlin) was fed directly into the arcing area, and the whole ex-
periment was completed in about five min. Then the as-made prod-
ucts were collected from the wall of the chamber, submerged in a
strong acid bath for 24 h to dissolve the ill-encapsulated metal nano-
Fig. 2. The XRD results of three ferromagnetic GEM materials. The core of Fe-GEM is
particles, and finally set in an intense magnetic field (4500 to 5000 G) mainly composed of bcc iron; while those of Ni-GEM and Co-GEM are composed of
to separate them from any carbonaceous debris. The purified prod- pure fcc Ni and Co, respectively.
ucts were then weighed, and characterized by X-ray diffraction
(PANalytical X'Pert Pro, CuKα), Raman spectroscopy (with 5 mW
laser power and 514.5 nm excitation wavelength Ar + laser), BET successfully prevented in this study. Because iron has greater carbon
(Quantachrome NOVA2000), SEM (FEI Quanta–200 F) and TEM solubility than nickel and cobalt do, a large portion of the dissolved
(JEOL JEM-2100 F-HR). The encapsulation efficiency is defined as the carbon in iron forms a stable solid solution instead of forming the en-
weight percentage of the residual powder that survived after the pu- capsulated graphite layers. (For more details on the formation mech-
rification procedure; i.e., the weight of the as-collected powders is anism, please refer to Elliott et al. [12] and Host et al. [15]) Therefore
100%. The encapsulation efficiency hereafter reported in the paper is the encapsulation efficiency of Fe-GEM is much lower than that of Ni-
determined by averaging the encapsulation efficiency derived from GEM and Co-GEM.
at least five experiments. The spontaneous combustion of Fe-GEM has long been considered
a problem, resulting from the oxidation of the ill-encapsulated Fe-
3. Results and discussion GEM nanoparticles. The exothermic reaction triggers a similar oxida-
tion reaction of other nearby ill-encapsulated particles, resulting in an
3.1. Sample characterization uncontrollable chain reaction of oxidation; thus, once the encapsula-
tion efficiency is enhanced, spontaneous combustion can be
The production rate of the as-made products by the new method prevented.
can be routinely maintained at as high an output as 0.8 g per min, Fig. 2 summarizes the XRD results of the three ferromagnetic GEM
which is almost 10 times higher than a previous study [13], and the crystals synthesized in this study, where the diffraction patterns of
encapsulation efficiency increases dramatically for Ni-GEM and Co- Ni-GEM and Co-GEM are consistent with the fcc structures for Ni
GEM from the original value of 20–30% to the satisfactory 70–80% and Co. The presence of the fcc phase cobalt at ambient condition
with high reproducibility; the standard deviation of the encapsulation was probably due to the small particle nature of particles [16],
efficiency is 2.50% and 4.83% for Co-GEM and Ni-GEM, respectively. As while the most prominent peak for Fe-GEM centered at around
has been anticipated, the encapsulation efficiency of Ni-GEM and Co- 44.65° is attributed to the (111) plane of bcc iron.
GEM is similar because of their analogous physical properties, i.e., Fig. 3 is the SEM image of Co-GEM; it clearly shows that the prod-
melting point, evaporation rate, etc., and their similar carbon-metal ucts synthesized in this study are almost spherical, as a result of the
phase diagrams. Although the encapsulation efficiency of Fe-GEM is coalescence process. Note that all of the samples synthesized in this
not as high as those of Ni-GEM and Co-GEM, it is quadrupled from
less than 10% to 45-50% on average, and the standard deviation of
the encapsulation efficiency is 7.01%; in addition, the spontaneous
combustion of Fe-GEM observed in a previous work [14] has been

Fig. 3. The SEM image of Co-GEM nanoparticles synthesized in this study; the shape of
Fig. 1. Schematics of the liquid ethanol inlet and the modified tungsten arc-discharge GEM is almost spherical, which indicates that the dominant growth mechanism of GEM
setup in a vacuum chamber. is coalescence.
 C.-C. Chiu et al. / Diamond & Related Materials 24 (2012) 179–183
Fig. 4. The Raman spectra of Ni-GEM nanoparticles synthesized in this study reveals
that the surface of the products is covered with graphitic structured carbon.
study reveal the same morphology; only the micrograph of Co-GEM is
presented in the paper.
Raman spectroscopy is a powerful tool for the characterization of
carbonaceous materials; the two salient peaks, G and D, are at around
1580 cm − 1 and 1360 cm − 1, respectively. The G peak originated from
the asymmetric stretching mode (E2g) of graphite in the basal plane;
the D peak resulted from the finite crystal size effect, while the sym-
metric stretching mode (A1g) became Raman active [17]; however, it
can also be considered as the defect mode of graphitic-structured car-
bon. The Raman spectra in Fig. 4 provide evidence that the surface of
GEM synthesized in this study is covered by graphitic structured car-
bon. GEM crystals are usually encapsulated by turbostratic graphitic
layers; therefore, the rather prominent D band peak might be a con-
sequence of a considerable amount of defects in the graphite shells
instead of any remaining carbonaceous debris. In fact, most of the car-
bonaceous debris has already been removed during the magnetic seg-
regation procedure. The linear increment of the background with
increasing wavenumber is attributed to luminescence.
An HRTEM image of Ni-GEM is shown in Fig. 5. The lattice fringes
that surround the metal nickel core clearly indicate that the outer
shell of the GEM is constituted of several graphitic layers, while the
electron diffraction spots indicate that the metal core is constituted
of fcc nickel. From the above characterization results, it is evident
that the morphology of the product GEM synthesized in this study
Fig. 5. TEM images of the Ni-GEM synthesized in this study. (a) HRTEM image of Ni-GEM. The core-shell structure is clearly identified, where the core is nickel and the outer shell is
graphite layers; the inset shows the electron diffraction pattern of the corresponding HRTEM image. (b) HRTEM image of the graphite shells.
181
is no different from that synthesized in previous studies. The particle
size distribution derived from TEM micrographs is shown in Fig. 6;
the average diameter is 30.19 nm. The log-normal distribution is
clearly identified, and it reveals that the dominant growth mecha-
nism of GEM crystals synthesized in this study is also coalescence
[11]. Note that only the histogram of Ni-GEM is shown in this study
since the physical properties affecting coalescence process, such as
melting point and evaporation rate, are analogous for iron group
metals [11]. The BET analysis shows that the specific surface area of
the Ni-GEM is around 30–35 m 2/g, and the average diameter is
around 25–30 nm, which is in good consistent with the results de-
rived from TEM. The theoretical density of the Ni-GEM 6.88 g/cm 3
could also be estimated by the thermogravimetric data, which indi-
cate that the average weight percentage of nickel and carbon of Ni-
GEM are about 91% and 9%, respectively.
3.2. Discussion
Two interesting phenomena were observed during the arcing ex-
periment in this study. First the color of the light emitted from the
arc-discharge instantly turned blue as soon as the liquid ethanol
was injected to the arc-discharge area, and second the shape of the
arc-discharge plasma changed from a bell-shape to a columnar-
shape, as illustrated in Fig. 7. (For more details on the morphology
of the conventional modified tungsten arc-discharge, please refer to
Teng et al. [7]). The former was owing to the light scattered by gas
molecules, and was mainly composed of short electromagnetic
waves; this provides direct evidence that when the injected ethanol
encounters the extreme heat of the arc-discharge, it is instantly evap-
orated and decomposed, and generates large quantities of molecules
or single atoms, such as hydrogen, oxygen and carbon near the arc-
discharge area.
The second phenomenon, i.e., the columnar-shape arc-discharge,
can be qualitatively elucidated as follows. When liquid ethanol is
injected directly into the arcing area, the ethanol decomposes into
carbon, hydrogen, oxygen and other hydrocarbon vapors, which sub-
sequently dominate the ionized species in the arc. The poorer thermal
conductivity and diffusivity of the complicated atmosphere, com-
pared to that of helium, are responsible for the presence of the ob-
served columnar shape arc. In this case, the morphology of a given
plasma plume is primarily determined by the composition and prop-
erties of gases, i.e., gas with poorer thermal conductivity and diffusiv-
ity yields a plasma plume with narrower spatial distribution. The
complicated atmosphere with poor thermal conductivity and diffu-
sivity traps more heat within the plasma plume, accelerates the
182 C.-C. Chiu et al. / Diamond & Related Materials 24 (2012) 179–183
Fig. 6. Particle size distribution histogram of Ni-GEM synthesized in this study.
evaporation process and, therefore, is responsible for the unusually
high production rate.
Fig. 8 shows the cross section of the metal bulk from the crucible
after the arcing experiment. The graphite plates and spherical holes
could be found within the region directly under the “columnar-
shape” arc-discharge, which contradicts a previous study [14] that
one should not find any visible graphite in this region. A possible con-
jecture is that this result is due to the occurrence of the “columnar-
shape” arc, which effectively focuses the thermal energy at the center
of the crucible. As a result, the highly localized energy would create
convection inside the molten metal, and “pump” the carbon source
from the circumference into this center region. Fig. 8 also reveals
that during the arcing experiment, a large amount of carbon could
dissolve and mix with the molten metal; therefore, the well-mixed
molten pool could act as a carbon source reservoir that continuously
provides carbon to the encapsulation processes.
Why does the synthesis efficiency suddenly become excellent but
only last for about five minutes? First of all, the energy needed to
Fig. 7. Schematics of the “columnar-shape” arc discharge. A column-like narrow plas-
ma plume appears at the region just above the molten pool, which is totally different
from that observed in all previous studies.
Fig. 8. An optical microscopic image shows the vertical section of solidified metal from
the crucible. The left hand side is the area directly under the columnar-shape arc-
discharge, where dense turbostratic graphite plates and some spherical holes can be
found. A lot of carbon from the decomposition of ethanol was dissolved in the molten
metal, and later crystallized into larger graphite plates in the cooler area (on the right
hand side of the picture).
evaporate and decompose the liquid ethanol is relatively small, com-
pared to that of the arc-discharge, so the injected liquid ethanol did
not disturb the energy flow of the system at the beginning. However,
once the liquid ethanol reached the molten metal, a series of process-
es began: (1) the ethanol decomposed into carbon, hydrogen, oxygen,
and other hydrocarbon gases; (2) the vast evaporated gases and
metal from the center of the crucible built up a conduction channel
for the arc-discharge; (3) a large amount of well-mixed carbon and
metal vapor condensed into GEM nanoparticles; and (4) soon the
complicated atmosphere and the graphite plates floating on top of
the molten metal “poisoned” the arc-discharge so that the electric re-
sistance became too large, and the arcing suddenly stopped. In order
to achieve mass production for any further scientific researches, this
“poisoning” problem and the addition of the unwanted oxygen prob-
lem must be solved.
In this study, we have verified that as long as the carbon-to-metal
ratio within the coalescence region is effectively controlled, the en-
capsulation efficiency can be substantially increased. Based on this
fact, it is expected that some metals that are difficult to encapsulate
by graphitic layers by using the modified tungsten arc-discharge
method, may still be encapsulated as long as the carbon-to-metal
ratio within the coalescence region is well-controlled.
4. Conclusions
In this study, we have demonstrated that a simple liquid ethanol
injecting method could greatly improve both the encapsulation effi-
ciency and the production rate of ferromagnetic GEM nanoparticles.
The production rates have increased almost ten-fold for all three fer-
romagnetic GEM nanoparticles, but the encapsulation efficiencies for
Ni-GEM and Co-GEM were found to be higher than that of Fe-GEM
due to iron's high carbon solubility. In addition, the spontaneous
combustion problem of these materials, in particular Fe-GEM, could
be prevented by using the ethanol injection method. XRD, Raman
spectra, SEM and TEM analyses confirm that the nanoparticles syn-
thesized with this method are identical to those of previous studies.
A peculiar “columnar-shape” arc-discharge was observed during the
experiment; although the detailed mechanism of this phenomenon
is still unclear, a preliminary estimation suggests that the energy den-
sity for the evaporation of source materials has been substantially in-
creased. In this study, it is shown that controlling the carbon-to-metal
ratio of an arc-discharge system is the key to the successful enhance-
ment of encapsulation efficiency of GEM nanoparticles.
Acknowledgment
This work was supported by the National Science Council, Taiwan,
Grant NSC99-2116-M-002-017 and NSC100-3113-E-002-008. Special
 C.-C. Chiu et al. / Diamond & Related Materials 24 (2012) 179–183 183

thanks to Prof. I-Nan Lin and Mr. Kuang-Yau Teng for their assistance [7] M.H. Teng, C.Y. Hsiao, Y.L. Hsiao, Diamond Relat. Mater. 18 (2009) 396–398.
[8] M.H. Teng, S.W. Tsai, W.A. Chiou, J. Alloys Compd. 495 (2010) 488–490.
with the TEM micrographs. [9] S.S. Li, M.H. Teng, Diamond Relat. Mater. 20 (2011) 183–186.
[10] V.P. Dravid, J.J. Host, M.H. Teng, B.R. Elliott, J.H. Hwang, D.L. Johnson, T.O. Mason,
References J.R. Weertman, Nature 374 (1995) 602.
[11] C.G. Granqvist, R.A. Buhrman, J. Appl. Phys. 47 (5) (1976) 2200–2219.
[1] M. Tomita, Y. Saito, T. Hayashi, Jpn. J. Applied Phys. 32 (1993) L280–L282. [12] B.R. Elliot, J.J. Host, V.P. Dravid, M.H. Teng, J.H. Hwang, J. Mater. Res. 12 (12)
[2] R.S. Ruoff, D.C. Lorents, B. Chan, R. Malhotra, S. Subramoney, Science 259 (1993) (1997) 3328–3344.
346. [13] M.H. Teng, S.W. Tsai, C.I. Hsiao, Y.D. Chen, J. Alloys Compd. 434–435 (31) (2007)
[3] W. Kräschmer, D.R. Huffman, Carbon 30 (8) (1992) I143–I147. 678–681.
[4] M. Bystrzejewski, K. Pyrzyńska, A. Huczko, H. Lange, Carbon 47 (2009) [14] J.C. Lo, J.C. Lu, M.H. Teng, Diamond Relat. Mater. 20 (3) (2011) 330–333.
1189–1206. [15] J.J. Host, V.P. Dravid, M.H. Teng, J. Mater. Res. 13 (9) (1998) 2547–2555.
[5] S.R. Chung, K.W. Wang, M.H. Teng, T.P. Perng, Int. J. Hydrogen Energy 34 (2009) [16] J.J. Host, M.H. Teng, B.R. Elliot, J.H. Hwang, T.O. Mason, D.L. Johnson, V.P. Dravid,
1383–1388. J. Mater. Res. 12 (5) (1997) 1268–1273.
[6] J.H. Hwang, V.P. Dravid, M.H. Teng, J.J. Host, B.R. Elliott, D.L. Johnson, T.O. Mason, [17] F. Tuinstra, J.L. Koenig, J. Chem. Phys. 53 (3) (1970) 1126–1130.
J. Mater. Res. 12 (4) (1997) 1076–1082.