Graphitization thermal treatment of carbon nanofibers

124
Accepted Manuscript Review Graphitization thermal treatment of carbon nanofibers Alberto Ramos, Ignacio Cameán, Ana B. García PII: S0008-6223(13)00250-9 DOI: http://dx.doi.org/10.1016/j.carbon.2013.03.031 Reference: CARBON 7912 To appear in: Carbon Received Date: 18 December 2012 Accepted Date: 6 March 2013 Please cite this article as: Ramos, A., Cameán, I., García, A.B., Graphitization thermal treatment of carbon nanofibers, Carbon (2013), doi: http://dx.doi.org/10.1016/j.carbon.2013.03.031 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Transcript of Graphitization thermal treatment of carbon nanofibers

Page 1: Graphitization thermal treatment of carbon nanofibers

Accepted Manuscript

Review

Graphitization thermal treatment of carbon nanofibers

Alberto Ramos, Ignacio Cameán, Ana B. García

PII: S0008-6223(13)00250-9

DOI: http://dx.doi.org/10.1016/j.carbon.2013.03.031

Reference: CARBON 7912

To appear in: Carbon

Received Date: 18 December 2012

Accepted Date: 6 March 2013

Please cite this article as: Ramos, A., Cameán, I., García, A.B., Graphitization thermal treatment of carbon

nanofibers, Carbon (2013), doi: http://dx.doi.org/10.1016/j.carbon.2013.03.031

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers

we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and

review of the resulting proof before it is published in its final form. Please note that during the production process

errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Page 2: Graphitization thermal treatment of carbon nanofibers

1

Graphitization thermal treatment of carbon nanofibers

Alberto Ramos, Ignacio Cameán, Ana B. García*1

Instituto Nacional del Carbón, CSIC, Francisco Pintado Fe 26, 33011-Oviedo, Spain

ABSTRACT

Carbon nanofibers (CNFs) and carbon nanotubes have revolutionized the world of the

nanotechnology due to their excellent mechanical, electrical and thermal properties. CNFs

are graphitic fibers made of stacks of graphene layers aligned perpendicular, tilted or

parallel to the fiber axis, thus resulting in different microstructures. Post-production

treatments can be applied to CNFs to improve their performance in several applications.

Among them, the heat treatment at high temperature to achieve the transformation of the

CNFs into graphite (graphitization) or graphitized CNFs (graphitization heat treatment) has

been studied in detail. This review covers the literature on this topic for the last 20 years,

analyzing the structural and textural changes showed by the CNFs during graphitization,

and how these changes influence their mechanical and electrical properties. Different

techniques, particularly, high-resolution transmission electron microscopy, have allowed to

determine the microstructure of these nanofilaments. A survey of the applications of

graphitized CNFs is provided, these including an additive for polymer reinforcement in

composites, an anode in lithium-ion batteries, a catalyst support in fuel cells, hydrogen

storage and others such as potential biosensors and catalysts in diverse reactions. In this

regards, special emphasis is placed on the advantages (or disadvantages) of using

graphitized CNFs instead of as-grown CNFs.

* Corresponding author. Tel/fax: +34 985 297662. E-mail address: [email protected]

Page 3: Graphitization thermal treatment of carbon nanofibers

2

CONTENTS

1. Introduction

2. High temperature treatment of carbon nanofibers

2.1. Evolution of the structural and textural properties

2.1.1. Graphitic structural order

2.1.2. Microstructure (structure and microtexture)

2.1.3. Porosity (surface area and pore volume)

2.1.4. Structural defects

2.2. Changes in mechanical and electrical properties

3. Applications of graphitized carbon nanofibers

3.1. Composites

3.2. Energy storage devices

3.2.1. Li-ion batteries

3.2.2. Fuel cells

3.3. Hydrogen storage

3.4. Other applications

4. Concluding remarks

Acknowledgements

References

Page 4: Graphitization thermal treatment of carbon nanofibers

3

1. Introduction

Carbon nanofibers (CNFs) are high aspect ratio graphitic fibers (≥ 100) with submicron

size diameters (typically below 100 nm) made up of stacks of graphene layers arranged in

different ways. These fibers possess outstanding thermal, mechanical and electrical

properties and have attracted a great deal of attention since Iijima published his article on

the closely-related carbon nanotubes (CNTs), thus revolutionizing the world of

nanotechnology in 1991 [1]. The CNTs alongside CNFs were often denoted in the

literature as carbon filaments, a term that described carbon fibers of submicron-size

diameters, before their microstructure (structure and microtexture) was elucidated.

Although it may seem a relatively new area of research, we can already find an account on

filamentous carbon materials in a patent in 1889 [2], more than a century before Iijima’s

discovery, in which the authors describe the production of carbon filaments through

decomposition of a C-containing gas in a metallic crucible at high temperature. But it was

not until the 1950s, with the development of the electron microscopy, that the nanometer

size of these carbons was confirmed [3,4]. Curiously, for a long period of time they were

considered more like a nuisance in the petrochemical and nuclear industries, where

deposits of these materials were often formed on metallic components in contact with hot

gases, such as hydrocarbons or COx [5,6]. In this regard, the early studies performed by

prestigious researchers in the 1970s, such as Baker [7], Oberlin and Endo [8], were more

focused on understanding the mechanisms of growth of such deposits to avoid their

formation. Nevertheless, it was with the advent of the nano-era that potential applications

for these carbon fibers were envisioned and efforts were made to tailor their synthesis in

order to enhance their properties, which resulted in an exponential increase of the number

of research articles on this topic published in the last 20 years.

Page 5: Graphitization thermal treatment of carbon nanofibers

4

Regarding the production of CNFs, the main method currently used is the catalytic

thermal chemical vapor deposition (CVD), which consists in the decomposition of a C-

containing gas (usually hydrocarbons or CO) over an elemental transition metal (Fe, Ni or

Co) or alloy as catalyst at temperatures in the range of 500-1200 ºC in a partial hydrogen

atmosphere [9]. The general mechanism for the catalytic growth of CNFs has been

proposed and confirmed several times and it has been explained in detail in previous

reviews [6,10,11]. It can be shortly described in three steps: (i) decomposition of the

hydrocarbon (or CO) on the metal surface, (ii) carbon dissolution and diffusion through the

bulk of the metal and (iii) precipitation on the form of graphite at the other side of the

metal particle.

We must also note that, although the catalytic thermal CVD is the most extended

and efficient method to produce CNFs, and it is even used in the large-scale production for

their commercialization [12], other promising alternative methods have been developed

more recently such as plasma-enhanced CVD and electrospinning. In the former, cold

plasma activation of the gas generated by electron impact is employed, thus leading to the

formation of vertically-aligned CNFs instead of entangled ropes of fibers, with potential

applications in the field of nanoelectronics [13,14]; whereas the latter consists in the

production of an electrostatically driven jet of a C-containing polymer solution, typically

polyacrylonitrile (PAN) in dimethylformamide. This solution is stabilized at temperatures

around 300 ºC to transform the thermoplastic nanofiber into a thermosetting fiber through

dehydrogenation, cyclization and polymerization processes. Subsequently, a carbonization

stage is carried out at temperatures usually in the range of 600-1200 ºC, which involves the

crosslinking, reorganization and coalescence of cyclized sections accompanying the

structural transformation from a ladder structure to a graphite-like one [15-17].

Focusing again on the thermal CVD, it is worth mentioning that the fine-tuning of

this process (temperature, metal and metal particle shape, carbon feedstock, etc.) allows the

Page 6: Graphitization thermal treatment of carbon nanofibers

5

production of all the different forms of carbon nanofilaments known to date, which differ

on their structure, texture, and properties, and can be classified as follows (Fig. 1): (a-b)

CNTs, which are comprised of graphene layer(s) rolled up in a cylindrical shape, forming

either single- or a multi-wall carbon nanotubes (SWCNTs or MCWNTs) depending on the

number of layers involved (one or more), in both cases bearing the metal particle at the tip;

(c-e) platelet, herringbone (or fishbone) and ribbon (or tubular) CNFs, consisting of

graphene layers perpendicular, tilted and parallel to the fiber axis, respectively, defining a

non-circular cross section, with the metal particles usually in the middle of the fiber in the

first two instances and in one extreme in the latter case, all of them known already since

the 1990s [18]; (f) stacked-cup and cone-stacked CNFs, formed by stacked truncated cones

leaving a big hollow core stacked-cup [19], or a small one cone-stacked [20]; (g)

cone-helix CNFs, constituted of a continuous helix-spiral graphite layer and an internal

hollow core, which some authors claim is the real structure of stacked-cup CNFs [21,22];

and finally (h) thickened CNFs and CNTs, comprised of an inner core with one of the

microstructures above described (a to g) and an outer pyrolytic layer of deposited carbon

formed after the catalytic growth of the filament. When this outer deposit leads to a fiber

with a diameter higher than 500 nm then it is considered as a vapor-grown carbon fiber

(VGCF) [23].

Page 7: Graphitization thermal treatment of carbon nanofibers

6

Fig. 1 - Depiction and/or TEM images of the different accepted structures for carbon

nanofilaments: (a) SWCNT and (b) MWCNT [9]; (c) platelet, (d) herringbone and (e)

ribbon or tubular CNFs [18]; (f) stacked-cup CNF [19]; (g) cone-helix CNF [9]; and

(h) thickened stacked-cup CNF [19].

Tailoring the structure, texture, and thermal, mechanical or electrical properties of

CNFs can be achieved not only by choosing the right production method and controlling

all the parameters involved as we have just commented, but also by post-production

Page 8: Graphitization thermal treatment of carbon nanofibers

7

treatments such as CVD of thin film coating, functionalization of the surface (etching in

air, soaking in acids, plasma etching, etc.), removal of the metal particles, and heat

treatment (HT), among others. Usually, one or a combination of several of them is often

necessary to improve the performance of the CNFs in all the different applications

accounted in the literature which can be classified in four different fields: (i) catalysts and

catalyst support materials [24-26]; (ii) polymer additives to form composites, improving

mechanical, thermal and electrical properties of the new materials [27-33]; (iii) electronic

devices, such as biosensors [28,34], anodes in lithium-ion batteries [35], or electrodes in

fuel cells [25,36]; and (iv) gas storage [11].

Amongst all the post-production treatments enumerated above, the HT at high

temperature of the CNFs to achieve their transformation into graphite (graphitization) or

graphitized CNFs (graphitization heat treatment) has been studied in detail. However,

although studies on the graphitization of filamentous carbon were briefly accounted on

Oberlin’s review from the 1980s [37], we are not aware of any other revision focused on

this topic since then. For that reason, the aim of this review is to cover the work on the

graphitization of CNFs (types b to f of the previous list, but also including some thickened

CNFs from type g) of the last 20 years, in which the development of the high-resolution

transmission electron microscopy (TEM) has allowed to determine their microstructure

(structure and microtexture) and to classify them as mentioned earlier. The structural and

textural changes showed by the CNFs during graphitization, and how these changes

influence their mechanical and electrical properties are analyzed by using different

experimental techniques such as X-ray diffraction (XRD), Raman spectroscopy, TEM and

other related microscopies, nitrogen adsorption/desorption isotherms, thermogravimetric

analysis, etc. Finally, a survey of the uses of graphitized CNFs is provided, just before the

concluding remarks.

Page 9: Graphitization thermal treatment of carbon nanofibers

8

2. High temperature treatment of carbon nanofibers

2.1. Evolution of the structural and textural properties

2.1.1. Graphitic structural order

XRD is widely used for the structural characterization of carbon materials. The average

values of the interlayer spacing, d002, and the crystallites sizes, height of layered stacking,

Lc, and basal plane length, La of these materials, which are calculated from XRD, are used

to estimate their graphitic structural order. The mean interlayer spacing is measured

through the position of the (002) peak applying Bragg’s equation while the mean

crystallites sizes, Lc and La, are mostly estimated from the (002) and (110) peaks,

respectively, using the Scherrer’s formula [38,39]. A more accurate procedure for the

measurements of the lattice parameters and the crystallite sizes of carbon materials by

XRD has been developed by Iwashita et al. [40]. The graphitization process will tend to

diminish the interlayer spacing down to the theoretical value for graphite (0.3354 nm) as

well as increase the mean crystallite sizes. Raman spectroscopy is also used to evaluate the

degree of structural order of different carbon materials [41-46], including CNFs [47,48]. It

is complementary to XRD although it has the advantage of surface specificity, thus

allowing the study of very heterogeneous materials. The most important parameter

calculated with this technique is the ratio of the integrated intensities of the D band (ID) at

1380 cm-1

, attributed to the defects of the graphitic structure, and the G band (IG) at

1580 cm-1

which is ascribed to a graphitic (ordered) structure, both bands belonging to the

first-order Raman spectrum for carbon materials. Obviously, the graphitization process will

reduce the intensity of the D band, therefore decreasing the ID/IG ratio. The degree of

Page 10: Graphitization thermal treatment of carbon nanofibers

9

structural order estimated by this technique possesses a bi-dimensional character, being

strongly dependent on the orientation of the crystallites, whereas in the case of XRD it has

a three-dimensional nature.

In this regard, Mochida and co-workers studied the graphitization process of

platelet [49-51] and tubular [50] CNFs produced by catalytic CVD, employing XRD and

Raman spectroscopy. The platelet CNFs, obtained by the decomposition of CO over a Fe

catalyst at 600 ºC [49], showed already a very high degree of structural order, with a d002 of

0.3363 nm, similar to that reported by Murayama and Maeda for filamentous graphite [52],

and Lc and La values of 28 and 22 nm, respectively. HT of these CNFs at 2800 ºC for 10

min involved little further graphitization with similar interlayer distance and slightly larger

crystallite size (30 x 30 nm). However, ball-milling and particularly, acidic treatment of the

graphitized CNFs were reported to increase the degree of structural order of these carbon

materials substantially, making them comparable to HOPG. Thus, CNFs with d002 of

0.3356 nm and Lc of 137 nm were obtained after treatment with nitric acid. A Raman study

was also carried out, which showed that the ID/IG ratio of the graphitized CNFs (0.24) was

much lower than that of the as-produced CNFs (1.35). The formation of loops between

adjacent end planes during the HT of the CNFs as confirmed by TEM (commented in

detail in the next section) reduces abruptly the number of free edges, thus accounting for

the fall of the ID/IG Raman ratio [44]. This effect also explained the increase of the ID/IG

ratio to 0.65 after the acidic treatment of the graphitized CNFs due to the rupture of some

of the loops (TEM observation, vide infra). In addition to this, the full width at half

maximum of the G band ( 1580 cm-1

) decreased slightly in the following order: as-

produced CNFs > graphitized CNFs > graphitized CNFs-milled ≈ graphitized CNFs-acid

treated, a fact that the authors claimed to be related to the degree of graphitization. Yoon et

al. [50] also found that altering the process variables, such as CO/H2 proportion (1:4 or 4:1

v/v) and catalyst composition (Fe or Fe/Ni 6/4 wt/wt), but especially the temperature (from

Page 11: Graphitization thermal treatment of carbon nanofibers

10

560 to 720 ºC) affected the morphology of the CNFs thus produced. In this regard, whilst

platelet CNFs were produced in the range of temperatures of 560-620 ºC, tubular (ribbon)

CNFs with hollow cores were obtained at higher temperatures (630-675 ºC). Both types of

CNFs were also subjected to HT at 2000 ºC and 2800 ºC, and the structural changes thus

induced were followed by XRD. In the case of the platelet-type CNFs, the d002 remained

unchanged, while the crystallite size Lc grew slightly upon increasing temperature. The

resultant graphitized platelet CNFs exhibited crystalline parameters similar to those above

commented [49]. The tubular CNFs also displayed a highly-ordered graphitic structure, as

denoted by the low d002 value of 0.3369 nm, which increased after HT: 0.3387 nm at 2000

ºC and 0.3375 nm at 2800 ºC. These values, together with scanning electron microscopy

(SEM) observations to be commented later on, discarded the possibility of a tubular

alignment of the planes, as in MWCNTs, because this type of alignment will give a

theoretical minimum d002 value of 0.339 nm [53]. In stark contrast, the Lc crystallite size

grew significantly from 9.5 nm in the as-produced CNFs to 16.2 nm in the heat-treated

ones at 2800 ºC, which is consistent with the removal of structural defects during the

graphitization process.

In 2006, Oya and Ono synthesized tubular CNFs applying the polymer blend

technique (electrospinning) to a naphthalene-based mesophase pitch dispersed in a

polymethylpentene matrix [54]. CNFs with a relative low degree of structural order (d002 of

0.3379 nm and Lc of 1.6 nm) were obtained. After HT at 3000 ºC, the CNFs showed an

interlayer spacing of 0.3367 nm and a crystallite size of 56.9 nm, thus suggesting a three-

dimensional crystal structure similar or close to that of graphite which was confirmed by

the presence of the (112) reflection in the selected area electron diffraction (SAED) pattern

obtained from the TEM images.

More recently, Fujikawa et al. [55] have studied the graphitization of rectangular

platelet CNFs in the presence (B-doped) and absence of elemental boron. XRD and Raman

Page 12: Graphitization thermal treatment of carbon nanofibers

11

spectroscopy measurements were performed to follow the structural evolution of the CNFs

during HT at 1900, 2200 and 2500 ºC. As in previous examples, the as-grown platelet

CNFs already showed a significant degree of structural order with d002 and Lc values of

0.3369 nm and 18.02 nm, respectively. The HT of the CNFs slightly improved their

graphitic order; thus, values of 0.3357 nm and 25.31 nm were determined for the latter

parameters in those treated at 2500 ºC. For the B-doped CNFs, a higher improvement of

the structural order was observed after HT, reflecting the fact that boron atoms are

graphitization accelerators for carbon materials.

Fig. 2 - Raman spectra of CNFs: (a) without B and (b) with B. From bottom to top,

results are shown for the as-grown and heat-treated at 1900, 2200 and 2500 ºC CNFs

[55].

The decrease of the half-width at half-maximum of the D and G Raman bands as well as

that of the ID/IG ratio of the CNFs after increasing the temperature of HT was also

indicative of the improvement of the graphitic order (Fig. 2). For example, ID/IG ratios of

1.390 and 0.241 were, respectively, calculated for as-prepared and heat-treated CNFs.

However, the intensity of the D band did not seem to decrease with temperature whereas

Page 13: Graphitization thermal treatment of carbon nanofibers

12

the D’ band ( 1620 cm-1

) disappeared, a fact attributed to the formation of loops observed

in the TEM images. In the case of the B-doped CNFs, the D and D’ bands were still intense

after HT and the ID/IG ratio did not diminish in the same proportion as the as-grown CNFs

did (0.705 at 2500 ºC), which is consistent with the incorporation of boron atoms to the sp2

carbon network as confirmed by X-ray photoelectron spectroscopy (XPS).

Habazaki et al. [56] reported the synthesis of platelet CNFs by liquid phase

carbonization of poly(vinyl)chloride (PVC) powders. The CNFs were treated at 1000-2800

ºC and the structural changes were followed by XRD (Fig. 3). The as-produced CNFs

exhibited a low degree of cristallinity (d002 of 0.348 nm which corresponds to a turbostratic

structure and Lc of 4 nm). By increasing HT temperature, the (002) peak of XRD profile

became more intense and narrower, finally splitting at 2800 ºC in two overlapped peaks

ascribed to the presence of carbons with different degree of graphitization. Thus, besides

graphitic structures with d002 and Lc of 0.336 nm and 43 nm, others with turbostratic

character (d002 of 0.340 nm, Lc of 27 nm) were also detected in the heat-treated CNFs.

Fig. 3 - XRD patterns of as-produced (600 ºC) and heat-treated (1000-2800 ºC)

platelet CNFs [56].

Page 14: Graphitization thermal treatment of carbon nanofibers

13

The graphitization of herringbone (or fishbone) CNFs, produced by methane

decomposition over Ni catalysts supported on SiO2, Al2O3, TiO2 or MgO, was studied in

detail by Garcia and co-workers [57-60]. In the first work [57], the CNFs were heat-treated

at 2400, 2600 and 2800 ºC and characterized by XRD and Raman spectroscopy. The as-

produced CNFs already showed a significant degree of structural order with d002 values of

0.339 nm and crystallite sizes (Lc and La) of up to 8 and 24 nm, respectively. As

expected, the HT decreased the interlayer distance ( 0.337 nm at 2800 ºC) and increased

the crystallite sizes (up to 17 and 39 nm) of the CNFs. The variation with temperature of

the Raman parameters followed a parallel way. Thus, a progressive narrowing of the G

band, indicating an increase in size and orientation of the graphitic domains, and a decrease

of the relative intensity of the D band, associated with an improvement of structural order

and crystalline orientation, were observed. The most structurally ordered materials were

obtained from those CNFs containing Si or Ti in addition to Ni, this implying the catalytic

role of these metal species. According to the model proposed for the catalytic

graphitization of hard carbons [61], the active constituents of CNFs (metals) would react

with the disordered carbons at the boundaries of the BSUs to form the respective carbides

which, upon further decomposition, would lead to graphite, thus increasing the size of the

already existing graphite layers. As a proof of principle, the presence of silicon and

titanium carbides was detected by XRD in the graphitized CNFs. The intensity of these

peaks was found to decrease by increasing the temperature of HT (decomposition

temperature > 2700 ºC) (Fig. 4).

Page 15: Graphitization thermal treatment of carbon nanofibers

14

Fig. 4 - XRD patterns of as-produced (NF01-NiCuTi) and graphitized (NF01-

NiCuTi/2400–2800) CNFs [57].

In subsequent communications, the catalytic effect of the inherent metals species

(namely Ni and Si) in the graphitization process of other fishbone methane-based CNFs

containing different proportions of Ni and Si was studied by the same authors [58, 60]. As

indicated by XRD and Raman spectroscopy, the degree of structural order of these

materials improved progressively with increasing HT temperature (Fig. 5). In fact, good

linear correlations between treatment temperature of the CNFs and both the interlayer

spacing, d002, and the crystallite size, Lc, of the materials prepared were found.

Page 16: Graphitization thermal treatment of carbon nanofibers

15

Fig. 5 - XRD parameters (a) and first-order Raman spectra (b) of the as-produced

(CNF-5) and heat-treated (CNF-5/1800-2800) CNFs. [60].

Moreover, for a given HT temperature more-ordered materials were obtained from

CNFs with higher Si/Ni weight ratio. As a result, graphite-like materials with structural

characteristics (interlayer spacing of 0.3364 nm and crystallite sizes Lc and La of 36 nm

and > 50 nm) comparable to oil-derived synthetic graphite were prepared. The formation of

silicon carbide (SiC) was apparent at temperatures ≥ 2400 ºC by XRD analysis; although

the intensity of the peaks attributed to this species decreased with increasing temperature.

In addition to SiC and graphite, nickel silicide peaks were also observed in the

Page 17: Graphitization thermal treatment of carbon nanofibers

16

diffractograms of the heat-treated CNFs. In an attempt to know the role played by the

inherent Ni in the graphitization, metals were firstly removed from the CNFs by an acidic

treatment with HNO3/HF followed by the addition of silica and the resulting CNFs ( 24

wt % of Si, no Ni content) were treated at 2400 and 2800 ºC. As compared with the

absence of metals, the presence of silica in the CNFs was found to significantly improve

the structural order of the materials prepared. However, the three-dimensional order

developed for these materials was still far from that achieved from the CNFs containing

lower concentration of Si in addition to Ni. Based on these results, the synergetic catalytic

effect of the Ni and Si species on the graphitization of these CNFs through the formation of

a nickel silicide phase as an intermediate state which further promotes the production of

silicon carbide was inferred. The formation of graphite at the expense of the subsequent

carbide decomposition was claimed to be a plausible mechanism of the catalytic

graphitization of CNFs.

The graphitization of stacked-cup CNFs was examined by Endo et al. by XRD [62]

and Raman spectroscopy [62,63]. These CNFs were synthesized by a floating reactant

method using ferrocene or iron pentacarbonyl as catalyst precursor, hydrogen sulfide as co-

catalyst, and natural gas as carbon feedstock [19]. They showed relatively low

graphitizability even at 3000 ºC as indicated by the absence of separation of (101) and

(100) peaks and the low intensity of the (004) reflection in the diffractograms. The authors

also noticed that a minimum d002 value of 0.3378 nm was assessed for the heat-treated

CNFs. Moreover, the interlayer spacing of the CNFs treated at 1800 ºC (0.3427 nm) was

slightly higher than that of the as-grown CNFs (0.3424 nm). They ascribed this increase to

a structural disruption at the edges of the graphene planes due to formation of loops and

also transformation into multi-loops as observed by TEM (discussed below). This effect

was reported to be a determining factor for the poor graphitization behavior of these CNFs

which they claimed was similar to that of glassy carbon. Raman spectra were consistent

Page 18: Graphitization thermal treatment of carbon nanofibers

17

with data obtained from XRD. Thus, little changes occurred above 1500 ºC because of the

formation of multi-loops, which impede the graphitization. As a result, relatively intense D

and D’ peaks were observed in the first-order Raman spectra of the heat-treated CNFs,

even at 3000 ºC. Moreover, the symmetric shape of the 2D peak at 2700 cm-1

in the

second-order spectrum was attributed to an incomplete three-dimensional graphitization.

Howe et al. [64] and Yoon et al. [65] also investigated the effect of annealing on the

degree of structural order of two types of stacked-cup CNFs thickened with an outer CVD

layer of turbostratic carbon which were produced by Applied Sciences Inc. (ASI), namely

Pyrograf® (PR-19 and PR-24) (see http://www.apsci.com), in a flow of natural gas, air,

ammonia, and the catalytic constituents hydrogen sulfide and iron pentacarbonyl. PR-19

CNFs were heat-treated in the range of 1100-2900 ºC and the structural evolution followed

by XRD [64]. As regards the interlayer spacing, no significant changes were appreciated

when treating the CNFs at 1300 and 1500 ºC, whereas the Lc decreased slightly. However,

after HT at 2900 ºC, a significant improvement of the graphitic structural order of the

CNFs was observed; the interlayer spacing dropped to 0.3350 nm while the crystallite size

Lc grew up to 21 nm. In a follow-up publication [65], PR-19 and PR-24 CNFs were heat-

treated at different temperatures up to 3227 ºC. Graphitized CNFs with a d002 value of

0.3350 nm were again obtained at 2500 ºC. However, further heating at higher

temperatures did not improve the degree of graphitization. In order to account for this fact,

they proposed a theoretical model, also based on TEM and SEM observations, implying

that HT causes a polygonization of these CNFs that will be explained in detail in the next

section.

Paredes et al. [66] also carried out an extensive graphitization study of similar

thickened CNFs obtained from the same commercial source (ASI). These fibers were

subjected to HT at 1800, 2300 and 2800 ºC. The global structural characterization was

performed by XRD and Raman spectroscopy. In the diffractograms, the most intense peak

Page 19: Graphitization thermal treatment of carbon nanofibers

18

was the (002) which in the as-grown CNFs exhibits a broad shoulder on its low-angle side

which was deconvoluted in two: one at 26.3º attributed to an ordered graphite phase, and

a broad band at 24.7º indicative of a less-ordered carbon phase (consistent with the

known dual structure of these CNFs observed by TEM). Moreover, the (100) and (101)

reflections as well as the (004) were also discernible. On increasing the HT temperature,

the (002) peak was observed to narrow, to become more symmetric as the shoulder

disappeared which was ascribed to the gradual conversion of the less ordered phase into

graphitic carbon, and to increase in intensity (Fig. 6). However, some degree of asymmetry

was even detected in the CNFs treated at 2800 ºC. Therefore, to quantify the extent of the

graphitization of these CNFs, the authors calculated the XRD crystalline parameters of the

graphitic and disordered phases as well as their area ratio. The interlayer spacing d002 and

crystallite size Lc of the graphitic phase were respectively found to decrease and increase

with HT temperature to finally reach values of 0.336 nm and 20 nm at 2800 ºC, thus

suggesting a highly graphitic material with some remaining structural defects. These values

were similar to those reported above for this type of CNFs [64,65].

Fig. 6 - XRD profiles of the as-grown and heat-treated CNFs. Top right inset:

deconvolution of the (002) peak for the as-grown CNFs [66].

Page 20: Graphitization thermal treatment of carbon nanofibers

19

As regards the Raman data, the widths of D, G and D’ bands decreased with

increasing temperature, indicating a progressive removal of the structural disorder. The

ratio between the integrated areas of the D and G bands, ID/IG, related to the graphitic order

of the material was also observed to fall dramatically. However, in consistency with the

XRD results, the D band in the Raman spectra did not completely disappear even at the

highest treatment temperature, this meaning that some degree of structural disorder was

still present in the graphitized CNFs. The 2D band in the second-order Raman spectra kept

a symmetric shape throughout the annealing process implying that full removal of defects,

especially those affecting long-range three-dimensional ordering, was not complete. The

authors also noticed the existence of a very weak band in all the Raman spectra at 182

cm-1

that they attributed to the presence of SWCNTs or MWCNTs, adventitiously

produced during the synthesis of these CNFs.

Kuvshinov et al. [67] studied the effect of HT on the degree of structural order of

three carbon nanofilaments with different microtextures as observed by TEM. All of them

were produced by catalytic decomposition of natural gas. The first one, obtained using a Ni

catalyst, was made of embedded cones of graphite with a solid core, thus fitting, according

to the authors, into the herringbone type of CNFs; whereas the second one, prepared in the

presence of a Ni-Cu catalyst, looked like a mixture of embedded-cone CNFs and octopus-

like platelet CNFs. Finally, the third one, produced over a Fe-Ni catalyst, was mainly

comprised of MWCNTs and some unusual chain-like CNFs. The CNFs were subjected to

HT at 1700, 2200 and 2600 ºC. The XRD analysis of the as-produced CNFs revealed the

graphitic-like nature of all of them with diffraction patterns showing reflections of (00l)

type and asymmetric diffuse (hk) peaks that correspond to turbostratic graphite. Average

interlayer distance d002 of 0.344 nm and Lc of 4.5 nm were calculated for all of these

CNFs. After HT, the (002) peak became narrower and more intense, evidencing the

improvement of the crystalline order in the CNFs. Therefore, crystallite sizes in the range

Page 21: Graphitization thermal treatment of carbon nanofibers

20

8.0-10.5 nm were determined for the CNFs heat-treated at 2600 ºC. As regards the

interlayer distance, a slight decrease was detected in the materials containing mainly CNFs,

leading to values of 0.342-0.341 nm. However, the interlayer distance of the MWCNTs did

not change by HT. The authors explained that the transformation of the nanotube structure

upon heating only occurs at the boundaries of its individual fragments and therefore, there

are not changes in interlayer distance.

Lee et al. investigated the annealing of CNFs produced by decomposition of

ethylene over a Ni-Cu catalyst at 600 ºC [68,69]. In their first report [68], the effect of HT

at 2200 ºC on these fibers (MJ CNFs), which appear to be of the herringbone type

according to TEM micrographs shown in this paper (vide infra), was compared with that

on commercial PR CNFs supplied by ASI with a stacked-cup microstructure. The first-

order Raman spectrum of CNFs [68,69] displayed two sharp peaks, at 1320 cm-1

,

attributed to the D band, and 1590 cm-1

, ascribed to a merger of G and D’ bands, as well

as a weak shoulder at 1179 cm-1

that was associated to the presence of functional groups

such as C=O formed during production and storage. Moreover, a broad band centered

around 1500-1550 cm-1

was also observed due to the presence of amorphous carbon with

higher intensity in the spectrum of PR CNFs. The D band in the spectrum of MJ CNFs

sharpened after HT, indicating a reduction of both functional groups and amorphous

carbon, the band at 1590 cm-1

split into the G and D’ bands, at 1580 and 1610 cm-1

,

respectively, and the ID/IG ratio decreased. Furthermore, a peak in the second-order Raman

spectrum at 2610 cm-1

, assigned to the G’ band of crystalline graphite appeared thus

confirming the graphitization of these CNFs. (Fig. 7). The Raman spectrum of the heat-

treated PR CNFs showed four sharp D, G, D’ and G’ bands, wherein the higher intensity of

the G’ band the authors claimed that denotes a more developed graphitic structure as

compared to the herringbone heat-treated MJ CNFs. The variation of the crystallite width

Page 22: Graphitization thermal treatment of carbon nanofibers

21

(La) as calculated from the ID/IG ratio was in agreement with this conclusion. Thus, a

bigger improvement of this parameter was assessed for PR CNFs (from 1.7 to 8.0 nm) than

for MJ CNFs (1.0 to 3.1 nm). The decrease of the interlayer spacing, d002, and the growth

of the crystallite size, Lc, calculated from XRD confirmed the significant improvement of

the degree of crystallinity of MJ CNFs and PR CNFs after HT at 2200 ºC. However, no

differences in the XRD parameters were observed. Thus, d002 values of 0.338 nm, and

crystallite thickness Lc of 11 nm were calculated for both heat-treated CNFs.

Fig. 7 - Raman spectra of as-prepared (PR, MJ), CVD-deposited (PRCVD, MJCVD)

and heat-treated (PRHT, MJHT) CNFs [68].

Finally, Weisenberger et al. investigated the effect of the graphitization temperature

on the structure of helical-ribbon CNFs [70], which are commercially produced by Grupo

Antolin Ingenieria (GAI) using a floating nickel catalyst. The CNFs were subjected to HT

at temperatures in the range of 1500-2800 ºC in a helium flow. On increasing temperature,

the (002) peak in the XRD profiles was observed to become sharper and to shift.

Page 23: Graphitization thermal treatment of carbon nanofibers

22

Therefore, the d002 value decreased from 0.3381 nm in the as-received to 0.3363 nm in the

heat-treated at 2800 ºC CNFs, whereas the crystallite thickness Lc increased from 8.8 nm to

13.0 nm, respectively. The decrease of the interlayer spacing was found to occur mainly in

the range of 1500-2500 ºC. The crystallite thickness, however, continued to grow over

2500 ºC. Another interesting point in the XRD study of these CNFs was the evolution of

the (112) reflection with HT. It was noticeable at 2250 ºC, and increasingly pronounced up

to 2800 ºC. Since this peak is characteristic of the three-dimensional order necessary for

AB stacking of graphite, it was inferred that HT above 2250 ºC induced the rearrangement

of the turbostratic graphene planes of the CNFs into alignment for this stacking structure

(Fig. 8). On the other hand, the intensities of the D and D’ Raman bands associated to

structural defects were observed to decrease with HT temperature. Therefore, the authors

claimed that the annealing of the CNFs appears an effective way to reduce the number and

severity of the graphitic defects, as illustrated by the decreasing ID/IG ratio.

Fig. 8 -Full scale XRD scans of as-received and heat-treated (1500-2800 ºC) CNFs

[70].

Page 24: Graphitization thermal treatment of carbon nanofibers

23

2.1.2. Microstructure (structure and microtexture)

As stated in the introduction of this review, the development of electron microscopy

techniques enabled the discovery of the nanometer size of carbon filaments in the early

1950s. Then, a further improvement in resolution was achieved in the 1980s and 1990s

with the introduction of high resolution TEM, thus allowing to ascertain the microstructure

of these filaments, which led to the distinction between CNTs and CNFs according to the

arrangement of the graphene planes in the filament (cylindrical or not, to state it in a

simplistic manner). Therefore, this technique is a powerful tool to monitor the evolution of

the microstructure of the CNFs during HT. Besides TEM, other techniques such as SEM or

scanning probe microscopy, especially scanning tunneling microscopy (STM) which

provides information at the atomic level, are also employed for a better and deeper

understanding of the graphitization process of CNFs.

Thus, we will start with the contributions of Mochida and co-workers who studied

the changes induced by HT in the morphology of platelet [49-51,71], herringbone [71] and

tubular [50,71] CNFs using SEM, TEM and STM techniques. As regards platelet CNFs,

both as-prepared and graphitized (at 2800 ºC) exhibited the typical alignment of graphene

planes for this type of morphology, arranged perpendicularly to the fiber axis, as seen in

the TEM images at high resolution (Fig. 9).

Page 25: Graphitization thermal treatment of carbon nanofibers

24

Fig. 9 - TEM images of platelet CNFs (a) as-prepared and (b) graphitized at 2800 ºC

[51].

In addition, the field emission SEM images showed the presence of platelet CNFs

of 80-300 nm width in as-prepared and graphitized CNFs, the former consisting of a

hexagonal column with flat surface bearing a hexagonal transverse shaped catalyst particle

on top. Graphitized CNFs maintained the same appearance, but did not carry the metal

particle, exhibiting a graphitic shell on top instead as a trace of the vaporized metal. TEM

images of graphitized CNFs showed a number of concentric loops (3-5 nm wide). When

observing the same spot upon tilting the sample +30º other loops appeared on the surface

of the fiber, this suggesting that the platelet CNFs consist of a structural unit with a

concentric loop end after graphitization. A low magnification STM scan across the

direction perpendicular to the fiber axis showed the transverse shape as a hexagon in the

graphitized CNFs. However, even the high magnification STM of as-prepared CNFs was

not clear enough to identify their shape probably due to insufficient arrangements of

carbon atoms. In contrast, rod-shaped units with a width of 2.5 nm were distinguished by

this technique in the graphitized CNFs, being all closely packed and stacked perpendicular

to the fiber axis to form a fiber (Fig. 10). Dome-like caps were formed in the end of these

Page 26: Graphitization thermal treatment of carbon nanofibers

25

rod units. A well-developed basal arrangement of carbon atoms was observed on the

surface along each unit, whose width was reported to correspond to that of the concentric

loop ends observed by TEM in the graphitized CNFs. This rod-like unit is considered by

the authors as a meso-dimensional substructure of platelet CNFs.

Fig. 10 - STM images of platelet CNFs (a) as-prepared and (b) graphitized showing

the presence of carbon nano-rod units [51].

The existence of carbon nano-rod units was further confirmed by TEM after their

separation from the graphitized platelet CNFs by mechanical milling (Fig. 11). Nano-rods

observed under STM and TEM showed variable lengths probably depending on the

catalyst particle size; and their diameter was easy to measure as each rod appeared to

consist of 4-5 graphene layers, accounting for a width of about 2.5 nm as above stated.

Page 27: Graphitization thermal treatment of carbon nanofibers

26

Fig. 11 - TEM image of nano-rod units in the graphitized CNFs [51].

All these microscopic observations, together with the XRD characterization

discussed in the previous section, allowed the authors to develop a three dimensional

model to explain the formation and the microstructure of the platelet CNFs. They proposed

that these nanofibers consist of nano-rods which are formed over the surface of the catalyst

particle, with a polygonal, especially hexagonal, cross-section, which enables their closed

packing perpendicularly to the fiber axis to obtain a single platelet CNF with high

crystallinity and density. When carbon nano-rods are graphitized, their ends are closed by

conical or pyramidal caps observed as concentrically-closed loops under TEM or dome-

like caps under STM. The group of nano-rods arrange in both the same and the

perpendicular fiber axis direction, giving larger crystallite thicknesses (Lc) and widths (La)

than those of the single nano-rods (Fig. 12). Graphitization was reported to emphasize such

arrangement increasing both values, although d002 values remained almost unchanged as

observed in the XRD characterization of these platelet CNFs.

Page 28: Graphitization thermal treatment of carbon nanofibers

27

Fig. 12 - Three dimensional models of nano-rods and platelet CNFs [51].

The HT at temperatures > 2000 ºC of the platelet CNFs was reported to remove

edge surface C-H bonds forming chemically active dangling sites on the edges of the

graphene-like layers [72]. Therefore, those layers of edges must be stabilized by bonding

each other in the way described above (loop formation), even when such bonding might

cause strain or tension due to the sharp curvature in the concentric loop alignment of

hexagons, thus limiting the graphitization extent and increasing the mechanical instability

as well as the chemical reactivity of such closed-end loops as was observed under TEM

after ball-milling and acid (HNO3) treatments, respectively. The former slightly distorted

the loop structure, whereas the latter definitely cut off the loop ends, exposing the free

edges and consequently improving the overall alignment of the hexagonal graphene planes

and therefore the graphitic order in agreement with the decrease of the interlayer spacing

d002 and the growth of the crystallites sizes measured by XRD [49].

The three-dimensional model was further generalized and extended to tubular and

herringbone CNFs in a later publication by Yoon et al. [71]. In this study, concentric loops

at the edge of the graphene layers forming nano-sized units were also observed to be

Page 29: Graphitization thermal treatment of carbon nanofibers

28

present in the as-prepared platelet CNFs under TEM and STM, thus inferring that these

units, that they called carbon nano-plates (CNPs), are structural building blocks of the

CNFs. These CNPs provided the same (002) lattice fringe pattern as that of graphites or

CNFs. The STM image of the as-prepared CNFs also confirmed the presence of carbon

nano-rod units (CNRs) but in a lesser extent. After graphitization, independent stacking

units in a fiber, each consisting of several graphenes, as well as the presence of transverse

shaped polygonal plate units and of the surface of carbon basal planes were imaged by

STM. Herringbone CNFs that were prepared by decomposition of ethylene over a Cu-Ni

catalyst at 580 ºC having the graphene planes tilted 50-70º with respect to the fiber axis

and diameters ranging from 50 to 450 nm, as well as lower degree of structural order than

the platelet CNFs were also found to be composed of CNRs and CNPs units. After

treatment at 2800 ºC, the graphene layers became well aligned and the surface edges were

closed with concentric loop-ends, similar to those of platelet CNFs but with lesser uniform

alignment. No change in the herringbone structure was detected in the graphitized CNFs.

Tubular CNFs with a high degree of graphitization and quite homogeneous diameters of

about 40 nm were formed by the association of CNRs as well. Detailed examination of

these CNFs by STM showed that the aligned CNRs keep their preliminary tubular

microstructure, whereas TEM images of graphitized tubular CNFs showed concentric loop

ends closed at both ends of the fiber (Fig. 13). Based on all of these observations by SEM,

TEM and STM, the authors concluded that the three types of CNFs studied were composed

of assemblages of nano-sized sub-structural units of carbon hexagon lattices, such as CNRs

or CNPs and their manner of assembly would define the macro three-dimensional structure

(i.e. platelet, herringbone, or tubular) (Fig. 14). The dimensions of the single CNRs units

were found to be approximately 2.5 nm in diameter and variable lengths depending on the

fiber dimensions. CNPs were proposed to be formed by association of several CNRs.

However, to ascertain their formation mechanism, dimensions and their relationships with

Page 30: Graphitization thermal treatment of carbon nanofibers

29

CNRs more studies will be required. Moreover, graphitization at 2800 ºC in all cases

resulted in the formation of dome-like caps (loops in TEM) at both ends of CNRs.

Fig. 13 - STM (a, b) and TEM (c) images of graphitized tubular CNFs [71].

Fig. 14 - Hypothetical model of single nano-rod and the relationship of rod- and plate-

type units [71].

Page 31: Graphitization thermal treatment of carbon nanofibers

30

More recently, Tamayo-Ariztondo et al. [73] have carried out a TEM study on

graphitized platelet and herringbone CNFs from commercial origin. The as-received

herringbone CNFs showed a preferred orientation, but the stacking was not regular. After

HT at 2750ºC, the graphene layers appeared parallel to each other, and the CNFs presented

a graphite-like structure with an interlayer distance of 0.337 nm. The formation of loops

joining the edges of the graphene layers was also observed, modifying the surface of the

fibers. The structural changes for the platelet CNFs were not as pronounced as those of the

herringbone although an interlayer spacing d002 of 0.337 nm was also measured after

graphitization. Similarly, the loop formation on the edges of the graphene layers of the

fibers was also observed (Fig. 15). Overall, the authors concluded that the degree of

crystallinity of the graphitized CNFs seemed to be very high.

Fig. 15 - TEM images of platelet CNFs (a) as-received and (b) heat-treated [73].

Chan et al. [74] reported the TEM analysis of the platelet-symmetric CNFs. These

CNFs were prepared through template-directed liquid crystal assembly and covalent

capture as described on a previous work [75]. They observed that the graphene layers in

most of the fibers were oriented approximately perpendicular to the fiber long axis. The as-

prepared CNFs showed short meandering graphene layers, typical of a low-temperature

carbon from a liquid crystal precursor [75]. At 2500 ºC, the platelet structure of the CNFs

Page 32: Graphitization thermal treatment of carbon nanofibers

31

was reported to be preserved in the interior of the fiber (both the strictly perpendicular and

the small proportion of tilted graphene layers) and the short, meandering fringes greatly

lengthened and straightened, but the free edges at the graphene layers disappeared (Fig.16).

Based on these observations, the authors claimed that the HT of the platelet-symmetric

CNFs was not the primary cause of the tilt, although they also reported that the available

data set did not allow for a trustworthy statistical ratio of the two types of structures.

Fig. 16 - TEM images of platelet CNFs: (a) as-prepared and (b) annealed at 2500 ºC

[74].

Habazaki et al. [56] studied the structural evolution during HT (1000-2800 ºC) of

PVC-based platelet CNFs by TEM. In the TEM images of the as-produced CNFs with a

diameter of about 30 nm (measured by SEM), spherical hollow regions with intervals of

several hundred nanometers were appreciated, probably owing to the formation of gas

during the decomposition of PVC. As regards the structure, these fibers were composed of

short layers of carbon atoms approximately normal to the fiber axis. After heating, such

orientation of the layers prevails and their stacking becomes more evident. Moreover, the

Page 33: Graphitization thermal treatment of carbon nanofibers

32

aspect of the CNFs surface changed from relatively smooth at 1500 ºC to ragged at 2800

ºC due to the formation of loops at the edges of the graphene layers. Apart from the loops,

the rather straight lattice fringes were also indicative of the improvement of the degree of

graphitization of the CNFs after heating at 2800 ºC (Fig. 17).

Fig. 17 - TEM images of CNFs: (a) as-produced (with SAED pattern) and (b, c, d)

heat-treated at 1000 ºC, 1500 ºC and 2800 ºC [56].

Fujisawa et al. [55] studied the effect of B-doping on the graphitization of platelet

CNFs by SEM and TEM techniques to complement the XRD and Raman data already

commented in the previous section. The as-grown CNFs were in the form of short rods

with a semi-rectangular cross-sectional morphology. The crystalline graphene layers were

stacked regularly along the fiber axis and the surface was covered with chemically active

edges planes. As expected, when the CNFs were treated at 2200 and 2500 ºC these active

edges were converted to energetically stable multi-loops. By doping the CNFs with boron,

the temperature at which these loops were formed decreased to 1900 ºC. Moreover, the

density of multi-loops on the outer surface of the CNFs was increased as observed in the

Page 34: Graphitization thermal treatment of carbon nanofibers

33

TEM images. From these observations, the authors concluded that the B atoms play an

important role in loop formation, leading to changes in the surface morphology during HT

at high temperature.

Cameán et al. [60] also observed by TEM the formation of loops between adjacent

active end planes during the graphitization of herringbone CNFs. This fact, according to

the authors, could account for the gaps in the expected continuously decreasing trend of

ID/It Raman ratio with increasing structural order, as assessed by XRD, since loops are

known to contribute to the D’ band intensity [55], attributed to end planes in carbon

materials [44].

The microstructural changes experimented by a mixture of platelet and herringbone

CNFs with diameters ranging from 10 to 300 nm after HT at 2500 ºC were investigated by

Zheng et al. [76] using TEM. The CNFs were prepared through the CVD method,

decomposing acetylene over a Ni catalyst at 700 ºC. The authors concluded that the type

and diameter of the as-prepared CNFs seemed to depend on the catalyst particle: the size

would affect the final diameter of the fiber, whereas shape and crystalline orientation

would establish the stacking pattern of the graphene layers comprising the fiber. Thus, they

observed that platelet CNFs were mainly formed by rectangular-shaped Ni particles,

whereas herringbone CNFs were formed by conical-shaped ones. After HT, the stacking of

the graphene layers was maintained and improved, but many unexpected ring-like (round-

head) structures like CNTs were observed on the surface of the CNFs, some of them

growing several nm out of the surface of the fibers because of the large strain caused by

their formation. These round heads were observed in both type of fibers (platelet and

herringbone) and consisted of 2-3 graphene layers curved to a diameter of 1-3 nm,

depending on the strain. It was claimed that the unstable atoms of the edges with dangling

bonds, present in the as-prepared CNFs, are prone to form round head structures at

elevated temperatures. However, this will generate strain at connected graphene layers

Page 35: Graphitization thermal treatment of carbon nanofibers

34

which would tend to enlarge the interlayer space, restrained at the same time by the Van

der Waals interaction between them. Therefore, it was concluded that if the diameter of the

CNFs is large enough, the round head structure would have little effect on the fiber, except

at the edges. However, if the diameter of the fiber is small enough the strain in the round

heads at the edges of the nanofiber will become so severe as to induce the enlargement of

the interlayer spacing between graphene layers therefore decreasing the structural order of

the graphitic structure.

The microstructural evolution during graphitization of three different types of CNFs

synthesized by Kuvshinov et al. [67] was tracked by TEM as well. The CNFs with a

diameter in the range of 15-120 nm showing a disordered graphite-like structure and basal

planes bent into embedded cones (herringbone type according to the authors) were

gradually transformed during HT, particularly as regards the microstructure of each fiber

which was observed to be now comprised of thick monolith conical structures. A wave-like

topography was formed on the surface due to locking of adjacent cones. Fibers with a

diameter of about 20 nm with a microstructure of stacks of graphene layers perpendicular

to the fiber axis (platelet) that were not present in the as-produced CNFs were observed

after HT at 2600 ºC. This fact was associated with the development of a graphite-like

structure (Fig. 18). CNFs with a mixed microstructure formed by octopus-like platelet and

embedded-cone fiber types having diameters in the range of 25-100 nm were restructured

by HT, specifically those of the latter type of 25 nm which changed their morphology to

platelet. Restructuring of the fibers was accompanied by the locking of adjacent graphene

layers and the formation of multilayered arch bridges. The HT of CNFs containing

MWCNTs and chain-like CNFs led to the appearance of new microstructures such as

individual enclosed fragments united by several common external graphene layers and

capsules built by graphene layers left after the volatilization of the metal particles which is

known as onion-like carbon. In summary, the authors reported that the degree of

Page 36: Graphitization thermal treatment of carbon nanofibers

35

restructuring of CNFs during HT depends on the initial microstructure, being significant in

CNFs having conical and platelet structures, gradually decreasing with the increase of the

fiber diameter.

Fig. 18 - (a) General morphology of as-produced herringbone CNFs built by

embedded cones, (b) the surface of a CNF and (c) individual CNF with a diameter of

~ 15 nm. (d) General morphology of heat-treated herringbone CNFs, (e) the structure

of a fiber with a diameter of ~ 70 nm and (f) an image of locked edges of graphene

layers on the fiber surface [67].

Tubular CNFs prepared by Ono and Oya [54] through the polymer blend technique

mentioned in the previous section were also characterized by SEM, field emission SEM

and TEM. Since these CNFs were stabilized by carbonization at 900 ºC prior to

graphitization at 3000 ºC, this microscopic characterization was focused on both. The

carbonized CNFs formed bundles, as it was seen in SEM micrographs, and presented

relatively uniform diameters of about 100 nm. TEM observations revealed that the surface

of these CNFs was smooth and no defects were observed. Moreover, they consisted of fine

carbon crystallites with a preferred orientation along the fiber axis. The microstructure of

Page 37: Graphitization thermal treatment of carbon nanofibers

36

the CNFs changed drastically after graphitization. Thus, SEM micrographs showed now

bundles of CNFs, some of them very thin, with approximately the same diameter as before.

Moreover, a highly crystalline structure was observed through TEM, with carbon crystals

well aligned along the fiber axis. These results were in good agreement with the XRD

analysis commented previously.

The structural changes induced on stacked-cup CNFs by HT at high temperatures

were studied in detail by Endo and co-workers [19,62,63,77] utilizing SEM and TEM

techniques. These CNFs exhibited long straight morphology, diameters in the range of 50-

150 nm and lengths up to 200 m, with a characteristic hollow core all along the fiber. In

addition, some differences in the wall thickness of the CNFs (outer / inner diameter ratio)

were observed through TEM micrographs, leading to the identification of two types of

microstructures: truncated conical graphene layers with graphite AB stacking (30-35

truncated cones with a spacing between layers of 0.34 nm, see fig. 1f) forming various

angles with the fiber axis and showing a large proportion of open edges on the outer

surface as well as in the inner channel, and coated nanofibers with a certain portion of

amorphous carbon (thickened CNFs, see fig. 1h) [19,62]. Significant morphological and

microstructural changes were reported to occur during HT of these stacked-cup CNFs at

high temperature (1800-3000 ºC), such as the transformation of the relatively smooth

tubular surface (wall) of the fibers into a rugged surface (saw-toothed with regular pitches

appearance) and the formation of energetically stable loops between the adjacent graphene

layers from the unstable edge planes in both the outer surface and the inner hollow core at

even the lowest treatment temperature of 1800 ºC (Fig. 19). Different types of loops were

identified: two by two, three by three, two by two with one unstable edge plane and one by

one; as well as different loop shapes especially in the outer surface of the fiber, such as

swelling type and plain type. Loop formation in the center of the sidewall of the

graphitized CNFs, which indicates the presence of a discontinuous defect phase, was also

Page 38: Graphitization thermal treatment of carbon nanofibers

37

observed by TEM. Finally, the authors concluded that (i) the morphology of the

graphitized CNFs resembles the stacking of truncated onion rings with an entirely hollow

core and (ii) the formation of multi-loops was responsible for the low degree of structural

order of the graphitized CNFs prepared in this work as assessed by XRD and Raman (see

previous section), since this fact was claimed to restrict the structural reorganization. Loop

formation at the edges of truncated cups, stacked one by one due to Van der Waals forces,

was proposed to occur via a zipping mechanism, theoretically investigated by Rotkin and

Gogotsi for other related graphitic carbons [78]. According to this mechanism, the

relatively unstable single loops would transform into more stable multi-loops below 2100

ºC. Above 2100 ºC, a somewhat decreased interlayer spacing caused by structural disorder

within the domains connected by loops, coupled with the structural stabilization between

domains, could be responsible for the jagged surface found for the graphitized CNFs. Endo

et al. also investigated the annealing of stacked-cup CNFs at lower temperatures (900-1500

ºC) [63]. No loops were observed after HT at 1000 ºC, but there were already some

undulated end planes. However, the presence of single or multi-loops was evident in the

TEM images of the CNFs treated at 1200 ºC. By increasing the temperature up to 1500 ºC,

the proportion of loops at the end of the edge planes increased as well. Based on these

TEM observations and also on XRD and Raman data, the authors proposed a model for the

growth of these CNFs during HT which consist in packets of 4-6 truncated cone graphene

layers interconnected forming long tubular structures during HT.

Fig. 19 - TEM images of (a) uncoated CNF after annealing at 3000 ºC and (b) coated

CNF after annealing at 3000 ºC [19].

Page 39: Graphitization thermal treatment of carbon nanofibers

38

Tzeng et al. [79] also studied the annealing in the temperature range of 1600-2400

ºC of stacked-cup CNFs by using SEM and TEM techniques. The CNFs showed a quite

uniform diameter distribution with a crooked morphology and a microstructure with

hollow cores separated by conical-shaped graphene layers. As reported by other authors

[19,62,63,77], loops were formed on both the outer and the inner surfaces of these stacked-

cup CNFs after HT at high temperature (≥ 1600 ºC), although those on the latter were

found to be fewer (Fig. 20). In Tzeng’s experiments, multi-loops were firstly observed

above 1800 ºC while unstable (destroyed under electron beam) single loops were already

found at 1600 ºC. As stated by Endo et al. [63], the loop formation and transformation

from single to multi-loops was dependent on many factors such as tube diameter, wall

thickness, crystallinity, truncated conical angles with respect to the tube axis, and the

amorphous carbon deposited on the outer surface of the CNFs.

Fig. 20 - TEM images of CNFs (a) grown by thermal CVD at 600 ºC and (b) annealed

at 2400 ºC [79].

Tubular-like CNFs showing a relatively wide hollow core surrounded by a

somewhat thinner carbon wall and an average diameter of 60 nm were prepared by Ci et al.

[80] employing the floating catalyst method from benzene and ferrocene at 1150 ºC. TEM

images showed that the walls of the CNFs were comprised of two layers: an inner layer,

catalytically grown from the iron particle, and an outer pyrolytic layer. The latter was also

Page 40: Graphitization thermal treatment of carbon nanofibers

39

composed of the less-crystalline interlayer near the inner layer and an outer amorphous

carbon layer. The HT at 2500 ºC led to significant microstructural changes, especially in

the inter-layer which becomes more ordered, with graphene sheets parallel to the fiber axis

(Fig. 21), and in the outer amorphous carbon layer that becomes thinner, whereas the inner

layer remained unchanged regarding the orientation (30º with respect to the fiber axis) and

the crystalline degree of the graphene sheets. Despite the fact that the authors did not report

it, it is evident that multi-loops have appeared in both surfaces of the inner layer of the

graphitized CNFs, similar to those described by Endo and co-workers in later publications

[19,62,63,77]. They also claimed that the graphene sheets of the inner-layer grew outward

toward the inter-layer, and the carbon atoms for its growth must come from the inter-layer.

It was assumed as well that the growth of inter-layer must depend on the carbon atoms of

its own or those of the outer-layer, as lower activation energy is needed for the diffusion of

amorphous carbon atoms than those of aromatic sheets of large size.

Page 41: Graphitization thermal treatment of carbon nanofibers

40

Fig. 21 - TEM images of CNFs (a) a wall of as-grown and (b) annealed. IN: inner-

layer; IT: inter-layer; O: outer-layer [80].

The graphitization of commercial stacked-cup CNFs was investigated by Shioyama

in 2005 [81]. The CNFs (GSI Creos Corp.) were heat-treated at 2800, 3000 and 3200 ºC.

TEM images of the graphitized CNFs showed evidence of structural changes resulting in a

composite texture: multi-graphene sheets rolled into concentric cylinders sheathing the

stacking morphology of truncated conical graphene layers. The texture of the outer sheath

was similar to that of MWCNTs and the stacking morphology of the inner structure was

the same as in the as-received CNFs. However, a detailed observation of the inner stacked-

cup microtexture in the graphitized CNFs revealed the presence of loops between the

adjacent graphene layers in both the inner (hollow core) and outer surfaces, in analogy

with Endo’s observations [19,62,63,77]. The author reported that the conversion from

stacked-cup texture to a MWCNT-stacked-cup composite texture was pronounced at

Page 42: Graphitization thermal treatment of carbon nanofibers

41

higher temperatures as a consequence of the release of the loop stress leading to the

rearranging of the carbon atoms into multi-graphene sheets rolled forming concentric

cylinders, although Endo et al. did not detect this sheath even after heating both coated and

uncoated stacked-cup CNFs at 3000 ºC [19]. Thus, a mean proportion of the sheath

thickness to the whole thickness of 0.69 was measured for the CNFs graphitized at the

highest temperature of 3200 ºC.

Local structural changes induced by HT on commercial thickened CNFs produced

by ASI were investigated in detail at the atomic level by Paredes at al. [66] through STM,

completing the XRD and Raman global structural characterization above commented. In a

general STM view of the as-grown CNFs, it was observed that they exhibited a very

smooth morphology with a diameter of about 100 nm. At the nanometer scale, these CNFs

displayed an isotropic platelet morphology which was transformed during HT into

different ones, firstly consisting of striped arrangements of increasing width, and finally

into large, atomically flat terraces up to several tens of nm wide separated by steps at 2800

ºC, similar to highly ordered graphites. As regards the STM at the atomic scale, the as-

grown CNFs were characterized by an absence of long-range graphitic order and only tiny

crystallites of < 2 nm were found. These crystallites were observed to coalesce into larger

yet defective units after HT at 1800 ºC. Atomic structures showing truly graphitic domains

were developed at 2300 ºC as evidenced by the appearance of the STM triangular pattern

with a periodicity of 0.25 nm although highly disordered sections were also generated at

this temperature attributed to the aggregation of mobile defects (atomic vacancies). Finally,

the atomic scale STM images on the terraces corroborated the long-range graphitic order of

the CNFs treated at 2800 ºC which displayed large crystalline areas exhibiting a low

coverage of small fragments of incompletely graphitized graphenes.

Howe et al. [64] have studied the effects of HT on commercial CNFs from ASI

(PR-19) by using TEM. The as-grown CNFs showed a hollow core about ½ to ⅔ of the

Page 43: Graphitization thermal treatment of carbon nanofibers

42

total fiber diameter (100-200 nm) surrounded by two layers of different morphologies; the

inner layer exhibiting herringbone morphology according to the authors, with graphene

layers canted about 25º with respect to the fiber axis and the outer layer that was formed by

turbostratic graphene planes which were on average parallel to the fiber axis. The Fe

catalyst particles appeared encapsulated in the hollow cores of the CNFs. The HT of the

CNFs was reported to progressively transform their initial duplex microstructure into a

new one composed of conical sections of 20 nm in size inclined 25º with respect to the

fiber axis with a higher degree of structural order (Fig. 22). These findings were in good

agreement with the XRD analysis reported in the previous section.

Fig. 22 - TEM images of CNFs (a) as-grown and (b) heat-treated at 2900 ºC [64].

In a subsequent article, Yoon et al. have reported the microstructural changes

experimented by commercial PR-24 and PR-19 CNFs during HT [65]. The authors

reported that the as-grown CNFs had stacked-cup geometry with circular cross sections.

They also described them as CNFs with a duplex morphology comprised of an inner core

covered with a layer of turbostratic carbon. TEM micrographs confirmed that the CNFs

were well graphitized and contained no metal impurities. Moreover, a theoretical model

based on the polygonization of the CNFs at sufficiently high temperatures and with

sufficiently large diameters, supported by TEM and SEM observations, was developed in

Page 44: Graphitization thermal treatment of carbon nanofibers

43

this work to explain the anomalous low values (< 0.3354 nm, the theoretical value of the

graphite crystal) determined by XRD for the interlayer spacing of the CNFs graphitized at

above 2500 ºC.

Scanning transmission electron microscopy (STEM) and TEM analyses were

employed by Lee et al. to follow the microstructural changes experimented by CNFs with

hollow and solid-core structures during HT at 2200 ºC [68], namely commercial thickened

stacked-cup (PR CNFs) and herringbone (MJ CNFs) which were specifically produced for

this study. In addition to the different core structure, the two types of CNFs selected were

reported to have very different surface morphology as observed in the STEM micrographs.

Thus, the surface of MJ CNFs was rougher than that of PR CNFs even after HT. As

reported in previous works, the formation of loops between the edges of graphene planes at

both walls (inner and outer) of the inner layer of the heat-treated PR CNFs was detected by

TEM. Loops together with more aligned interlayers were also observed to appear in the

heated MJ CNFs which was associated with the enhancement of the graphitic structure.

However, according to the authors, the most interesting TEM observation was the

significant spatial discontinuity appearing after the HT of both types of CNFs which they

claimed resulted from the reorganization of graphene layers as a consequence of the pore

collapsing and folding of graphene layers. Finally, a more significant improvement of the

crystalline structure of the hollow-core PR CNFs was observed by TEM in consistency

with the XRD and Raman data (Figs. 23-24).

Page 45: Graphitization thermal treatment of carbon nanofibers

44

Fig. 23 - TEM images of the double layer structure of hollow-core PR CNFs (a-b) as-

prepared and (c-d) heat-treated [68].

Fig. 24 - TEM images of the double layer structure of solid-core MJ CNFs (a-b) as-

prepared and (c-d) heat-treated [68].

SEM and TEM analyses were conducted for pyrolytically-stripped commercial PR-

24 CNFs, as-fabricated and heat-treated at 2800 ºC, by Ozkan et al. [82]. A description of

their microstructure, similar to that proposed by other authors previously mentioned in this

Page 46: Graphitization thermal treatment of carbon nanofibers

45

section, was also made. These CNFs displayed the typical duplex layer composition

(turbostratic outer layer, oblique graphene inner layer) with a hollow core and the already

known changes occur after HT: (i) loop formation on both inner and outer surfaces of inner

layer, thus provoking wedge discontinuities between layers and, (ii) graphitization of the

turbostratic outer layer, which is thinner than that of the initial CNFs, now exhibiting

longitudinally-aligned graphene layers. Additionally, SEM images (Fig. 25), showed the

fracture of the outer turbostratic layer in the as-fabricated CNFs as well as that of the co-

axial outer layer in the heat-treated CNFs. In both cases, protruding graphene layers from

the inner core with a truncated-cone structure were observed. Moreover, the cone angle of

the protruding segment of the fracture cross-sections is close to those shown in the

aforementioned TEM micrographs, indicating mutual sliding of the graphene planes of the

stacked-cup inner fiber structure.

Fig. 25 - SEM images of the fracture of commercial PR-24 CNFs (a) as-fabricated and

(b) heat-treated [82].

A very thorough and detailed TEM study on the structural transformation of

commercial PR-19 and PR-24 CNFs during HT at low (1500 ºC) and high (3000 ºC)

temperatures was carried out by Lawrence et al. [83]. The principal aim of this study was

to ascertain the real microstructure of these commercial CNFs produced by the vapor

growth process which have been the subject of a great number of research articles in the

Page 47: Graphitization thermal treatment of carbon nanofibers

46

last years, as well as the changes induced by different post-processing treatments,

including, among others, HT. In the first place, the authors found that most of the as-grown

CNFs had stacked-cup or cone-helix structure [22], but around a quarter of them presented

a segmented, bamboo-shaped structure that had been previously reported by others [84].

The relative proportion of the two microstructures was reported to be independent of the

production conditions. The conical CNFs showed the double-layer microstructure already

described in this section. It must be noted that the outer layer was present in most of the

cone fibers studied under TEM, the only difference between PR-24 and PR-19 being the

thinner diameter of the outer layer for the former. As for the microstructure of the inner

layer, regarded as a stacked-cup type by Endo and co-workers [19,62,63,77], Lawrence et

al. [83] found that it was incompatible with the wide variety of cone angles measured for

this inner layers by TEM [22] and proposed a cone-helix one instead. Subtle structural

changes were observed in these CNFs under HT at low temperature; thus, the ends of the

inner conical layers begin to curl and join. These changes were more pronounced by

increasing the temperature up to 3000 ºC. In fact, highly ordered outer layers often

consisting of MWCNTs were observed, although sometimes they were not straight or even

present at all. The structural changes of the inner layers were even more significant since

they were found to adopt now a multiwall structure with jagged edges which were reported

to be only compatible with stacked-cup morphology, as denoted also by the cone-angle

distribution. Since a similar angle distribution was not found in the conical CNFs treated at

low temperature, the authors concluded that the structural transformations of the inner

layer of these CNFs from cone-helix to stacked-cup morphology begins at 1500 ºC and

was almost complete at 3000 ºC. As regards the bamboo-shaped CNFs, the SEM images

clearly showed their distinctive segmented structure and the presence of metal catalyst

particles in some of the segments. Each segment was found to have a two-layer structure:

an outer disordered layer and an inner layer of MWCNTs with continuous multishell

Page 48: Graphitization thermal treatment of carbon nanofibers

47

fullerenes capping each segment. The authors proposed a reaction-diffusion mechanism to

explain the growth and structure of the bamboo-like shaped CNFs. The morphology of

these CNFs was strongly affected by HT up to 3000 ºC. Thus, the disordered outer layers

were transformed into ordered MWCNTs merging with the MWCNTs structure of the wall

of each segment, whereas the fullerene layers capping each segment were transformed to

sets of graphene layers joined together at oblique angles.

Finally, the structural rearrangement of commercial helical-ribbon CNFs with a

large hollow core and diameters in the range 40-140 nm [9] from GAI were investigated as

a function of graphitization temperature by Weisenberger et al. [70] using TEM. No

appreciable structural changes were detected after HT at 1400 ºC. Morphological changes

appeared clear at 2000 ºC: loop development on the surface of the fibers due to

dehydrogenation of internal and external edges of the helical graphite ribbon was observed.

By increasing the treatment temperature up to 2400 ºC, the graphene layers formed groups

of 15-30 planes layered in facets enabling the AB graphite stacking sequence, and the

cross-section became clearly polygonal, whereas further heating at 2800 ºC did not

provoke substantial changes in the fibers. The polygonization phenomenon was already

described by M. Yoon et al. for (supposedly) stacked-cup fibers [65]. Weisenberger et al.

stated that the rearrangement towards polygonal cross sections is morphologically in

agreement with the AB stacking structure within the planar facets composed of flattened

layers of the helical-ribbon layers and groups of layers, as verified by XRD with the

appearance of the (112) reflection (see Fig. 8).

2.1.3. Porosity (specific surface area and pore volume)

Mochida and co-workers [49,51,85] studied the evolution of the porosity of platelet CNFs

during graphitization at 2800 ºC by measuring the specific surface area (BET surface area)

Page 49: Graphitization thermal treatment of carbon nanofibers

48

which decreased markedly from 91 to 32 m2g

-1. Further milling and acidic treatment of the

graphitized CNFs were found to recover in part the initial surface area of the as-produced

CNFs. Although the authors did not discuss this point at all, the loops at the ends of the

graphene layers which were reported to uniformly cover the surface of the graphitized

CNFs appeared to be, among other possible factors, responsible for the decrease of the

surface area since distortion in the alignment of these loops and even disappearance of the

latter, recovering the free graphene edges were, respectively, observed by TEM after the

above mentioned treatments.

Similarly, a decrease of the surface area was also noticed by Tamayo-Ariztondo et

al. after HT at 2750 ºC of herringbone and platelet CNFs [73]. This fall was much higher

for herringbone (from 164.81 to 22.4 m2g

-1) than for platelet CNFs (from 72.43 to 51.09

m2g

-1). The authors conjectured that this fact was due to the vaporization and further

disappearance during HT of the smallest CNFs. Therefore, the mean diameter of the

graphitized CNFs was larger, thus reducing the surface area.

The porosity changes experimented by non-CVD produced platelet CNFs under HT

in the 1000-2800 ºC range of temperature were studied by Habazaki et al. [56]. The

specific surface area of as-produced CNFs deeply decreased (from 250 to 70 m2g

-1) with

increasing temperature up to 1500 ºC. However, above this temperature, it remained

basically constant. Based on this and considering the adsorption/de-sorption nitrogen

isotherms, the authors concluded that the CNFs heat-treated above 1000 ºC did not have

porosity. Therefore, since micropores were detected in the as-produced CNFs, their

elimination during HT appeared responsible for the fall of the surface area (Fig. 26).

Page 50: Graphitization thermal treatment of carbon nanofibers

49

Fig. 26 - Variation of the specific surface area (BET) of platelet CNFs with HT

temperature [56].

The variation of the textural characteristics of platelet, herringbone and tubular

CNFs during HT in the temperature range 1700-2600 ºC were reported by Kuvshinov et al.

[67]. The as-produced CNFs were mostly mesoporous, with some microporosity as well.

The mesopores were formed by free spaces between interlaced nanofibers, whereas

micropores were created by relatively large defects on the surface of the fibers. As

expected, the surface area of CNFs decreased when the treatment temperature was

increased. However, the significance of this decrease, which was associated with the

removal of the surface defects and the thinnest CNFs, was found to be related to the

surface structure of the CNFs, with those having lower energies (formed by graphene

layers) being more stable towards the influence of the temperature than surfaces with

higher energy (i.e. formed by edges of graphene layers). Therefore, platelet and

herringbone CNFs with edge graphene layers showed higher surface area fall than those

having cylindrical graphene layers such as MWCNTs. The authors finally proposed that

the surface area decrease could be due to partial recrystallization of carbon since carbon

Page 51: Graphitization thermal treatment of carbon nanofibers

50

transfer may occur by surface diffusion as well as through the gas phase. The driving force

of the process was the difference in saturated partial pressures of carbon above the surface

of the fibers of different curvature. This partial pressure should be higher in fibers with a

small diameter and highly bent surface. The mechanism was finally confirmed by the

dependence of pore distribution on temperature as seen in Fig. 27 in which the area below

pores of 3-5 nm decreased significantly by increasing HT temperature.

Fig. 27 - Pore size distributions of as-produced and heat-treated (1700-2600 ºC)

herringbone CNFs [67].

Garcia et al. studied the effect of the HT at high temperature (2400-2800 ºC) on the

porosity of the herringbone CNFs mentioned in previous epigraphs [57]. The authors

reported the surface area as well as its distribution in meso- and micro-porosity. The CNFs

were mainly mesoporous with surface areas ranging from 97 to 107 m2g

-1. The HT reduced

their surface area significantly (up to 2.8 times at 2800 ºC) at the expense of both the

mesopores located in the nanofiber interior along the axis (up to 2.9 times) and the

micropores caused by graphene layer defects (up to 2.2 times). On the basis of the parallel

evolution of the texture (surface area) and structure (interlayer spacing, crystallite sizes) of

the CNFs during HT, it was stated that the decrease of the surface area was associated with

Page 52: Graphitization thermal treatment of carbon nanofibers

51

the increase of the degree of structural order as a consequence of the removal of defects

and the growth of the crystallite size.

Lee et al. compared the porosity changes induced by HT at 2200 ºC on herringbone

MJ CNFs and commercial stacked-cup PR CNFs [68]. First of all, it was found that the

herringbone MJ CNFs possessed much larger surface area and pore volume (up to 8.5

times) than PR CNFs. A remarkable porosity decrease was detected after the HT of MJ

CNFs. In contrast, no significant changes in the values of the surface area / pore volume

were appreciated for PR CNFs when heated. Based on these results and considering the

crystalline parameters of the graphitized CNFs above reported, the authors suggested that

the observed reduction of the porosity (surface area and pore volume) of the MJ CNFs

during HT led to pore collapsing by diffusion of carbon species, thus limiting their

graphitization and at the same time, providing an explanation for the lower degree of

structural order finally observed for these CNFs as compared to the commercial PR CNFs.

In a different publication, the same authors measured the surface area and pore volume for

the same type of herringbone CNFs after HT showing the progressive decrease on both

magnitudes by increasing the temperature [69]. But, according to SEM images, there was

no significant change in surface roughness even after HT at 2200 ºC.

The variation with temperature (500-2800 ºC) of the specific surface area and the

corresponding micropore surface area of stacked-cup CNFs was investigated by Endo et al.

[63]. An abrupt increase of the surface area, closely related to the development of

micropores, was detected at 900 ºC, followed by a progressive decrease to reach a

minimum value of about 50 m2g

-1 at temperatures ≥ 1500 ºC which, in fact, was similar to

that of the CNFs heat treated at 500 ºC. The authors claimed that the variation of the

surface area was directly connected with the morphological changes of the graphene edge

sites on the outer / inner surface of the CNFs imaged by TEM. That is, the single-loop

formation between the adjacent truncated conical layers at 900-1200 ºC after the instability

Page 53: Graphitization thermal treatment of carbon nanofibers

52

produced by the evolution of hydrogen gas and the transition from single- to more

energetically favorable multi-loops above 1500 ºC. After the formation of multi-loops,

little morphological changes were observed (Fig. 28).

Fig. 28 - Variation of specific surface area and micropore area of stacked-cup CNFs

with the HT temperature (established from the nitrogen absorption at 77 K) [63].

Zhou et al. [86] also tracked the changes on porosity experimented by CVD-

produced fishbone CNFs with mesoporous character during HT at 1700 ºC. A decrease of

the BET surface area of the CNFs attributed to the removal of surface defects in

accordance with the XRD and HR-TEM data was reported. However, a parallel increase of

the total pore volume of the CNFs was also measured, but not explained by the authors.

2.1.4. Structural defects

Thermogravimetric techniques performed under an air (or oxygen) atmosphere permit to

evaluate the stability of carbon materials towards oxidation, which is closely related to the

number of defects on their surface since the reaction with oxygen for carbons progresses at

structural defects such as vacancies, face edges and others [87]. Therefore, in this section

Page 54: Graphitization thermal treatment of carbon nanofibers

53

we will present the scarce number of studies found in the literature reporting the effect of

graphitization on the oxidation temperature of CNFs.

Garcia et al. [57] reported that the HT of herringbone CNFs decreased their oxygen

reactivity considerably as shown by the shift of the temperature of maximum oxidation rate

to higher values ( 150 ºC) which was associated with the removal of structural defects

(Fig. 29). Unlike XRD and Raman measurements which have an average character, the

analysis of temperature programmed oxidation (TPO) profiles of the graphitized CNFs

revealed the presence of two carbon species with different degree of structural order, the

less ordered being progressively transformed into the more ordered carbon structure by

increasing the temperature.

Fig. 29 - TPO profiles of as-produced (CNF01-NiCuTi) and graphitized (CNF01-

NiCuTi/2400-2800) CNFs [57].

Lee et al. studied the effect of HT at 2200 ºC on the oxygen reactivity of

herringbone MJ CNFs and commercial stacked-up PR CNFs by performing the thermal

gravimetric analysis in a CO2 environment, followed by the measurements of the onset

(temperature at which 5 wt % loss was reached) and the complete oxidation temperatures

[68]. Increases of both temperatures were detected after the HT of MJ and PR CNFs, the

Page 55: Graphitization thermal treatment of carbon nanofibers

54

latter showing higher values which the authors reported as an indication of their more

ordered structure in agreement with the XRD and Raman spectroscopy results. Therefore,

it was concluded that HT at 2200 ºC was more effective in changing the structure of the

stacked-up PR CNFs than that of the herringbone MJ CNFs.

Similarly, Zhou et al. [86] carried out the thermal gravimetric analysis in an air

atmosphere of fishbone CNFs. A significant increase of the onset temperature from 530 ºC

to 660 ºC was detected after HT at 1700 ºC that was attributed to the higher degree of

structural order of the heat-treated CNFs.

2.2. Changes in mechanical and electrical properties

The HT at high temperature of CNFs modifies, often enhancing, their mechanical and

electrical properties, a fact to take into account for further applications such as polymer

additives to produce novel composites or in the field of electronics and energy storage

devices. Therefore, in this section we will review the studies on this subject, more

specifically those related to the electrical conductivity/resistivity and the tensile strength of

the CNFs.

Kuvshinov et al. [67] investigated the specific electrical resistance changes

experimented by two CNFs with herringbone and a mixture of platelet/herringbone

microstructures during HT at 2600 ºC. The value of this electrical property was reported to

decrease significantly (by more than 2 times) as a consequence of the increase of the

crystallinity degree. However, it was still ten times higher than that of graphite. The

authors explained that this difference was mainly due to the different texture of the

materials. They also suggested that the higher resistance of the graphitized CNFs could be

linked with an additional contact resistance between individual fibers and the reduced cross

section of the sample due to porosity.

Page 56: Graphitization thermal treatment of carbon nanofibers

55

Fujisawa et al. measured the electrical resistivity in the bulk state of platelet CNFs

(determined by both the contact resistivity between the fibers and the intrinsic resistivity of

the individual fibers) and the electrical resistivity of an individual nanofiber (as-grown and

heat-treated in the range of 1900-2500 ºC) [55]. Although no general relationship between

the treatment temperature and the electrical resistivity of an individual nanofiber was

observed (possibly due to the inhomogeneous diameter distribution of the CNFs as the

authors claimed), those treated at the highest temperature (2500 ºC) showed the lowest

value, even though the decrease as regards the as-grown nanofiber was not remarkable.

The electrical resistivity of these CNFs was found to be extremely high (ca. 280 k)

compared to nanofilaments with other microstructures such as MWCNT (ca. 2.4 k) [88].

The platelet morphology of the CNFs was reported to account for this difference since the

current flow between graphene layers in graphite along the fiber length direction is six

times lower than that in the graphene plane [89].

A progressive decrease of the electrical resistivity in the bulk state of stacked-cup

CNFs by increasing HT temperature was reported by Endo et al. [62]. A drastic decrease

was observed by heating the CNFs at 1800 ºC. According to the authors, the elimination of

hydrogen or water absorbed from air could explain it because these CNFs were found to

have a higher proportion of active edge sites on the outer surface. The evolution of the

electrical resistivity of the CNFs treated at temperatures above 1800 ºC was related to the

formation of loops on the outer surface. Therefore, no large changes of this property were

observed in the region of 2300-3000 ºC, which was consistent with the XRD, Raman and

microstructure data commented in previous sections.

The variation of the intrinsic electrical conductivity of commercial PR-19 CNFs

during HT in the temperature interval 1100-2900 ºC was studied by Howe et al. [64]. In

this work, the CNFs were compressed in a cylindrical die, and the intrinsic electrical

resistivity was measured as a function of the volume fraction at a certain pressure.

Page 57: Graphitization thermal treatment of carbon nanofibers

56

Measurements made at very low volume fractions were observed to suffer from contact

problems as well as sensitivity to the details of loading. Therefore, the discussion was

focused on data collected at volume fractions above 5.5 %. The electrical conductivity was

found to reach a maximum at 1500 ºC with values ranging from 1 -1

cm-1

up to 11 -1

cm-1

for volume fractions of 6 % and 50 %, respectively. Surprisingly, although the

crystallite size of heat-treated CNFs increased at temperatures above 2000 ºC, the electrical

conductivity was found to drop. The authors reported that the electron scattering at the

well-defined grain boundaries growing between the conical crystallites formed during the

HT of the CNFs as observed by TEM was responsible for this drop.

Lee et al. [69] measured the volume resistivity of pellets of herringbone CNFs (as-

grown and heat-treated at 2200 ºC) by applying a pressure of 6.8 MPa. The HT increased

significantly the intrinsic electrical conductivity of the CNFs as shown by the fall of the

resistivity from 0.47 cm to 0.18 cm, in agreement with the results reported by Endo et

al. for stacked-cup CNFs recently commented [62].

A detailed study on the mechanical properties of commercial PR-24 CNFs

(thickened stacked-cup nanofibers), as-grown and heat-treated at 2800 ºC, was conducted

by Ozkan et al. [82]. In this work, the elastic modulus and the strength of the CNFs were

measured by a Microelectromechanical Systems-based (MEMS) mechanical testing

platform. The HT increased the elastic modulus of the CNFs which was attributed to the

graphitization of the turbostratic layer; however, the average and the characteristic

strengths were reduced. Therefore, the authors concluded that the improvement in the

thermal and electrical properties of the CNFs by HT was achieved at the expense of the

mechanical strength. On the other hand, the measured strength data of heat-treated CNFs

exhibited lower standard deviation than that of the as-grown CNFs, implying a broad flaw

distribution in the turbostratic layer, which was annealed during HT.

Page 58: Graphitization thermal treatment of carbon nanofibers

57

3. Applications of graphitized carbon nanofibers

In this section, we will review the applications of graphitized CNFs reported so far in the

literature, placing special emphasis on the advantages (or disadvantages) of employing

these heat-treated nanofibers instead of the as-grown ones provided that the authors give

this information. Therefore, four different types of applications will be considered: (i)

composites (additive for polymer reinforcement, metal-coated nanofibers, etc.); (ii) energy

storage devices: mainly Li-ion batteries (anodes or fillers) and fuel cells (cathode support);

(iii) hydrogen storage, and (iv) other applications (potential biosensors and catalysts in

diverse reactions).

3.1. Composites

Many efforts have been devoted to the study of composites based on CNFs in recent years.

Most of this work has been covered in two reviews by Tibbetts et a. [29] and Alsaleh et al.

[30]. Moreover, results on the preparation and properties of heat-treated CNFs/polymer

composites were also included in the latter. The present epigraph covers all of the works

reported in the literature about this type of composites with regard to the effect of the

graphitization or high-temperature HT of the CNFs on their final properties, particularly on

mechanical and electrical properties. It must be noted that most of the research in this field

has been focused on the use of commercial Pyrograf® (PR) CNFs supplied by ASI, widely

mentioned already in this review, although there have been a few studies involving other

types of CNFs as well, that will be commented at the end of this section.

An important part of the research on composites of PR CNFs was carried out by

Tibbetts and co-workers, who fine-tuned the industrial scale production of this type of

Page 59: Graphitization thermal treatment of carbon nanofibers

58

fibers early in the 1990s [90-95]. Among other factors, the effect of the graphitization of

the CNFs on the electrical conductivity of CNFs/polypropylene (PP) and CNFs/nylon

composites was studied [93]. The CNFs were of PR III type showing a diameter of 0.2 m

and they were graphitized at 3000 ºC. It was found that HT of the fibers at this temperature

strongly decreased the electrical resistivity of both types of composites, specifically at fiber

volume fractions above 3 %. Under these conditions, composites with electrical properties

suitable for applications such as electrostatic painting or static discharge were prepared.

The authors concluded that the electrical conductivity of the composites depended on the

degree of graphitization of the fibers and on their surface conductivity. In a subsequent

report, they studied in addition to the electrical, the mechanical properties of the above

mentioned CNFs/PP composites [94]. Once again, it was observed that the composites

made with the graphitized CNFs had higher electrical conductivity. Moreover, the average

fiber length was reported to also affect the electrical properties with those longer

graphitized fibers leading to composites with the lowest resistivity. In this sense, the

authors stated that composites with room-temperature electrical resistivity slightly larger

than that of the single crystal graphite were prepared in this work by using graphitized

fibers of 10 m. However, the poor values of the tensile strength and the stiffness of the

CNFs/PP composites were not improved at all by employing graphitized CNFs. In a

follow-up article [95], they investigated the mechanical properties of PP composites having

CNFs with different degree of graphitization. It was concluded that both the tensile

strength and the modulus tend to decrease with the graphitization index of the CNFs

(calculated from the value of the interlayer spacing d002) as a consequence of the poor

adherence to the PP matrix as observed in the SEM images. The effect of HT (1100-2900

ºC) of commercial PR-19 CNFs on the mechanical and electrical properties of PP

composites (containing CNFs from 3 to 12 vol. %) was studied in greater detail by Howe

et al.[64]. The electrical conductivity of the composites was lower than that measured for

Page 60: Graphitization thermal treatment of carbon nanofibers

59

the CNFs, already commented in a previous section, since the polymer makes fiber-fiber

contact more difficult. However, the variation of the neat fibers and CNFs/PP composites

conductivities with HT temperature followed similar trend, the latter showing a maximum

closer to 1300 ºC instead of 1500 ºC (the conductivity peak found for the fibers). In line

with this, superior mechanical properties as regards tensile strength and Young’s modulus

were measured for those composites made with CNFs heat-treated at 1500 ºC. Higher HT

temperatures were observed to have a negative effect on these composite properties which

in the case of the tensile strength was attributed to the poor PP binding with the lower

surface energy of the graphitized CNFs. The authors concluded that CNFs with a

filamentary core of conically nested graphene planes graphitize into discontinuous conical

crystallites as it was observed by TEM. Apparently, this discontinuous structure does not

give optimal mechanical or electrical properties to the composites in which these CNFs are

used.

The effect of HT of commercial PR-24-XT-PS CNFs from ASI on the electrical

properties and the crystallization behavior of CNFs/PP composites were studied by Lee et

al. [96]. A significant drop of the composite volume resistivity (from 1011

-cm in pure

PP to 104 -cm) was observed after the addition of only a 3 wt % of heat-treated (2300

ºC) CNFs. However, higher amounts of as-grown CNFs were needed to make composites

conductive. This difference was due to the decrease of the intrinsic volume resistivity of

the CNFs after HT. Furthermore, the heat-treated CNFs were found to be more effective

nucleation agents, thus leading to composites with higher crystallization temperature which

the authors reported to be related with the more graphitic structure of these CNFs as

inferred from TEM and STM images. However, the crystallinity and crystallization rates

for the composites with the heat-treated and the as-grown CNFs did not show significant

differences, suggesting that these CNFs did not affect the growth of the polymer crystalline

structure in a significant manner.

Page 61: Graphitization thermal treatment of carbon nanofibers

60

Kuriger et al. [97] measured the thermal conductivity and electrical resistivity in

both the longitudinal and the transverse directions of PP composites reinforced with heat-

treated (3000 ºC) PR-19 CNFs. The resistivity decreased with the volume fraction of the

CNFs in the composite, finally reaching lower values than those of glass fiber-reinforced

polymers. As regards thermal conductivity, it was found to progressively increase with the

volume fraction of the CNFs, particularly in the longitudinal direction. The authors

claimed that the values measured for this property were considerable higher that those

obtained by other researchers using as-grown CNFs. This difference was attributed to the

higher conductivity of the heat-treated CNFs.

Lafdi et al. also studied the effect of HT (1500-3000 ºC) of commercial PR-24

CNFs on the physical properties of CNFs/epoxy resin composites [98]. The strength of

adhesion between the CNFs and the epoxy matrix was characterized by the flexural

strength and modulus of the composite. The latter was found to increase with increasing

CNFs loading (4, 8 and 12 %) and CNFs heating temperature up to 1800 ºC. However,

higher temperatures which led to an increase of the graphitic order of the CNFs resulted in

a decrease of the composite modulus, particularly for 12 wt % loading, thus being

indicative of poorer adhesion to the epoxy resin matrix. The authors stated that it was a

consequence of the elimination of truncated free edges of the graphene layers on the

nanofiber surface during graphitization as observed by TEM, thus decreasing the number

of sites available for bonding with the polymer matrix. As regards the nanocomposite

strength, again it was found to increase with the treatment temperature of the CNFs up to

1800 ºC. However, unlike the modulus, it decreased with the increase of the load of CNFs

in the composite. Therefore, as compared to the modulus, a minimum improvement of the

composite strength was attained by reinforcing with these heat-treated CNFs which

according to the authors was an indication of the lack of optimization of the chemical

adhesion between the CNFs and the epoxy matrix. A drastic improvement of the thermal

Page 62: Graphitization thermal treatment of carbon nanofibers

61

and electrical properties of the composites was achieved by using the heat-treated PR-24

CNFs. Thus, as compared to the neat epoxy resin, the electrical resistivity of the

composites was decreased by nine orders of magnitude in the case of the CNFs treated at

3000 ºC. In a similar manner, the thermal conductivity of the neat epoxy resin was found to

increase significantly with the addition of these CNFs. The authors explained that the

alignment of the graphene layers of the CNFs during graphitization led to a more efficient

transfer of phonon and electron, thus resulting in a decrease in electrical resistivity and an

increase in thermal conductivity of the CNFs/epoxy resin composites. They finally

concluded that a compromise should be reached for the utilization of these

nanocomposites: if good mechanical properties are desired, then the CNFs should not be

heat-treated above 1800 ºC prior to the reinforcement of the epoxy resin, but HT to 3000

ºC would be beneficial if good thermal and electrical properties should be attained.

Memon and Lafdi [99] prepared buckypaper thermal interface materials (TIMs)

with various heat-treated (1100 ºC, 1500 ºC and 3000 ºC) PR CNFs, using polyvinyl

alcohol as binder. They explored the potential of these materials to sustain transient power

spikes by efficiently dissipating maximum heat to the sink by measuring their thermal

resistance. All TIMs were subjected to two types of heat loads: uniform pulse and transient

power spike. The parametric study showed significant decrease in the thermal resistance of

these TIMs with increasing HT temperature of the CNFs. Moreover, they outperformed

other TIMs like thermal pastes o greases with more than 50 % enhancement in thermal

transport from heat source to heat sink because of the graphitic nature of the CNFs which

made them more conductive.

Xu et al. [100] prepared CNFs/vinyl ester composites employing PR-19 CNFs heat-

treated at 3000 ºC. The composites were characterized by measuring the flexural strength

and modulus as well as the electrical resistivity as a function of CNFs loading. Moreover,

the results were compared to those using as-grown PR-19-PS CNFs. Since the heat-treated

Page 63: Graphitization thermal treatment of carbon nanofibers

62

CNFs were more conductive materials than the as-grown ones, it was expected that the

corresponding composites would maintain this difference. However, the composites based

on PR-19-PS CNFs exhibited lower volume electrical resistivity at fiber content above the

percolation threshold. The authors suggested that this could be due to the aspect ratio

differences existing after mixing during the preparation of the composite, since the heat-

treated CNFs appeared more brittle. In the same vein, they claimed that the higher surface

activity of these heat-treated CNFs could lead to a thicker vinyl ester coating during

mixing, thus inducing poorer electrical contact between fibers. However, they do not rule

out other possible variables accounting for this contradictory result.

Chen et al. [101] synthesized boron nitride (BN) coatings on commercial PR-PS

CNFs (as-grown and heat-treated at 3000 ºC) through a simple two-step process employing

boric acid and ammonia. They observed that the surface structure of the fibers had a strong

influence on the morphology of the boron nitride coating. Thus, while a uniform

crystalline, smooth, homogenous and symmetric coating was observed by TEM to be

formed on the surface of the heat-treated CNFs, a polycrystalline coating appeared on the

surface of the as-grown CNFs. As the surface of the latter CNFs was turbostratic, the

nucleation of BN crystals occurs more easily in many places. However, the seeds had

different orientations producing a polycrystalline coating (Fig. 30).

Fig. 30 - TEM images of BN-coated PR-PS CNFs (a) as-grown and (b) heat-

treated (3000 ºC) [101].

Page 64: Graphitization thermal treatment of carbon nanofibers

63

Composites with other types of CNFs different from the above reported commercial

PR were also investigated by other researchers. In this context, Lee et al. [69] prepared PP

composites containing different loadings of as-produced and heat-treated (1200 ºC, 1800

ºC and 2200 ºC) CNFs of herringbone type [70]. In this work, the effect of HT of CNFs on

the electrical properties (volume resistivity) of PP composites was studied. In general, a

significant drop of the volume resistivity of the composite was observed for loadings > 5

wt % of both the as-produced and heat-treated CNFs. However, higher values were

measured for the composites made with the latter CNFs, particularly with those prepared

with 10 wt % of CNFs treated at 2200 ºC. The authors concluded that although the HT of

these CNFs improved their crystallinity, specifically at 2200 ºC, as assessed by XRD, the

observed parallel decrease of the surface area and the length reduction by mixing during

the composite preparation could account for this unexpected result, suggesting that the

morphology of CNFs plays an important role in the electrical and thermal properties of the

CNFs/polymer composites.

Kang et al. [102] investigated the effect of HT at low temperatures (700-1000 ºC)

of herringbone CNFs that were prepared by a CVD process on the mechanical properties of

CNFs(Ni/Y)-Cu composites by determining the compressive yield strength values. It was

observed that the yield strength of the composite increased substantially employing the

heat-treated CNFs. At temperatures below 800 ºC, this improvement was continuous with

increasing temperature. However, at 1000 ºC or higher, the strength of the composite was

reduced. The authors pointed out that the carbon fraction in the composites decreased with

increasing temperature due to the elimination of amorphous and disordered carbon, thus

leading to thinner CNFs. By considering the same fraction of CNFs in the composite, the

strengthening efficiency with heat-treated CNFs at 800 ºC was 88 % higher than that using

the as-produced CNFs. However, at 1000 ºC or above, the catalyst alloyed with Cu and

Page 65: Graphitization thermal treatment of carbon nanofibers

64

formed large bead-like particles of hundreds of nanometers in diameter as observed by

TEM, thus reducing the yield strength of the composite.

Tamayo-Ariztondo et al. [73] reinforced copper with commercial platelet and

herringbone CNFs, as-produced and heat-treated at 2750 ºC, as well as already heat-treated

longitudinally aligned CNFs, to reduce the coefficient of thermal expansion of this metal

which is of importance for its application as heat sink material. In this work, the CNFs

were coated with Cu using electrochemical deposition. As seen by SEM, the best Cu

coating (dense and uniform) was obtained for longitudinally aligned CNFs (already heat-

treated) and for the heat treated herringbone CNFs, particularly, after chemical oxidation

treatments of the surface to create anchor points (carbonyl, carboxyl or hydroxyl groups).

Concerning platelet CNFs, Cu was not deposited in the as-produced CNFs, whereas it was

barely deposited in the form of agglomerates in the heat-treated ones. From these findings,

the authors concluded that the quality of the Cu coating was very dependent on the

structure of the CNFs, with those having graphene layers parallel to the fiber axis showing

the best results due to the exposure of bonds in the outer surface. On the contrary, the Cu

deposition was inhibited on surfaces perpendicular to the graphene layers, as is the case for

platelet CNFs. In general, the HT of the CNFs was found to improve the quality of the Cu

coating dramatically, particularly in the case of the herringbone type. The authors related

this fact to the formation of loops at the ends of the graphene layers, thus providing a good

surface for Cu deposition similar to that encountered in the longitudinally aligned CNFs.

Page 66: Graphitization thermal treatment of carbon nanofibers

65

3.2. Energy storage devices

3.2.1. Lithium-ion batteries

Carbon materials have been used in lithium-ion battery systems either as the electrode

itself or as an additive due to their excellent mechanical, electrical and thermal properties.

More specifically, the attractive features of these materials for such applications are their

high electrical conductivity and good corrosion resistance in many electrolytes. In this

sense, CNFs also possess these properties and were first tested for these applications by

Endo et al. [35]. They used what they called submicron VGCFs comprised of a central

filament surrounded by a pyrolytic outer deposit of amorphous carbon, which straightens

out as the HT temperature increases to form an annular tree-like structure (thickened

nanofilaments) with a diameter of 0.2 m. The anode performance of the graphitized (2800

ºC) CNFs and their additive effect in conventional anode materials for Li-ion batteries

were studied in this work. As regards the anode performance, battery discharge capacities

of 283 mAhg-1

and cycle efficiencies of about 77 % were attained. The authors claimed

that as compared to data from VGCFs with diameters of 2 m, these graphitized CNFs

showed a relatively higher capacity as a consequence of their large surface area which is

caused by the smaller diameter. Moreover, they exhibited fairly good cyclic efficiency

even after 200 charge-discharge cycles (Fig. 31). The graphitized CNFs also showed

acceptable properties as fillers for electrodes. Thus, the cyclic efficiency of synthetic

graphite was found to increase continuously with the added proportion (up to 10 wt %) of

the graphitized CNFs and it was maintained at almost 100 % up to 50 cycles. It was

concluded that the addition of these CNFs to the anode improved (i) the conductivity due

to the high electrical conductivity of the fibers themselves and the network formation of

the fibers with the graphite particles, and (ii) the ability to absorb and to retain significant

Page 67: Graphitization thermal treatment of carbon nanofibers

66

electrolyte because of the large surface area and small diameter which also provided a

homogenous fiber distribution and consequently, good electrolyte penetration. Moreover,

they provided resiliency and compressibility to the electrode structure.

Fig. 31 - The variation of discharge capacity when graphitized submicron VGCFs are

used as anode material in the range 0 to 1.5 V with a current density of 0.2 mAcm-2

[35].

Yoon et al. [85] studied the effect of the graphitization of platelet CNFs on their

anodic performance in Li-ion batteries by galvanostatic charge/discharge cycling. The as-

produced CNFs already showed a relatively high discharge capacity in the 1st cycle (278

mAhg-1

). Annealing of the CNFs at 2000 ºC and 2800 ºC slightly increased the discharge

capacity (300 and 324 mAhg-1

, respectively), all comparing well with the commercial

graphite used in this study as a reference material (300 mAhg-1

). However, a fairly large

reversible charge was observed for both the CNFs and the reference graphite. Thus, first

cycle coulombic efficiency values of 69 %, 63 % and 65 % were calculated for the heat-

treated CNFs, the as-produced CNFs and the graphite, respectively. The authors suggested

that since the formation of loops between the edges of the graphene planes during the

graphitization of the CNFs as observed by TEM did not appear to influence on the

Page 68: Graphitization thermal treatment of carbon nanofibers

67

coulombic efficiency (irreversible capacity), the very low values calculated for this

parameter should be related to the large surface area of the CNFs (both the as-produced

and the graphitized). Finally, it was concluded that CNFs are promising candidates for Li-

ion battery anodes provided that fine-tuning and control of their properties were achieved.

No data about efficiency along prolonged cycling was reported in this work.

Heat-treated platelet CNFs, this time prepared by liquid phase carbonization of

PVC, were also tested as anodes for Li-ion batteries by Habazaki et al. [103]. In this work,

particular attention was paid to the rate capability (charge-discharge cycle performance) of

the CNFs as a function of the diameter and the degree of crystallinity. The CNFs heat-

treated at 1000 ºC with a very low degree of structural order as seen by TEM images

showed higher capacities and rate capabilities compared to those heat-treated at 1500ºC

and even 2800ºC, the latter exhibiting a typical graphitic structure in which loops between

the edges of the graphene layers were formed. These highly graphitic CNFs provided a

reversible capacity of less than 200 mAhg-1

at a current density of 50 mAg-1

which is far

from the theoretical capacity of graphite (372 mAhg-1

). The authors proposed that this low

value was related to the presence of loops which could hinder the intercalation of lithium

ions into all of the graphene layers. Moreover, the reversible capacity and the cycle

performance were improved by reducing the diameter of the CNFs with the only

disadvantage being the large loss of capacity during the first cycle due to the formation of

the solid-electrolyte interface. In fact, it was found that the irreversible capacity increased

as the CNFs diameter decreased.

Fujisawa et al. also investigated the anodic performance in lithium-ion batteries of

heat-treated (1900-2500 ºC) CNFs, both as-grown and boron-doped, with platelet

morphology [55]. An increase of the lithium discharge/charge plateau below 0.2 V, which

was related to the reversible capacity, was observed upon increasing the HT temperature of

the CNFs. Moreover, those graphitized B-doped CNFs provided higher capacities and were

Page 69: Graphitization thermal treatment of carbon nanofibers

68

found to degrade to a lesser extent at high-discharge current density. The authors explained

that the presence of B atoms in the graphitized CNFs was responsible for the improvement

of lithium ion adsorption since they were found to have higher electrical conductivity and

degree of crystallinity as well as more graphene layers loop-ended surfaces which were

more stable (or inert) against electrolytes. The large amount of Li adsorption onto B-

substituted planar carbon material was theoretically confirmed in a previous work also

from Kurita and Endo [104] by molecular orbital calculations. Unlike other reports from

the Endo et al. [35], no data about efficiency along prolonged cycling was reported.

The electrochemical performance as potential anodes in lithium-ion batteries of

graphitic materials that were prepared by HT (2800-2900 ºC) of herringbone CNFs was

investigated by Cameán et al. through galvanostatic cycling [59]. The graphitized CNFs

provided reversible capacities up to 320 mAhg-1

after 50 discharge/charge cycles, these

values being comparable to those of oil-derived graphite, currently employed as anode in

the commercial lithium-ion batteries (Fig. 32). In addition to that, they showed excellent

cyclability and cyclic efficiency (> 99 %). The authors claimed that the nanometric size of

these CNFs seemed to favor the diffusion of the lithium ions into the materials, thus

improving their electrochemical performance, in accordance with the latter works above

commented. It was concluded that apart from the degree of crystallinity of the graphitized

CNFs, the presence of loops at the end of the graphene edges observed by TEM appeared

to affect the reversible capacity of the battery. Moreover, the surface area and mesopore

volume of these materials were reported to influence on the irreversible capacity and on the

capacity retention along cycling, respectively.

Page 70: Graphitization thermal treatment of carbon nanofibers

69

Fig. 32 - Extended galvanostatic cycling of the graphitized CNFs and of the SG

(commercial graphite employed as anode in lithium-ion batteries) [59].

Finally, the electrochemical properties of PAN-derived CNFs as anodes in lithium-

ion batteries as a function of the HT temperature (700-2800 ºC) were evaluated by Kim et

al. [105]. No binder or conductive filler was used in this case in the fabrication of the

active anode material. A very low reversible capacity (130 mAhg-1

) was measured for the

CNFs heat-treated at 2800 ºC. This fact was attributed to the low degree of crystallinity of

the thermally treated PAN-based CNFs as deduced from the XRD and Raman parameters

of these materials, which are in fact classified as typical non-graphitic carbons. However, it

was remarked that the CNFs treated at only 1000 ºC showed a very good reversible

capacity of 450 mAhg-1

(larger than the 372 mAhg-1

theoretical value of graphite) that the

authors explained was originated from the peculiar morphology of these materials showing

a highly disordered structure, with defects and dangling bonds. Moreover, based on the

absence of any distinct redox peak in the derivative voltage profiles, it was stated that the

lithium ions were inserted/removed into/from the CNFs by a doping/de-doping process.

Furthermore, when the rate capability at increasing current density (30-100 mAg-1

) was

considered, no large capacity degradation as compared to graphite was observed, which

confirmed that the lithium ion diffusion path within the anode was highly reduced as a

Page 71: Graphitization thermal treatment of carbon nanofibers

70

consequence of the small diameter of fibers. From these results, together with the constant

and slightly inclined charge potentials found for the low-temperature treated CNFs, it was

concluded that these materials were excellent candidates for anodes of high-power lithium-

ion batteries.

3.2.2. Fuel cells

The potential of CNF-supported platinum catalyst as electrode for fuel cell applications

was first investigated by Bessel et al. [36] in 2001. They used the oxidation of methanol at

40 ºC as a probe reaction and as-produced platelet, herringbone and ribbon CNFs that were

previously purified to remove the residual catalyst metals. As observed by TEM, the

platinum crystallites dispersed in the CNFs were relatively thin, showing highly faceted

structures, which was associated with a strong metal-support interaction. Based on the

results of methanol oxidation, the authors concluded that the nature of the carbon support

influenced on the activity of the metal catalyst. Thus, it was found that the oxidation

activity of Pt (5 wt %) supported on platelet or ribbon CNFs was highly improved (400 %)

as compared to the commercial standard Vulcan carbon. Additionally, a decrease of

catalysts self-poisoning was observed. The authors suggested that this improvement in

catalyst performance was linked to the fact that the metal particles adopt specific

crystallographic orientations when dispersed on the structure of the CNFs, which is an

important factor since it is known that certain faces of the metal were more active than

others for methanol oxidation. Finally, it should be remarked that although in this work

heat-treated (graphitized) CNFs were not studied, it was included in the present review

attending to its foremost character.

In 2008, Gang et al. [106] reported a significant improvement of the activity

towards oxygen reduction reaction (ORR) of direct methanol fuel cells of Pt catalyst by

Page 72: Graphitization thermal treatment of carbon nanofibers

71

using heat-treated (2800 ºC) herringbone CNFsas support instead of the as-produced

CNFs obtained from Yoon et al. [71] and previously described in this review). They

suggested that the formation of loops at the ends of the graphene layers during HT as seen

by TEM could enhance the surface diffusion of oxygen on the CNFs. More recently [107],

the same group has carried out a thorough study in which the activity of the above

mentioned Pt/graphitized CNFs and Pt/as-produced CNFs catalysts in ORR was compared

to that of the commercial Pt/Vulcan. Once again, Pt/graphitized CNFs outperformed Pt/as-

produced CNFs, followed by Pt/Vulcan in tests at 30ºC and 60 ºC. The authors concluded

that the high degree of graphitization of the CNFs improved the performance and

durability of the Pt supported catalyst, this implying a higher electrical conductivity and

removal of O-containing groups from the surface of the fibers and therefore a more

hydrophobic surface for water removal, as well as the formation of loops that possess

topological defects for uniform metal deposition,

Ko et al. [108] investigated the effect of HT temperature in the interval of 1600-

2800 ºC on the electrochemical corrosion of CNFs in polymer electrolyte membrane fuel

cells by monitoring the generation of CO2 at constant potential during electrochemical

oxidation. No mention of the fiber morphology was given in the paper as the Pt/CNFs

catalyst was obtained from a commercial source. The resistance of the CNFs to

electrochemical corrosion was found to increase by increasing treatment temperature, in

particular at temperatures ≥ 2400 ºC. Based on the XPS and XRD characterization of the

catalysts, the authors concluded that this improvement was only related to the reduction of

the oxygen functional groups which increases the hydrophobic nature of carbon surfaces

since the degree of structural order of the CNFs did not change significantly during HT.

Page 73: Graphitization thermal treatment of carbon nanofibers

72

3.3. Hydrogen storage

An early study by Chambers et al. [109] unveiled that purified as-produced CNFs were

able to sorb and retain hydrogen in an amount over an order of magnitude higher than that

found with conventional hydrogen storage systems. Platelet, herringbone and tubular CNFs

with a graphitic character were employed in this study. The authors stated that the unique

crystalline arrangement existing within the graphite nanofiber structure, where the platelets

generate a system comprised entirely of slit-shaped nanopores in which only edge sites are

exposed, could easily overcome diffusion limitations, thus accounting for the vast amount

of hydrogen sorbed. Moreover, due to the weak bonding of the platelets, the non-rigid wall

nanopores can expand to accommodate hydrogen in a multilayer configuration with the

presence of delocalized -electrons on the graphene layers being also a major contributory

factor to the gas adsorption. Finally, a major fraction of the hydrogen stored by the CNFs

was reported to be released at room temperature by lowering the pressure. As stated above,

although heat-treated (graphitized) CNFs were not studied in this work, it was included in

the present review attending to its foremost character. These results generated controversy

in the scientific community and prompted Tibbetts et al. [110] to carefully measure the

sorption of hydrogen by various types of carbon materials, including commercial PR CNFs

from ASI both as-produced and graphitized at 3000 ºC. According to their measurements,

none of the materials showed hydrogen sorption appreciably above background and

concluded that all the claims of more than 1 wt % of hydrogen sorption by carbon

materials at room temperature are erroneous or simply due to experimental errors.

Two years after the latter publication, Zhu et al. [111] measured the sorption of

hydrogen in thickened CNFs with hollow cores prepared by the floating catalyst method

and subjected to HT at different temperatures, up to 2200 ºC. It was found that hydrogen

sorption increased with increasing HT temperature of the CNFs, reaching a maximum

Page 74: Graphitization thermal treatment of carbon nanofibers

73

value of 4 wt % for the CNFs heat-treated at 2200 ºC, thus contradicting the previous

results obtained by Tibbetts [110]. To account for this, they claimed that three features

favored hydrogen sorption in the CNFs, these being (i) a suitable crystalline state, (ii) the

exposed edges on the surface and (iii) the absence of O-containing functional groups on the

surface of the fibers.

3.4. Other applications

CNFs were also investigated for other potential applications such as biosensors and drug

delivery vehicles. For this purpose, it is of paramount importance to know how CNFs

interact with antibodies and proteins. Thus, Naguib et al. [112] examined the effect of the

surface structure of CNFs on the adhesion of monoclonal CD3 antibodies. In their

experiments they used as-produced (pyrolytically stripped) and heat-treated at 3000 ºC

commercial CNFs from ASI. Binding of the proteins to nanofibers was enhanced by poly

(L-Lysine) (PLL) and improved by increasing disorder and hydrophilicity of the

nanofibers’ surface. Therefore, the more disordered as-produced CNFs showing a higher

presence of O-containing functional groups improved the wetting and attachment to the

PLL and proteins compared to the hydrophobic, well-ordered heat-treated CNFs, with an

oxygen-free surface full of loops. The authors concluded that the wall structure is a major

factor that needs to be taken into account in order to determine the potential of CNFs for

their use in biomedical applications.

Parrot et al. [113] used the Terahertz time domain spectroscopy to study the

electrical and optical properties of a series of PR-19 CNFs (as-produced pyrolytically

stripped and heat-treated at 1500 ºC and 3000 ºC) to account for their different catalytic

activity in the oxidative dehydrogenation of ethylbenzene to produce styrene. The

adsorption coefficient and real refractive index of the CNFs were measured by this

Page 75: Graphitization thermal treatment of carbon nanofibers

74

technique, and a relationship between these values and their structural order was

established. The least graphitic CNFs (pyrolytically stripped) were found to be the most

active in the catalytic process followed by the heat-treated at 1500 ºC and 3000 ºC.

Wang et al. [114] examined the effect of HT of activated CNFs (obtained by

stabilization, carbonization and activation of electrospun PAN-based CNFs) on their

catalytic activity in the oxidation of NO to NO2. The oxidation conversion was

dramatically improved by employing the heat-treated (graphitized) CNFs. The authors

explained that the graphitization of the CNFs provided better active oxidation sites and

more topological defects (observed under SEM and TEM) which were reported to be also

useful for NO oxidation. It was concluded that pore size and volume, surface functional

groups, and graphitization potential could be tailored by the extent of the activation and

graphitization process of the CNFs, which resulted in different catalytic activities for NO

oxidation.

4. Concluding remarks

In this review the structural and textural changes showed by CNFs during graphitization,

and how these changes influence their mechanical and electrical properties as well as

potential applications have been considered. Despite all the different types of CNFs

according to their microstructure (platelet, herringbone, tubular, etc.) or synthetic process

(CVD, electrospinning, etc.) some clear patterns can be established for all. Thus, in terms

of structure, an overall improvement of the graphitic three-dimensional order often occurs

as denoted by the decrease of the interlayer spacing (d002) accompanied by the growth of

the crystallite sizes (Lc and La) as measured by XRD, and a decrease of the ID/IG Raman

ratio, implying substantial removal of defects from the lattice as well. In any event, the

extent of this improvement turned out to be very variable depending on the CNFs under

Page 76: Graphitization thermal treatment of carbon nanofibers

75

study and no apparent trends could be observed to account for this. The observations under

microscopic techniques, specially by TEM, are in good agreement with the latter, generally

showing better aligned stacks of graphene planes after HT in the range of 2500-3000 ºC

and the characteristic formation of loops (and multi-loops) at the edges of the graphene

planes connecting adjacent layers already at temperatures below 2000 ºC regardless of the

microstructure of the CNFs. With regard to the porosity and the surface area, a general

decrease of both was detected upon HT which is associated with the removal of surface

defects together with the vaporization of the smallest CNFs. Thermogravimetric studies of

the oxidation temperature of CNFs are also in line with the results obtained with other

techniques showing that the heat-treated CNFs, with a higher degree of structural order,

also displayed higher oxidation temperatures. Concerning the electrical properties, despite

the scarce number of works in the literature it can be concluded that an overall

improvement in the electrical conductivity (or decrease in the electrical resistivity) occurs

after the HT of the CNFs. The only study of mechanical properties of individual thickened

stacked-cup CNFs showed that HT at 2800 ºC increased the elastic modulus, but decreased

the nanofiber strength, thus concluding that an improvement in the thermal and electrical

properties is achieved at the expense of the strength.

Finally, in terms of the applications of CNFs, the degree of graphitization seems to

play a very important role in polymer reinforcement as, in most cases, employing small

loadings of graphitized CNFs dramatically enhance the conductivity and thermal properties

in the novel composites as compared to as-produced CNFs, although the mechanical

properties are often worse due to poorer contact between the polymer and the CNFs and

therefore a compromise in HT temperature needs to be reached. The better electrical

properties and the higher surface inertness of graphitized CNFs apparently improved their

performances in energy storage devices such as anodes in lithium-ion batteries or catalyst

supports in fuel cells. As regards hydrogen storage, the results so far obtained are

Page 77: Graphitization thermal treatment of carbon nanofibers

76

inconclusive and somewhat contradictory and more research needs to be done in order to

know the exact effect of HT on CNFs for this application. The activity of as-produced and

heat-treated CNFs in other applications such as catalysts or biosensors was also tested with

divergent results.

Acknowledgements

Financial support from the Spanish Ministry of Economy and Competitiveness MINECO

(under Projects ENE2008-06516 and ENE2011-28318) is gratefully acknowledged. A.R.

thanks the Spanish Research Council for Scientific Research (CSIC) for a JAE-Doc

contract, co-funded by the European Social Fund (ESF).

REFERENCES

[1] Iijima S. Helical microtubules of graphitic carbon. Nature 1991;354(6348):56-8.

[2] Hughes TV, Chambers CR, inventors; Manufacture of carbon filaments. US patent

405480, 1889.

[3] Radushkevich LV, Lukyanovich VM. About the structure of carbon formed by

thermal decomposition of carbon monoxide on iron substrate. Zurn Fisic Chim

1952;26:88-95.

[4] Davis WR, Slawson RJ, Rigby GR. An unusual form of carbon. Nature

1953;171:756.

[5] Baker RTK, Harris PS. The formation of filamentous carbon. In: Walker PL, Jr.,

Thrower PA, editors. Chemistry and physics of carbon, vol 14; New York; Dekker;

1978, p. 83-165.

[6] Baker RTK. Catalytic growth of carbon filaments. Carbon 1989;27(3):315-23.

Page 78: Graphitization thermal treatment of carbon nanofibers

77

[7] Baker RTK, Barber MA, Harris PS, Feates FS, Waite RJ. Nucleation and growth of

carbon deposits from the nickel catalyzed decomposition of acetylene. J Catal

1972;26(1):51-62.

[8] Oberlin A, Endo M, Koyama T. Filamentous growth of carbon through benzene

decomposition. J Cryst Growth 1976;32(3):335-49.

[9] Martin-Gullon I, Vera J, Conesa JA, Gonzalez JL, Merino C. Differences between

carbon nanofibers produced using Fe and Ni catalysts in a floating catalyst reactor.

Carbon 2006;44(8):1572-80.

[10] Rodriguez NM. A review of catalytically grown carbon nanofibers. J Mater Re

1993;8(12):3233-50.

[11] de Jong KP, Geus JW. Carbon nanofibers: catalytic synthesis and applications.

Catal Rev-Sci Eng 2000;42(4):481-510.

[12] Nadarajah A, Lawrence JG, Hughes TW. Development and commercialization of

vapor grown carbon nanofibers: a review. Key Eng Mat 2008;380:193-206.

[13] Teo KBK, Singh C, Chhowalla M, Milne WI. Catalytic synthesis of carbon

nanotubes and nanofibers. In: Nalwa HS, editor. Encyclopedia of Nanoscience and

Nanotechnology, vol X; American Scientific Publishers; 2003, p.1-22.

[14] Melechko AV, Merkulov VI, McKnight TE, Guillorn MA, Klein KL, Lowndes

DH, et al. Vertically aligned carbon nanofibers and related structures: controlled

synthesis and directed assembly. J Appl Phys 2005;97(4):041301-39 p.

[15] Chronakis IS. Novel nanocomposites and nanoceramics based on polymer

nanofibers using electrospinning process—A review. J Mater Process Tec

2005;167(2-3):283-93.

[16] Liu C-K, Lai K, Liu W, Yao M, Sun R-J. Preparation of carbon nanofibres through

electrospinning and thermal treatment. Polym Int 2009;58(12):1341-9.

Page 79: Graphitization thermal treatment of carbon nanofibers

78

[17] Nataraj SK, Yang KS, Aminabhavi TM. Polyacrylonitrile-based nanofibers—A

state-of-the-art review. Prog Polym Sci 2012;37(3):487-513.

[18] Rodriguez NM, Chambers A, Baker RTK. Catalytic engineering of carbon

nanostructures. Langmuir 1995;11(10):3862-6.

[19] Endo M, Kim YA, Hayashi T, Fukai Y, Oshida K, Terrones M, et al. Structural

characterization of cup-stacked-type nanofibers with an entirely hollow core. Appl

Phys Lett 2002;80(7):1267-9.

[20] Terrones H, Hayashi T, Muñoz-Navia M, Terrones M, Kim YA, Grobert N, et al.

Graphitic cones in palladium catalysed carbon nanofibres. Chem Phys Lett

2001;343(3-4):241-50.

[21] Vera-Agullo J, Varela-Rizo H, Conesa J, Almansa C, Merino C, Martin-Gullon I.

Evidence for growth mechanism and helix-spiral cone structure of stacked-cup

carbon nanofibers. Carbon 2007;45(14):2751-8.

[22] Ekşioğlu B, Nadarajah A. Structural analysis of conical carbon nanofibers. Carbon

2006;44(2):360-73.

[23] Endo M, Takeuchi K, Hiraoka T, Furuta T, Kasai T, Sun X, et al. Stacking nature

of graphene layers in carbon nanotubes and nanofibres. J Phys Chem Solids

1997;58(11):1707-12.

[24] Rodriguez NM, Kim M-S, Baker RTK. Carbon nanofibers: a unique catalyst

support medium. J Phys Chem 1994;98(50):13108-11.

[25] Serp P, Corrias M, Kalck P. Carbon nanotubes and nanofibers in catalysis. Appl

Catal A-Gen 2003;253(2):337-58.

[26] Bitter JH. Carbon nanofibers in catalysis – Fundamental studies and scope of

application. In: Murzin DY, editor. Nanocatalysis; Research Signpost; 2006, p. 99-

125.

Page 80: Graphitization thermal treatment of carbon nanofibers

79

[27] Lozano K. Vapor-grown carbon-fiber composites: Processing and electrostatic

dissipative applications. JOM 2000;52(11):34-6.

[28] Kang I, Heung YY, Kim JH, Lee JW, Gollapudi R, Subramaniam S, et al.

Introduction to carbon nanotube and nanofiber smart materials. Compos Part B-

Eng 2006;37(6):382-94.

[29] Tibbetts GG, Lake ML, Strong KL, Rice BP. A review of the fabrication and

properties of vapor-grown carbon nanofiber/polymer composites. Compos Sci

Technol 2007;67(7-8):1709-18.

[30] Alsaleh MH, Sundararaj U. A review of vapor grown carbon nanofiber/polymer

conductive composites. Carbon 2009;47(1):2-22.

[31] Tran PA, Zhang L, Webster TJ. Carbon nanofibers and carbon nanotubes in

regenerative medicine. Adv Drug Deliver Rev 2009;61(12):1097-114.

[32] Saito N, Aoki K, Usui Y, Shimizu M, Hara K, Narita N, et al. Application of

carbon fibers to biomaterials: a new era of nano-level control of carbon fibers after

30-years of development. Chem Soc Rev 2011;40(7):3824-34.

[33] Hammel E, Tang X, Trampert M, Schmitt T, Mauthner K, Eder A, et al. Carbon

nanofibers for composite applications. Carbon 2004;42(5-6):1153-8.

[34] Huang J, Liu Y, You T. Carbon nanofiber based electrochemical biosensors: a

review. Anal Methods 2010;2(3):202.

[35] Endo M, Kim YA, Hayashi T, Nishimura K, Matusita T, Miyashita K, et al. Vapor-

grown carbon fibers (VGCFs): Basic properties and their battery applications.

Carbon 2001;39(9):1287-97.

[36] Bessel CA, Laubernds K, Rodriguez NM, Baker RTK. Graphite Nanofibers as an

Electrode for Fuel Cell Applications. J Phys Chem B 2001;105(6):1115-8.

[37] Oberlin A. Carbonization and graphitization. Carbon 1984;22(6):521-41.

Page 81: Graphitization thermal treatment of carbon nanofibers

80

[38] Biscoe J, Warren BE. X-ray study of carbon black J Appl Phys. 1942;13(6):364-

71.

[39] Franklin R. The structure of graphitic carbons. Acta Crystallogr 1951;4(3):253.

[40] Iwashita N, Park CR, Fujimoto H, Shiraishi M, Inagaki M. Specification for a

standard procedure of X-ray diffraction measurements on carbon materials. Carbon

2004,42:701-714.

[41] Tuinstra F, Koenig JL. Characterization of graphite fiber surfaces with Raman

spectroscopy. J Compos Mater 1970;4:492-9.

[42] Zerda TW, John A, Chmura K. Raman studies of coals. Fuel 1981;60(5):375-8.

[43] Lespade P, Marchand A, Couzi M, Cruege F. Caracterisation de materiaux

carbones par microspectrometrie Raman. Carbon 1984;22(4–5):375-85.

[44] Katagiri G, Ishida H, Ishitani A. Raman spectra of graphite edge planes. Carbon

1988;26(4):565-71.

[45] Cuesta A, Dhamelincourt P, Laureyns J, Martinez-Alonso A, Tascon JMD. Raman

microprobe studies on carbon materials. Carbon 1994;32(8):1523-32.

[46] Afanasyeva NI, Jawhari T, Klimenko IV, Zhuravleva TS. Micro-Raman

spectroscopic measurements on carbon fibers. Vib Spectrosc 1996;11(1):79-83.

[47] Endo M, Nishimura K, Kim YA, Hakamada K, Matushita T, Dresselhaus MS, et al.

Raman spectroscopic characterization of submicron vapor-grown carbon fibers and

carbon nanofibers obtained by pyrolyzing hydrocarbons. J Mater Res

1999;14(12):4474-7.

[48] Wang Y, Serrano S, Santiago-Avilés JJ. Raman characterization of carbon

nanofibers prepared using electrospinning. Synth Met 2003;138(3):423-7.

[49] Lim S, Yoon S-H, Mochida I, Chi J-H. Surface modification of carbon nanofiber

with high degree of graphitization. J Phys Chem B 2004;108(5):1533-6.

Page 82: Graphitization thermal treatment of carbon nanofibers

81

[50] Tanaka A, Yoon S-H, Mochida I. Preparation of highly crystalline nanofibers on Fe

and Fe–Ni catalysts with a variety of graphene plane alignments. Carbon

2004;42(3):591-7.

[51] Yoon S-H, Lim S, Hong S-h, Mochida I, An B, Yokogawa K. Carbon nano-rod as a

structural unit of carbon nanofibers. Carbon 2004;42(15):3087-95.

[52] Murayama H, Maeda T. A novel form of filamentous graphite. Nature

1990;345(6278):791-3.

[53] Charlier JC, Michenaud JP. Energetics of multilayered carbon tubules. Phys Rev

Lett 1993;70(12):1858-61.

[54] Ono H, Oya A. Preparation of highly crystalline carbon nanofibers from

pitch/polymer blend. Carbon 2006;44(4):682-6.

[55] Fujisawa K, Hasegawa T, Shimamoto D, Muramatsu H, Jung YC, Hayasaki T, et

al. Boron atoms as loop accelerator and surface stabilizer in platelet type carbon

nanofibers. ChemPhysChem 2010;11(11):2345-8.

[56] Habazaki H, Kiriu M, Hayashi M, Konno H. Structure of the carbon nanofilaments

formed by liquid phase carbonization in porous anodic alumina template. Mater

Chem Phys 2007;105(2-3):367-72.

[57] Garcia AB, Cameán I, Suelves I, Pinilla JL, Lázaro MJ, Palacios JM, et al. The

graphitization of carbon nanofibers produced by the catalytic decomposition of

natural gas. Carbon 2009;47(11):2563-70.

[58] Garcia AB, Cameán I, Pinilla JL, Suelves I, Lázaro MJ, Moliner R. The

graphitization of carbon nanofibers produced by catalytic decomposition of

methane: Synergetic effect of the inherent Ni and Si. Fuel 2010;89(8):2160-2.

[59] Cameán I, García AB, Suelves I, Pinilla JL, Lázaro MJ, Moliner R. Graphitized

carbon nanofibers for use as anodes in lithium-ion batteries: Importance of textural

and structural properties. J Power Sources 2012;198:303-7.

Page 83: Graphitization thermal treatment of carbon nanofibers

82

[60] Cameán I, García AB, Suelves I, Pinilla JL, Lázaro MJ, Moliner R, et al. Influence

of the inherent metal species on the graphitization of methane-based carbon

nanofibers. Carbon 2012;50(15):5387-94.

[61] Oberlin A, Rouchy JP. Transformation des carbones non graphitables par traitement

thermique en presence de fer. Carbon 1971;9(1):39-46.

[62] Endo M, Kim YA, Hayashi T, Yanagisawa T, Muramatsu H, Ezaka M, et al.

Microstructural changes induced in ―stacked cup‖ carbon nanofibers by heat

treatment. Carbon 2003;41(10):1941-7.

[63] Endo M, Lee BJ, Kim YA, Kim YJ, Muramatsu H, Yanagisawa T, et al.

Transitional behaviour in the transformation from active end planes to stable loops

caused by annealing. New J Phys 2003;5:121.1-9.

[64] Howe JY, Tibbetts GG, Kwag C, Lake ML. Heat treating carbon nanofibers for

optimal composite performance. J Mater Res 2006;21(10):2646-52.

[65] Yoon M, Howe J, Tibbetts G, Eres G, Zhang Z. Polygonization and anomalous

graphene interlayer spacing of multi-walled carbon nanofibers. Phys Rev B

2007;75(16):165402.1-6.

[66] Paredes JI, Burghard M, Martínez-Alonso A, Tascón JMD. Graphitization of

carbon nanofibers: visualizing the structural evolution on the nanometer and atomic

scales by scanning tunneling microscopy. Appl Phys A 2005;80(4):675-82.

[67] Kuvshinov GG, Chukanov IS, Krutsky YL, Ochkov VV, Zaikovskii VI, Kuvshinov

DG. Changes in the properties of fibrous nanocarbons during high temperature heat

treatment. Carbon 2009;47(1):215-25.

[68] Lee S, Kim T-R, Ogale AA, Kim M-S. Surface and structure modification of

carbon nanofibers. Synth Met 2007;157(16-17):644-50.

[69] Lee S, Da S-Y, Ogale AA, Kim M-S. Effect of heat treatment of carbon nanofibers

on polypropylene nanocomposites. J Phys Chem Solids 2008;69(5-6):1407-10.

Page 84: Graphitization thermal treatment of carbon nanofibers

83

[70] Weisenberger M, Martin-Gullon I, Vera-Agullo J, Varela-Rizo H, Merino C,

Andrews R, et al. The effect of graphitization temperature on the structure of

helical-ribbon carbon nanofibers. Carbon 2009;47(9):2211-8.

[71] Yoon S-H, Lim S, Hong S, Qiao W, Whitehurst DD, Mochida I, et al. A conceptual

model for the structure of catalytically grown carbon nano-fibers. Carbon

2005;43(9):1828-38.

[72] Moriguchi K, Munetoh S, Abe M, Yonemura M, Kamei K, Shintani A, et al. Nano-

tube-like surface structure in graphite particles and its formation mechanism: A role

in anodes of lithium-ion secondary batteries. J Appl Phys 2000;88(11):6369-77.

[73] Tamayo-Ariztondo J, Córdoba JM, Odén M, Molina-Aldareguia JM, Elizalde MR.

Effect of heat treatment of carbon nanofibres on electroless copper deposition.

Compos Sci Technol 2010;70(16):2269-75.

[74] Chan C, Crawford G, Gao Y, Hurt R, Jian K, Li H, et al. Liquid crystal engineering

of carbon nanofibers and nanotubes. Carbon 2005;43(12):2431-40.

[75] Jian KQ, Shim HS, Schwartzman A, Crawford GP, Hurt RH. Orthogonal carbon

nanofibers by template-mediated assembly of discotic mesophase pitch. Adv Mater

2003;15(2):164-7.

[76] Zheng G-B, Sano H, Uchiyama Y. New structure of carbon nanofibers after high-

temperature heat-treatment. Carbon 2003;41(4):853-6.

[77] Kim YA, Hayashi T, Fukai Y, Endo M, Yanagisawa T. Microstructural change of

cup-stacked carbon nanofiber by post-treatment. Mol Cryst Liq Crys

2002;387(1):[381]/157-[385]161.

[78] Rotkin SV, Gogotsi Y. Analysis of non-planar graphitic structures: from arched

edge planes of graphite crystals to nanotubes. Mat Res Innovat 2002;5(5):191-200.

[79] Tzeng S-S, Wang P-L, Wu T-Y, Chen K-S, Chyou S-D, Lee W-T, et al. Formation

of loops on the surface of carbon nanofibers synthesized by plasma-enhanced

Page 85: Graphitization thermal treatment of carbon nanofibers

84

chemical vapor deposition using an inductively coupled plasma reactor. J Mater

Res 2006;21(10):2440-3.

[80] Ci L, Zhu H, Wei B, Xu C, Liang J, Wu D. Graphitization behavior of carbon

nanofibers prepared by the floating catalyst method. Mater Lett 2000;43(5-6):291-

4.

[81] Shioyama H. The production of a sheath around a stacked-cup carbon nanofiber.

Carbon 2005;43(1):203-5.

[82] Ozkan T, Naraghi M, Chasiotis I. Mechanical properties of vapor grown carbon

nanofibers. Carbon 2010;48(1):239-44.

[83] Lawrence JG, Berhan LM, Nadarajah A. Structural transformation of vapor grown

carbon nanofibers studied by HRTEM. J Nanopart Res 2008;10(7):1155-67.

[84] Blank VD, Polyakov EV, Batov DV, Kulnitskiy BA, Bangert U, Gutiérrez-Sosa A,

et al. Formation of N- containing C-nanotubes and nanofibres by carbon resistive

heating under high nitrogen pressure. Diamond Relat Mater 2003;12(3-7):864-9.

[85] Yoon S-H, Park C-W, Yang H, Korai Y, Mochida I, Baker RTK, et al. Novel

carbon nanofibers of high graphitization as anodic materials for lithium ion

secondary batteries. Carbon 2004;42(1):21-32.

[86] Zhou J-H, Sui Z-J, Li P, Chen D, Dai Y-C, Yuan W-K. Structural characterization

of carbon nanofibers formed from different carbon-containing gases. Carbon

2006;44(15):3255-62.

[87] Radović LR, Walker Jr PL, Jenkins RG. Importance of carbon active sites in the

gasification of coal chars. Fuel 1983;62(7):849-56.

[88] Wei BQ, Vajtai R, Ajayan PM. Reliability and current carrying capacity of carbon

nanotubes. Appl Phys Lett 2001;79(8):1172-4.

[89] Kelly BT. Physics of graphite. London: Appl. Sci. Publ. Ltd.; 1981.

Page 86: Graphitization thermal treatment of carbon nanofibers

85

[90] Tibbetts GG, Doll GL, Gorkiewicz DW, Moleski JJ, Perry TA, Dasch CJ, et al.

Physical properties of vapor-grown carbon fibers. Carbon 1993;31(7):1039-47.

[91] Tibbetts GG, Gorkiewicz DW, Alig RL. A new reactor for growing carbon fibers

from liquid- and vapor-phase hydrocarbons. Carbon 1993;31(5):809-14.

[92] Tibbetts GG, Bernardo CA, Gorkiewicz DW, Alig RL. Role of sulfur in the

production of carbon fibers in the vapor phase. Carbon 1994;32(4):569-76.

[93] Finegan IC, Tibbetts GG. Electrical conductivity of vapor-grown carbon

fiber/thermoplastic composites. J Mater Res 2001;16(6):1668-74.

[94] Tibbetts GG, Finegan IC, Kwag C. Mechanical and electrical properties of vapor-

grown carbon fiber thermoplastic composites. Mol Cryst Liq Cryst

2002;387(1):129-33.

[95] Finegan IC, Tibbetts GG, Glasgow DG, Ting JM, Lake ML. Surface treatments for

improving the mechanical properties of carbon nanofiber/thermoplastic composites.

J Mater Sci 2003;38(16):3485-90.

[96] Lee S-H, Hahn J-R, Ku B-C, Kim J-K. Effect of carbon nanofiber structure on

crystallization kinetics of polypropylene/carbon nanofiber composites. Bull Korean

Chem Soc 2011;32(7):2369-76.

[97] Kuriger RJ, Alam MK, Anderson DP, Jacobsen RL. Processing and

characterization of aligned vapor grown carbon fiber reinforced polypropylene.

Composites Part A 2002;33(1):53-62.

[98] Lafdi K, Fox W, Matzek M, Yildiz E. Effect of carbon nanofiber heat treatment on

physical properties of polymeric nanocomposites—Part I. J Nanomater

2007;Article ID 52729:6 p.

[99] Memon MO, Lafdi K. Use of carbon nanostructures in transient spike power

applications. Intl J Therm Sci 2012;53:1-7.

Page 87: Graphitization thermal treatment of carbon nanofibers

86

[100] Xu J, Donohoe JP, Pittman Jr CU. Preparation, electrical and mechanical properties

of vapor grown carbon fiber (VGCF)/vinyl ester composites. Composites Part A

2004;35(6):693-701.

[101] Chen L, Ye H, Gogotsi Y. Synthesis of boron nitride coating on carbon nanotubes.

J Am Ceram Soc 2004;87(1):147-51.

[102] Kang J, Nash P, Li J, Shi C, Zhao N, Gu S. The effect of heat treatment on

mechanical properties of carbon nanofiber reinforced copper matrix composites. J

Mater Sci 2009;44(20):5602-8.

[103] Habazaki H, Kiriu M, Konno H. High rate capability of carbon nanofilaments with

platelet structure as anode materials for lithium ion batteries. Electrochem

Commun 2006;8(8):1275-9.

[104] Kurita N, Endo M. Molecular orbital calculations on electronic and Li-adsorption

properties of sulfur-, phosphorus- and silicon-substituted disordered carbons.

Carbon 2002;40(3):253-60.

[105] Kim C, Yang KS, Kojima M, Yoshida K, Kim YJ, Kim YA, et al. Fabrication of

electrospinning-derived carbon nanofiber webs for the anode material of lithium-

ion secondary batteries. Adv Funct Mater 2006;16(18):2393-7.

[106] Gan L, Du H, Li B, Kang F. Enhanced oxygen reduction performance of Pt

catalysts by nano-loops formed on the surface of carbon nanofiber support. Carbon

2008;46(15):2140-3.

[107] Gan L, Du H, Li B, Kang F. Surface-reconstructed graphite nanofibers as a support

for cathode catalysts of fuel cells. Chem Commun 2011;47(13):3900-2.

[108] Ko Y-J, Oh H-S, Kim H. Effect of heat-treatment temperature on carbon corrosion

in polymer electrolyte membrane fuel cells. J Power Sources 2010;195(9):2623-7.

[109] Chambers A, Park C, Baker RTK, Rodriguez NM. Hydrogen storage in graphite

nanofibers. J Phys Chem B 1998;102(22):4253-6.

Page 88: Graphitization thermal treatment of carbon nanofibers

87

[110] Tibbetts GG, Meisner GP, Olk CH. Hydrogen storage capacity of carbon

nanotubes, filaments, and vapor-grown fibers. Carbon 2001;39(15):2291-301.

[111] Zhu H, Li X, Ci L, Xu C, Wu D, Mao Z. Hydrogen storage in heat-treated carbon

nanofibers prepared by the vertical floating catalyst method. Mater Chem Phys

2003;78(3):670-5.

[112] Naguib NN, Mueller YM, Bojczuk PM, Rossi MP, Katsikis PD, Gogotsi Y. Effect

of carbon nanofibre structure on the binding of antibodies. Nanotechnology

2005;16(4):567-71.

[113] Parrott EPJ, Zeitler JA, McGregor J, Oei S-P, Unalan HE, Tan S-C, et al.

Understanding the dielectric properties of heat-treated carbon nanofibers at

terahertz frequencies: a new perspective on the catalytic activity of structured

carbonaceous materials. J Phys Chem C 2009;113(24):10554-9.

[114] Wang M-X, Huang Z-H, Shen K, Kang F, Liang K. Catalytically oxidation of NO

into NO2 at room temperature by graphitized porous nanofibers. Catal Today

2012; 201:109-14.

Page 89: Graphitization thermal treatment of carbon nanofibers

88

FIGURE CAPTIONS

Fig. 1 - Depiction and/or TEM images of the different accepted structures for carbon

nanofilaments: (a) SWCNT and (b) MWCNT [9]; (c) platelet, (d) herringbone and (e)

ribbon or tubular CNFs [18]; (f) stacked-cup CNF [19]; (g) cone-helix CNF [9]; and (h)

thickened stacked-cup CNF [19].

Fig. 2 - Raman spectra of CNFs: (a) without B and (b) with B. From bottom to top, results

are shown for the as-grown and heat-treated at 1900, 2200 and 2500 ºC CNFs [55].

Fig. 3 - XRD patterns of as-produced (600 ºC) and heat-treated (1000-2800 ºC) platelet

CNFs [56].

Fig. 4 - XRD patterns of as-produced (NF01-NiCuTi) and graphitized (NF01-

NiCuTi/2400–2800) CNFs [57].

Fig. 5 - XRD parameters (a) and first-order Raman spectra (b) of the as-produced (CNF-5)

and heat-treated (CNF-5/1800-2800) CNFs [60].

Fig. 6 - XRD profiles of the as-grown and heat-treated CNFs. Top right inset:

deconvolution of the (002) peak for the as-grown CNFs [66].

Fig. 7 - Raman spectra of as-prepared (PR, MJ), CVD-deposited (PRCVD, MJCVD) and

heat-treated (PRHT, MJHT) CNFs [68].

Fig. 8 Full scale XRD scans of as-received and heat-treated (1500-2800 ºC) CNFs [70].

Page 90: Graphitization thermal treatment of carbon nanofibers

89

Fig. 9 - TEM images of platelet CNFs (a) as-prepared and (b) graphitized at 2800 ºC [51].

Fig. 10 - STM images of platelet CNFs (a) as-prepared and (b) graphitized showing the

presence of carbon nano-rod units [51].

Fig. 11 - TEM image of nano-rod units in the graphitized CNFs [51].

Fig. 12 - Three dimensional models of nano-rods and platelet CNFs [51].

Fig. 13 - STM (a, b) and TEM (c) images of graphitized tubular CNFs [71].

Fig. 14 - Hypothetical model of single nano-rod and the relationship of rod- and plate-type

units [71].

Fig. 15 - TEM images of platelet CNFs (a) as-received (b) heat-treated [73].

Fig. 16 - TEM images of platelet CNFs: (a) as-prepared and (b) annealed at 2500 ºC [74].

Fig. 17 - TEM images of CNFs: (a) as-produced (with SAED pattern) and (b, c, d) heat-

treated at 1000 ºC, 1500 ºC and 2800 ºC [56].

Fig. 18 - (a) General morphology of as-produced herringbone CNFs built by embedded

cones, (b) the surface of a CNF and (c) individual CNF with a diameter of ~ 15 nm. (d)

General morphology of heat-treated herringbone CNFs, (e) the structure of a fiber with a

Page 91: Graphitization thermal treatment of carbon nanofibers

90

diameter of ~ 70 nm and (f) an image of locked edges of graphene layers on the fiber

surface [67].

Fig. 19 - TEM images of (a) uncoated CNF after annealing at 3000 ºC and (b) coated CNF

after annealing at 3000 ºC [19].

Fig. 20 - TEM images of CNFs (a) grown by thermal CVD at 600 ºC and (b) annealed at

2400 ºC [79].

Fig. 21 - TEM images of CNFs (a) a wall of as-grown and (b) annealed. IN: inner-layer;

IT: inter-layer; O: outer-layer [80].

Fig. 22 - TEM images of CNFs (a) as-grown and (b) heat-treated at 2900 ºC [64].

Fig. 23 - TEM images of the double layer structure of hollow-core PR CNFs (a-b) as-

prepared and (c-d) heat-treated [68].

Fig. 24 - TEM images of the double layer structure of solid-core MJ CNFs (a-b) as-

prepared and (c-d) heat-treated [68].

Fig. 25 - SEM images of the fracture of commercial PR-24 CNFs (a) as-fabricated and (b)

heat-treated [82].

Fig. 26 - Variation of the specific surface area (BET) of platelet CNFs with HT

temperature. [56].

Page 92: Graphitization thermal treatment of carbon nanofibers

91

Fig. 27 - Pore size distributions of as-produced and heat-treated (1700-2600 ºC)

herringbone CNFs [67].

Fig. 28 - Variation of specific surface area and micropore area of stacked-cup CNFs with

the HT temperature (established from the nitrogen absorption at 77 K) [63].

Fig. 29 - TPO profiles of as-produced (CNF01-NiCuTi) and graphitized (CNF01-

NiCuTi/2400-2800) CNFs [57].

Fig. 30 - TEM images of BN-coated PR-PS CNFs (a) as-grown and (b) heat-treated (3000

ºC) [101].

Fig. 31 - The variation of discharge capacity when graphitized submicronVGCFs are used

as the anode material in the range 0 to 1.5 V with a current density of 0.2 mAcm-2

[35].

Fig. 32 - Extended galvanostatic cycling of the graphitized CNFs and of the SG

(commercial graphite employed as anode in lithium-ion batteries) [59].