Introduction

As a semiconductor with a distinctive crystal structure, graphite has a wide range of useful properties, making it an interesting material in technology [1–3]. The degree of graphitization plays an important role in the thermal conductivity, electrical conductivity, and mechanical properties. Catalytic graphitization with numerous catalysts has been proposed to accelerate the graphitization process and improve the degree of graphitization at relatively low temperature [3–14].

Recently, alloys or compounds have been investigated as catalysts owing to their synergistic catalytic effect on the degree of graphitization of carbon materials [15–26]. The majority of studies have focused on the synergistic catalytic effect of catalysts containing one metallic additive and one inorganic additive, although some studies have investigated catalysts with two metallic additives. Tzeng et al. reported that the degree of graphitization is enhanced in electroless Ni–P-coated polyacrylonitrile-based carbon fibers heat treated at 1400 °C compared with uncoated fibers heat treated at 2400 °C [15, 16]. Garcia et al. found that Ni–Si has a synergistic catalytic effect on the structural properties of carbon nanofibers produced by catalytic decomposition of methane [17]. Moreover, the microstructures of carbon fibers were improved after being coated with Ti–B, or Ni–B, or doped with Mo–B [18–20]. Carbon materials doped with Ti–Si have also attracted attention because of the notable enhancement of the ordered graphitic structure [3, 21]. Synergistic catalysis corresponded to the combination type of catalysts which contain one metallic additive and one inorganic additive mentioned above. It was considered that formation–decomposition of carbide occurred in these carbon materials along with formation of a new phase by reaction of the metallic catalyst with another additive. However, the mechanism of the combination type of catalysts which contain two metallic additives follows the rule that both dissolution–precipitation and formation–decomposition of carbide simultaneously occur. Significant effects have been attained in carbon materials using catalysts of Fe–Ni, Fe–Co, and Ti–Ni alloys/compounds [7, 22–26].

In our previous research, we found that a small amount of a light rare earth (RE) element significantly improves the properties of carbon material [27, 28]. For instance, 1 wt% praseodymium oxide significantly catalyzes graphitization of the graphite anode. However, there are few reports of synergistic catalytic graphitization of carbon materials using light RE elements and other additives. In this study, in combination with light RE element (La, Ce or Pr) additives, Ti and Ni were selected to represent two categories whose mechanisms have been proposed to follow different routes [3, 29]. Boron was also considered as an additive with light RE element because of its special homogeneous effect on catalytic graphitization of carbon materials [30]. The influence of the heat-treatment temperature and the effect of combinatorial catalysts on the microstructure and mechanical properties of graphite were investigated.

Materials and methods

Preparation

Coal-based needle coke and coal tar pitch were used as an aggregate and a binder, respectively. The additives were added to the mixture (coke:pitch = 4:1, wt/wt). The typical procedure to produce artificial graphite was then carried out, including kneading, hot molding, baking and impregnation, and finally graphitization. The sources and properties of the raw materials and the detailed experimental procedures are described elsewhere [28]. The samples are labeled as G–X1Y1X2Y2/Z, where G is graphite, X1 and X2 are the additives, Y1 and Y2 are the amounts of the additives (wt%), and Z is the temperature at which the sample was treated.

Characterization

The effect of combinatorial catalysts on the microstructures of graphite was investigated by X-ray diffraction (XRD, Riguku D/max-2550 diffractometer with Cu Kα1 radiation, λ = 1.5406 Å). The interlayer spacing (d 002), degree of graphitization (g), and mean crystalline size along c-axis (L c) of the samples were examined with silicon as the internal standard, and then calculated using the appropriate equations (Bragg, Mering-Maire and Alexander equations) [31, 32].

Raman measurements of surface sections of the samples were performed with a Thermo DXR spectrograph at an excitation wavelength of 445 nm. Ten Raman spectra were collected for each sample because the lattice vibration is strongly influence by charged impurities [33, 34]. The D, G, and 2D peaks were fitted with Lorentz functions. The degree of graphitization was determined by the intensity ratio of the D to G peaks (I D /I G). The mean crystalline size along the a-axis (L a) was calculated using the Cançado equation [35].

The morphologies of the samples were observed by transmission electron microscopy (TEM, JEOL/JEM 2100).

The electrical resistivity of graphite at room temperature was measured by the DC four-probe method.

Three-point bending flexural tests were performed on a universal testing machine (3300 Instron) for samples with dimensions of 10 × 10 × 40 mm and a span-length of 30 mm. For each system, three samples were tested at a loading speed of 2 mm min−1 to obtain an average value. The bending strength was calculated by (1):

$$ \sigma = 1.5{\text{PL}}\left( {{\text{BH}}^{2} } \right)^{ - 1}, $$
(1)

where P is the maximum flexural load, L is the span-length, and B and H are the width and the height of samples, respectively.

Results

The crystal parameters of the graphitized samples calculated from the XRD patterns and Raman spectra are listed in Table 1, including the d 002, g, I D/I G, L a, and L c values.

Table 1 Crystal parameters of the graphitized samples determined from the XRD patterns and Raman spectra
Full size table

Compared with the G/2800 sample, all of the graphite samples prepared with the Ti additive have lower d 002 values, larger crystalline sizes (both L a and L c), and higher degrees of graphitization. The L c value shows a similar trend to the d 002 value. Compared with the G–Ti4.0/2800 sample, the samples with the same combinatorial catalyst content (3 wt% Ti and 1 wt% light RE element) have lower d 002 values and higher degrees of graphitization. The g values are higher at the higher heat-treatment temperature in all the cases. The effect of the heat-treatment temperature on the L c value is similar to that on the g value, while the effect on the L a value is more prominent. The effect of the heat-treatment temperature on the L a value in samples prepared with the Ti additive is the most significant, followed by that in the samples prepared with Ni and B additives. For the heat-treatment temperature of 2400 °C, all of the samples prepared with combinatorial catalysts have lower d 002 values, higher degrees of graphitization, and larger L c values than the corresponding samples prepared with a single additive. However, the L a values are similar for samples prepared with a single catalyst and the combinatorial catalysts.

Compared with the G/2800 sample, the as-prepared samples with Ti and RE catalysts have similar d 002 values but larger L c values when they were heat treated at 2400 °C. Similar results are observed in the cases of G–Ni3.0RE1.0/2400 and G–B3.0RE1.0/2400. Moreover, the L a values of the samples with Ni and RE catalysts heat treated at 2400 °C are larger than that of the samples without catalyst heat treated at 2800 °C. However, the L a values of the G–Ti3.0RE1.0/2400 and G–B3.0RE1.0/2400 samples are all lower than that of the G/2800 sample.

A TEM image showing the lamellae morphology of the G–Ti4.0/2800 sample is shown in Fig. 1. All of the layers are parallel to the lamellar plane, which is clearly shown in the enlarged image of the area indicated by the white rectangle. Highly ordered stacked layers with a d 002 value of 0.3360 nm are observed, indicating that a high degree of graphitization was achieved.

Figure 1
figure 1

HRTEM image of the graphitized sample with 4.0 wt% Ti after heat treatment at 2800 °C. The inset shows the d 002 interlayer spacing of 0.3360 nm

Full size image

To investigate the synergistic catalytic effect of the combination catalysts, wide-angle XRD patterns were firstly recorded to confirm the structure and phases of the graphitized samples, as shown in Fig. 2. The sharp peaks of hexagonal graphite are easily observed. With increasing heat-treatment temperature, the (100) peak becomes sharper and more symmetrical, as expected. Along with the peaks of graphite, diffraction peaks of TiC and RE carbide (REC2) are observed in Fig. 2a. For the G–Ni–RE samples (Fig. 2b), signals corresponding to the nickel RE carbide (RENiC2) phase are detected in the samples heat treated at 2400 °C. In the G–B–RE samples heat treated at 2400 °C (Fig. 2c), the peak at around 32° is attributed to REC2. However, REC2 is not detected for the heat-treatment temperature of 2800 °C, as shown in Fig. 2f.

Figure 2
figure 2

XRD patterns of graphite prepared with additives at a–c 2400 °C and d–f 2800 °C

Full size image

Figure 3 shows the representative Raman spectra. When the heat-treatment temperature was increased from 2400 to 2800 °C, the D and G peaks in the Raman spectrum of the G–Ti4.0 sample became sharper and the I D /I G ratio increased, indicating that the degree of graphitization increased. This tendency is also observed for the G–Ni4.0 and G–B4.0 samples. The positions of the peaks in samples prepared with the co-catalysts are similar to the corresponding samples prepared with the single catalysts.

Figure 3
figure 3

Raman spectra of graphite prepared with single catalysts at a 2400 °C and b 2800 °C

Full size image

The as-prepared samples heat treated at 2800 °C were processed by the machining method to obtain specimens suitable for measuring the electrical resistivity and mechanical properties. The bulk density and electrical resistivity of the as-prepared samples are listed in Table 2. All of the graphitized samples prepared with catalysts show lower electrical resistivity than the sample prepared without a catalyst.

Table 2 Bulk densities and electrical resistivity of graphite samples prepared with and without catalysts at 2800 °C
Full size table

The effects of Ti and RE elements doped at the same amount on the mechanical properties of the samples were compared, as shown in Fig. 4a. Figure 4b shows a comparison of the effect of different catalysts on the bending stress of the samples.

Figure 4
figure 4

a Representative bending stress–displacement curves of G–Ti–RE samples heat treated at 2800 °C. b Comparison of the effect of different additives on the bending stress of the samples. M represents the metal

Full size image

Discussion

Determination of the amount of catalysts added into the carbon materials

As the results show, Ti enhances the crystallinity of carbon materials at 2800 °C, which is consistent with other publications [18, 21]. However, the catalytic effect of Ti did not increase with increasing Ti content, which is supported by the weakly fluctuant values of g, I D /I G, and L a. This is consistent with the results reported by He et al., who proposed that formation of titanium carbide (TiC) is the reason for a similar phenomenon in carbon fibers doped with Ti [18]. Consequently, 3 wt% Ti was added to the as-prepared samples with 1 wt% of a light RE element as a second additive to investigate their synergistic catalytic effect on graphite. The reason for choosing 1 wt% of the light RE element was based on the finding that 1 wt% of praseodymium oxide significantly influenced the graphitization process and then improved the electrical properties of the graphite anode in our previous publication [28]. In addition, other catalysts (Ni and B) with the same content were also prepared with the same process to compare the synergistic catalytic effects.

Graphitized samples with additives heat treated at different temperatures

The highest degree of graphitization (with a d 002 value of 0.3358 nm) was achieved in the sample with 3 wt% Ti–1 wt% Pr. This indicates that the combination of Ti and a light RE element has an effect on the microstructure of graphite.

However, this phenomenon was not observed for the G–Ni–RE and G–B–RE samples. A single Ni or B additive had a more significant effect on d 002 than the corresponding combinatorial catalysts, which can be attributed to the L c value, as mentioned above. The reason for this is related to their different catalytic mechanisms, which will be discussed latter.

Generally, the effect of temperature on the graphitization of carbon materials is significant, and catalysts are used to accelerate the graphitization process at relatively low temperature [1, 3]. The degree of structural order of samples increased as the heat-treatment temperature of the precursor increases, finally leading to graphitic materials. In previous publications [15–20], the carbon fibers, whose raw materials were hard carbon, such as methane, furan resin, and polyacrylonitrile, were catalyzed by the combinatorial catalysts which contain one metallic additive and one inorganic additive at about 2400–2800 °C for 2 h. Since hard carbon was easier to accept the assistance of catalysts than soft carbon [36], the graphite, whose raw materials were coke and mesophase pitch, doped with Ti and Si were heat treated at 2400–2600 °C and under 8–10 MPa hot-pressing simultaneously [21]. In this study, the carbon precursor were needle coke and coal tar pitch, which were both the soft carbon raw materials, and the graphite was obtained from these carbon precursor by adding light rare earth elements and other typical catalysts and heat treated at 2800 °C for 0.5 h. This indicated that the light rare earth elements exhibited the more significant catalytic ability than other catalysts and may assist the other catalysts to catalyze the soft carbon materials.

The difference between the L a value and other crystalline parameters can be explained as follows: the L a value calculated by Raman spectroscopy is affected by the technique based on propagation of phonons. This propagation is only stopped by the boundaries of the layers (or by a “physical end” or by defects bordering the layers) [1]. Vacancy defects formed after the departure of boron can be detected, and the L a values are smaller for B than the other additives at both temperatures.

Comparison of the heat-treatment results in Table 1 leads to the conclusion that for a given temperature, more-ordered structures are obtained for samples prepared with combinatorial catalysts than a single catalyst, which indicates a synergistic catalytic effect on the graphitization of carbon materials. For instance, the G–Ti4.0/2400, G–Ti3.0Pr1.0/2400, G–B4.0/2400, and G–B3.0Pr1.0/2400 samples have L c values of ~78, ~89, ~80, and ~86 nm, respectively. The sample prepared with a B–Ce co-catalyst has a d 002 value of 0.3363 nm and an L c value of 86.4 nm after heat treatment at 2400 °C, suggesting a synergistic catalytic effect. For the samples prepared with Ti catalysts, the d 002 values are lower than that of the G–Ti4.0 sample at both temperatures, and the L c values are higher in the G–Ti–RE samples. However, this trend is only observed for the Ni and B catalysts at a heat-treatment temperature of 2400 °C.

According to the values of the crystal parameters, the catalysts played an important role during the graphitization of carbon materials at relatively low temperature. The combinational catalysts synergistically accelerate the graphitization process, and thus the more-ordered graphitic structure can be obtained at lower temperature.

Synergistic catalytic effect during the graphitization process

For the Ti catalysts, as well as the peaks from the graphite phases, there are extra peaks corresponding to TiC shown in Fig. 2, which is consistent with the previous studies [2, 9]. The chemical stability of TiC means that it is difficult to decompose. Therefore, a decrease in the intensity of the TiC phase with increasing temperature indicates enhancement of graphitization.

Because REC2 peaks are present in the XRD patterns of the samples heat treated at 2400 °C, formation of graphite at the expense of carbide decomposition seems to be a plausible mechanism for catalytic graphitization of these carbon materials. The intensities of the REC2 peaks decrease when the heat-treatment temperature of the samples increases, and the peaks are not present in the XRD patterns of the samples prepared at 2800 °C. However, it cannot be concluded that REC2 decomposes to the metal and graphite with increasing temperature. Compared with G/2800, the residual TiC and REC2 at 2400 °C results in an increase of grain boundaries and defects, and subsequently leads to a decrease of the L a value calculated from the Raman spectra.

Nickel carbide was not detected, although it probably formed. It decomposes at temperatures above 500 °C [24, 29]. Ni accelerates graphitization followed a dissolution–precipitation mechanism, which means the disordered carbon dissolves into the liquid Ni phase and subsequently precipitated as graphite [6, 12, 19]. Similar to REC2, RENiC2 peaks are not present in the XRD patterns of the samples prepared at 2800 °C. Combined with the results for the G–Ti–RE samples, these results demonstrate that the catalytic behavior of the metals in the combinatorial catalysts involves formation of carbide by the reaction of the metal catalyst with carbon and its subsequent decomposition to graphite. This explains why the L a values of the G–Ni–RE samples heat treated at 2400 °C are larger than that of the G/2800 sample (i.e., because of evaporation of Ni and little residual carbide).

No boron or boron carbide phases were detected, which is similar to nickel carbide. In previous publications, boron was considered to accelerate graphitization by creating local strain, which is the driving force when boron attacks and replaces specific free carbon atoms [30, 38]. Boron was not observed to react with the RE elements to form a new phase. Boron would result in defects, irrespective of its departure or substitution, leading to a decrease of the L a value.

The results indicate that in the G–Ti–RE system, TiC and REC2 are the primary catalysts that accelerate the graphitization process of carbon materials with high-temperature heat treatment. TiC is difficult to decompose once it is generated, while REC2 exhibits different physical properties where formation and decomposition as well as melting occur. In the G–Ni–RE and G–B–RE systems, departure of Ni and B from graphite can entrain REC2 to some degree, resulting in a decrease in the intensity of REC2 at relatively high temperatures. REC2 accelerates the graphitization process by not only its decomposition but also assisting dissolution–precipitation by accelerating the speed of oversaturation.

Although the XRD data are consistent with the Raman results, the carbides are not detected in the Raman spectrum [39]. All of the positions of the peaks correspond to graphite, including the sharp G peak and the weak D peak. The spectrum also shows a second-order peak of the D band at ~2700 cm−1, which is sometimes called the 2D or G′ band because it appears in more graphitic carbons materials. Dresselhaus et al. [40] suggested that the G′ band is a Raman-allowed mode for sp 2 carbons. Based on the concept of propagation of phonons in the Raman spectrum, the effect of carbides may be contained in the D and G bands owing to obstruction of the catalysts, which increase the edges of the layers or acts as heteroatom to connect continuous layers even if they are more or less wrinkled [37].

In the samples prepared with RE catalysts, the parameters calculated from the Raman spectra can be referenced to changes in the intensity (Table 1). Compared with G/2800, the G–Ni–RE samples have more graphitic order when they are heat treated at 2400 °C, which can also be detected by Raman spectroscopy. However, similar results were not obtained for the G–Ti–RE and G–B–RE systems. Based on the above experimental observations and analysis, light RE elements and other catalysts have a synergetic effect on graphitization of carbon materials. Synergistic catalytic mechanisms are proposed along with the corresponding binary phase diagrams.

It can be inferred that Ti and light RE element act as co-catalysts in the graphitization process of carbon materials based on coke and pitch by formation of TiC and REC2, respectively. All of the decomposition temperatures of REC2 are much lower than that of TiC (~2250 and 2776 °C, respectively), which means that light RE elements would accelerate the graphitization progress of carbon materials at relatively low temperature. In our previous publication [28], it was proposed that the light RE elements began to exhibit the catalytic effects when the carbon materials heat treated at 1600 °C. Therefore, the light RE elements and Ti catalyzed the carbon materials at different temperature ranges, respectively. Adding light RE elements widen the range of the catalytic temperature, improving the degree of graphitization. The catalytic mechanism was supported by the fact that all of the samples for G–Ti–RE systems have lower d 002 values, higher degrees of graphitization, and larger L c values than the corresponding samples prepared with a single additive both at different temperatures. The increase of the L a values for G–Ti–RE systems after heat treated at 2400 °C confirmed that the light RE elements assisted the growth of the graphitic order structures at relatively low temperature.

The light RE elements in the G–Ni–RE systems had a similar effect on development of the graphitic structure at 2400 °C. For the Ni systems, RENiC2 is in the liquid state where carbon can dissolve to form an oversaturated solution. When the carbon materials heat treated above 1327 °C, the molten catalyst Ni was not only assist the diffusion of light RE elements but also reacted with part of light RE elements to create new phase RENiC2. Then the decomposition of solid phase RENiC2 occurred at higher heat-treatment temperature to release both light RE elements and graphite. This light RE elements can further catalyzed other less-ordered carbon. However, unlike RENiC2, which further decomposed to graphite, the role of REC2 in the Ni systems is similar to that in the Ti systems. However, the effect of it appears to be less significant than that in the Ti systems, as listed in Table 1. The main catalytic mechanism of Ni for carbon materials is preferential dissolution of disordered carbon and precipitation of graphite [18]. Hence, in all of the G–Ni–RE samples, the catalytic process follows the sequence of dissolution–precipitation and simultaneously carbide formation and decomposition.

However, light RE elements and B do not have a synergetic catalytic effect because no carbides were generated. Therefore, the microstructures of the G–B–RE samples are not attributed to the synergistic effect of the catalysts.

According to the above analysis, the active metals preferentially react with disordered carbons at the boundaries of the graphitic sheets to form the corresponding carbide, and further decomposition would lead to graphite. The presence of the metal catalysts increased the crystalline sizes of the already existing graphitic layers. When added to carbon materials with other catalysts, RE elements would assist catalytic graphitization.

Effect of combinatorial catalysts on the electrical and mechanical properties of graphitized samples

The electrical and mechanical properties of the graphitized samples were investigated based on the analysis of the crystal parameters and the catalytic process.

As summarized in our previous study [28], the L c value is the most important factor for the electrical resistivity, followed by the bulk density and the degree of graphitization. The lowest electrical resistivity was obtained for the G–Ti3.0Pr1.0/2800 sample. Based on the microstructures and electrical properties of the samples, the light RE elements assist catalytic graphitization of samples with other additives, and the light RE elements and other catalysts have a synergistic effect.

Ti has a more significant enhancing effect on the mechanical properties than the RE elements. This was also found in Ref. [21], where the bending stress in graphite was significantly enhanced by doping with Ti. However, when Si was introduced as an additive in graphite doped with Ti, the value of the bending stress sharply decreased, demonstrating the different effects of the catalysts. RE elements showed the opposite effect in the G–Ni–RE and G–B–RE systems, especially in G–B–RE, as shown in Fig. 4(b).

The synergistic catalytic effect of Ti and RE elements was exhibited in the microstructures and electrical properties of graphite because of acceleration of catalytic graphitization within a wide temperature interval and the low electrical resistivity and high fracture strength of residual TiC and REC2 [21]. The mechanical properties of graphite improved using Ni–RE catalysts because of the synergistic catalytic effect of Ni and the RE element. The departure of Ni would be harmful to graphite but this is compensated for by the RE element, which is also observed for the B–RE catalysts.

Conclusions

Three light RE elements (La, Ce, and Pr) were combined with other catalysts (Ti, Ni, and B) and added to carbon materials as co-catalysts to investigate their synergistic catalytic effect. The effect of the combinatorial catalysts on the degree of graphitization in the carbon materials was significant. Compared with reference samples containing the same amount of the single catalyst, more-ordered graphite was obtained using the combinatorial catalysts at a heat-treatment temperature of 2400 °C. Moreover, the as-prepared samples obtained with combinatorial catalysts had similar d 002 values and larger L c values when heat treated at 2400 °C compared with the sample without an additive heat treated at 2800 °C. This indicates that the RE elements and other catalysts have a synergistic catalytic effect on carbon materials. For instance, the combination of Ti and Pr had a synergistic catalytic effect on the microstructure of graphite, and the lowest electrical resistivity of 5.4 μΩ m was obtained after heat treatment at 2800 °C. In the Ti–RE catalytic system, RE elements would accelerate graphitization of carbon materials by widening the range of the catalytic temperature. In the G–Ni–RE and G–B–RE systems, departure of the Ni and B can entrain REC2, and the mechanisms include dissolution–precipitation and carbide formation and decomposition. The RE element compensates for the harm caused by departure of Ni. This was also was observed for the B–RE catalysts.