In situ fabrication of graphene from a copper–carbon nanoneedle and its electrical properties

Abstract

Herein, we present a direct observation of the formation of graphene from a single copper–carbon nanoneedle (Cu–CNN) during the measurement of current–voltage (I–V) and direct heating via in situ transmission electron microscopy (TEM). Significant structural transformation of Cu–CNN was observed with an applied potential in a two probe system. Under a high current flow between 4.9 μA to 49.0 μA, the Cu nanoparticles melted and evaporated due to Joule heating. The amorphous carbon began crystallizing and transformed into sp² hybridized hollow graphitic carbon, which was catalyzed by the dispersed Cu nanoparticles. The temperature generated during the current flow was estimated to be 1073 K, as revealed by an in situ TEM heating experiment. The graphene nanoneedle formed exhibited a high current density of 10⁶ A cm⁻², which is comparable to Cu in normal interconnect applications. Thus, the graphene nanoneedle formed will be promising for future alternative interconnect materials.

---

Introduction

Graphene, a sp² hybridized carbon with a honeycomb lattice structure, has been in the spotlight in the area of nanomaterials due to its outstanding mechanical, electrical and chemical properties.^1–4 These unique properties have promoted interest in the use of graphene to replace existing materials in a plethora of applications, including interconnect materials.^5–9 In order to achieve these applications, there have been many efforts to understand the fundamental background of graphene formation using various catalyst metals.^7–10 In our previous work, we demonstrated the graphitization process in solid phase reactions using in situ transmission electron microscopy (TEM) for metal nanoparticle and Cu film-coated amorphous carbon nanofibers (CNFs).^11–14 Depending on the catalytic activity of the metal, during the migration or evaporation of metal nanoparticles, amorphous carbon transforms into graphene, partially graphitized carbon structures and multi-wall carbon nanotubes (CNT) during electron current flow for Cu-coated CNFs, Ag- and Au-incorporated CNFs, and Fe-incorporated CNFs, respectively.^11–14

For a thorough understanding of the structural transformation process, the next step is the elucidation of the temperatures at which migration or the evaporation of metal nanoparticles and graphitization of amorphous carbon occur. For this purpose, in situ heating experiments in TEM would be helpful. In the previous work, however, due to the limitation of the heating temperature (up to ∼1173 K) in TEM, we could not determine these temperatures. Very recently, we noticed that this type of experiment would be possible for copper–carbon nanoneedles (Cu–CNN) due to the higher catalytic property in graphene formation for Cu than that for Ag and Au, as well as the decrease in evaporation temperature for the nanosized Cu particles in Cu–CNN. Herein, we tackle this subject by investigating the Cu and C atom interactions at the nanoscale to synthesize graphene nanoneedles by in situ current–voltage (I–V) measurement in TEM, together with the determination of the graphitization temperature by the TEM heating holder. The suitability of graphene nanoneedle synthesis from Cu–CNN for interconnect applications will also be discussed in this paper.

Experimental

For the preparation of Cu–CNN, we used commercially available Cu foils with the dimensions of 5 mm × 25 mm × 20 μm and a Kaufmann-type ion gun (Iontech. Inc. Ltd, model 3-1500-100FC). The Cu foil was mounted on a sample stage and a graphite plate, the carbon atom supplier was placed perpendicular to the edge of the Cu foil (Fig. 1a). The edges of the Cu foils were irradiated with argon ions (Ar⁺) at 45° from normal to the surface for 60 min at room temperature. The diameter and ion beam energy employed for this experiment were 6 cm and 1 keV, respectively. The basal and working pressures were 1.5 × 10⁻⁵ and 2.0 × 10⁻² Pa, respectively. The growth mechanism of ion-induced CNF has been explained elsewhere in detail.¹⁵ The surface morphology of the sample was observed carefully using a scanning electron microscope (SEM, Jeol JEM-5600). Fig. 1b shows that the Cu–CNN grew on the tips of conical protrusions. It should be noted that only one Cu–CNN formed at each conical protrusion. For the in situ TEM experiment, we used a TEM sample holder (JEOL; EM-Z02154T) with a tungsten (W) nanoprobe as the anode controlled by a nano-manipulator. The Cu–CNN sample was directly mounted as the cathode on the stationary stage of the TEM holder. The TEM holder was equipped with electrical biasing equipment which allowed us to observe the morphological transformation of the Cu–CNN sample. For in situ TEM heating, a heating sample holder was used. Both TEM analyses were performed on a JEOL JEM-2010 with a vacuum chamber pressure less than 2.5 × 10⁻⁵ Pa and acceleration voltage of 200 kV. TEM images in video mode were continuously recorded using an image recording application (the TEM images taken from the video clip are referred to as video clip TEM afterwards). While manipulating the bias voltage, the structural behavior of the Cu–CNN sample was observed.

Fig. 1 (a) Schematic diagram of the experimental set up for Cu–CNN fabrication. (b) SEM image of Cu–CNN at the Cu foil edge. (c) Schematic diagram of the in situ I–V measurement experimental set up in TEM.

Results and discussion

Fig. 2a–d show the characterization of a single Cu–CNN. The Cu–CNN (cathode) was brought into contact with the W nanoprobe (anode) prior to I–V measurement. The Cu–CNN was approximately 800 nm in length and 40 nm in diameter (Fig. 2a). The high-magnification image of the Cu–CNN in Fig. 2b reveals the dispersion of fine Cu nanoparticles with a crystallite size of around 15 nm in the amorphous carbon matrix. The selected area electron diffraction (SAED) pattern and X-ray energy dispersive spectrum (EDX) of the Cu–CNN are shown in Fig. 2c and d, respectively. SAED shows the polycrystalline ring pattern of Cu, which indicates that the Cu–CNN is mainly composed of a mixture of amorphous carbon and randomly oriented Cu crystallites of Cu (111), Cu (200) and Cu (220). This was confirmed by the EDX spectrum in Fig. 2d, which indicates that the Cu–CNN is mainly composed of 20% C and 80% Cu.

Fig. 2 (a) TEM image of the initial Cu–CNN used for I–V measurement. (b) High magnification TEM image of the Cu–CNN. The inset of (b) shows a high magnification TEM image of the Cu nanoparticle. (c) SAED pattern and (d) EDX spectrum of the Cu–CNN taken in the region abbreviated as A.

The electric property of the Cu–CNN was measured by applying a bias voltage between the anode W probe and cathode Cu–CNN in TEM. A bias voltage was applied up to 1.5 V with incremental steps of 7 mV s⁻¹ while observing the structural change of the Cu–CNN. Fig. 3a shows the I–V characteristic of the Cu–CNN. It was observed that there was no significant current flow through the Cu–CNN, only around 4.9 μA (at 182 s), at an applied voltage up to 1.34 V. Electrical transport in amorphous carbon is conventionally attributed to the hopping mechanism where tunneling transitions between localized states occur and the energy difference between the initial and final states is bridged by the electron-phonon scattering process.¹⁶ The estimated resistance just before the sudden increase in current flow is 0.3 MΩ and this value is similar to the lower value reported for amorphous carbon.¹⁶ Applying more than 1.34 V, there was a sharp rise in current flow (49.0 μA) with a significant drop in resistance from 0.3 MΩ to 226 Ω.

Fig. 3 (a) I–V curve of the Cu–CNN with a high bias voltage in the two probe system (0–1.5 V). (b) Video clip TEM images of the Cu–CNN with a change in applied potential from 0.0–1.5 V, same as that of the I–V characteristics. TEM images present the structural transformation of Cu–CNN with a gradual increase in applied potential. Formation of a graphene nanoneedle across the cathode and anode is obtained with a length of around 800 nm and width of 40 nm.

Fig. 3b shows a series of video clip TEM images of the Cu–CNN during I–V measurement. The video clip TEM image just before 185 s was very similar to the initial stage (0 s) and a dramatic structural change was observed at 185 s instantaneously, after an abrupt current increase (182 s), as shown in Fig. 3b. The current just before and after the abrupt current increase was about 4.9 μA and 49.0 μA, respectively, at an applied voltage of about 1.34 V. Hence, the Joule heat generated before the abrupt current increase would be about 10 times smaller than that after the abrupt current increase, because the Joule heat is proportional to current × voltage. Thus, the temperature around the apex area of the Cu–CNN would be also about 1/10 before the abrupt current increase, which would be too low to form a graphitized structure around the whole apex area. The weak contact between the W nanoprobe and the Cu–CNN would be broken by the tunneling effect at a high local bias voltage at the contact point of the Cu–CNN. Thus, a large current flow would occur abruptly. This large current would induce enough Joule heat to form a graphitized structure.

It should be noted that the Cu particles evaporated abruptly around the apex area of the Cu–CNN to leave a graphene nanoneedle structure after the large current flowed for 185 seconds (more than 1.34 V), as shown in Fig. 3b. The Cu evaporation started from the apex area as a result of the contact resistance between the Cu–CNN and W probe as well as the fact that the apex area had a smaller diameter compared to the base area. The Cu evaporation started from the apex area then continued to the base of the Cu–CNN. With a further increase in the applied voltage, the graphene nanoneedle structure grew with the evaporation of Cu, as shown in Fig. 3b (189–198 s). Finally, a graphene nanoneedle with the length of around 800 nm and width of 40 nm was achieved by in situ TEM. The observed growth process of the graphene nanoneedle structure with the induced Joule heating effect is significant to understand the Cu interaction with C atoms in solid phase reactions.

Fig. 4a shows a TEM image the graphene nanoneedle after complete evaporation of the Cu nanoparticles. Similar to the video clip TEM image in Fig. 3b (198 s), as shown in Fig. 4a, it is clear that the Cu particles disappeared from the Cu–CNN structure and left only a graphene structure. The inset in Fig. 4a shows the SAED analysis of the graphene nanoneedle, which showed a ring pattern, thus indicating a very fine polycrystalline structure. This SAED result clarifies the absence of Cu particles and the graphitization of the Cu–CNN to a graphene nanoneedle. To further confirm this, higher magnification TEM observations were made. Fig. 4b shows the TEM studies of the sheet like structure of the as-synthesized graphene nanoneedle. The carbon atoms created sp² hybridized structure catalyzed by Cu nanoparticles to form a sheet like structure rather than CNTs. An inter-planer spacing of about 0.35 nm is estimated for the graphene layers, which corresponds to the graphite (0002) spacing.

Fig. 4 TEM image (a) after complete growth of the graphene nanoneedle and (b) sheet like structure of the synthesized graphene nanoneedle, in contrast to the hollow tube like structure. The carbon atoms created an sp² hybridized structure catalysed by Cu nanoparticles to form a sheet like structure rather than CNTs.

After the TEM observation, we applied a voltage again. Fig. 5 shows a comparison of the current properties between the Cu–CNN and graphene nanoneedle formed. The maximum current flow for the applied bias of 1 V for the Cu–CNN is far lower compared to the graphene nanoneedle, which was 0.4 μA and 27.0 μA, respectively. The high resistance of amorphous carbon in the Cu–CNN significantly affects the flow of current. Whereas, for graphene it is well known that the sp² graphitic structure has very good current conductivity.¹⁷ The linear nature of the I–V curve suggests good ohmic contact between the W electrode and graphene nanoneedle, which allows the electrical current to reach up to 27 μA under the small bias voltage of 1 V. The resistance in the initial Cu–CNN was very high, 2.2 MΩ, while the graphene nanoneedle exhibited a low resistance, 37 kΩ. This implies the potential application of the graphene nanoneedle synthesized by Cu–CNN as an interconnect in nanoelectronic circuits.

Fig. 5 I–V curves of Cu–CNN and graphene nanoneedle with an applied voltage of 1 V.

The current carrying capacity of the graphene nanoneedle is a key parameter that would determine whether the graphene nanoneedle could be used in interconnect applications. Here, the current carrying capacity and electrical breakdown behavior of the graphene nanoneedle were tested by applying a bias voltage of 2.0 V. Fig. 6a shows that the current flow increases until it reaches the maximum current of 0.15 mA at 1.73 V before the sudden and sharp decrease of the current to 0 A. As can be seen in Fig. 6b, the graphene nanoneedle broke at the middle part with the maximum current density of 0.46 × 10⁶ A cm⁻². This value is lower than the electromigration threshold for conventional metallization lines in microelectronic circuits.¹⁶ Therefore the obtained value, which is comparable with Cu interconnects used in previous research, will open a new route in the application of the graphene nanoneedle synthesized from Cu–CNN in interconnect applications.¹⁸ Fig. 6c shows a high magnification image of the broken and disconnected part of graphene nanoneedle. It is proven that the graphene nanoneedle consists of 3 average layers of graphitic layers. The breaking of the graphene nanoneedle is due to the significant amount of heat generated at the low dimensional metal–metal contact, which affects the stability of crystalline based materials.¹⁹

Fig. 6 (a) I–V characteristic with a higher applied potential across the anode and cathode. (b) TEM image of the broken graphene nanoneedle in the middle part after the growth of around 800 nm in length. (c) Higher resolution TEM image of the broken and disconnected graphene sheets.

To investigate in detail the graphitization of amorphous C and evaporation of Cu, a constant low bias voltage was applied at 0.5 V for 10 minutes for another Cu–CNN. Fig. 7a shows the TEM image of the Cu–CNN before the current flow, in which the Cu nanoparticle size and grain boundary can be seen clearly. Fig. 7b shows the Cu–CNN after current flow. It can be seen that the front part evaporates and leaves a graphitic hole structure. The high magnification TEM image in Fig. 7c and d reveal that the Cu particles in the Cu–CNN evaporated, leaving holes with shapes similar to those of the Cu particles. Furthermore, whole part of the amorphous carbon around the Cu particles was transformed to a graphitic structure because of the Joule heating induced by the electron current flow. Careful observation of the carbon around the remaining Cu nanoparticle reveals that the graphitic carbon starts to form before the Cu evaporated (Fig. 7e). It is worth noting that all the C around the Cu nanoparticle transformed into a graphitic layer with an average of 3 layers. This is in contrast with the results published by Y. Yaakob et al. for Ag-incorporated CNF, which showed that only amorphous carbon around the Ag transforms into graphitic layers while the other regions remain in the amorphous state. This can be explained by the difference in the rate of evaporation between Cu and Ag. H. A. Jones et al. calculated the rate of metal evaporation using Langmuir and Mackay's data and showed that at the same given temperature, the rate of Ag evaporation is 50 times higher than Cu.²⁰ Thus, carbon will have more time to dissolve properly in Cu rather than Ag. It should be also noted that the carbon solubility in Cu is higher than Ag which helps to make graphitization complete.^21,22

Fig. 7 TEM images of the Cu–CNN (a) before and (b) after the constant current flow of 0.5 V for 10 minutes. Higher resolution TEM images of Cu–CNN (c) before and (d) after the constant current flow of 0.5 V for 10 minutes. (e) TEM image of the Cu–CNN after the constant current flow of 0.5 V with some Cu nanoparticles remaining in the graphitic structure.

To investigate the temperature that evaporated the Cu nanoparticles in the Cu–CNN structure, in situ heating of the Cu–CNN in a TEM without current flow was performed using a heating sample holder. Fig. 8a shows the Cu–CNN used for in situ heating in the TEM. The Cu–CNN was approximately 800 nm in length and 80 nm in diameter. While heating from room temperature to 973 K, no any significant changes occurred in the structure of Cu–CNN. However, when the Cu–CNN was heated to 1023 K, the evaporation of Cu nanoparticles started to begin, leaving a graphitic hole structure (Fig. 8b).

Fig. 8 TEM images of the Cu–CNN (a) before in situ TEM heating experiment, (b) after reaching the heating temperature of 1073 K, (c) at the heating temperature of 1073 K for 30 minutes and (d) at high magnification after the heating temperature of 1073 K for 30 minutes.

Fig. 8c shows the TEM image of the Cu–CNN after 30 minutes heating at the temperature 1023 K. It is clearly seen that most of the Cu nanoparticles evaporated to leave a graphene nanoneedle structure. The highly graphitic structure of the graphene nanoneedle formed is shown in Fig. 8d, which is similar to the results from the above-described I–V experiment (Fig. 4). Thus, it is concluded that the during I–V measurement, the temperature rise due to Joule heating is around 1023 K. Severin et al. reported that the evaporation of Cu occurs in the temperature range of 1160–1511 K under high vacuum conditions (pressure < 10⁻⁵ Pa).²³ Thus, it is confirmed that the Cu nanoparticles evaporated at a temperature below its normal evaporation temperature in TEM. It should also be highlighted here that the size of the Cu nanoparticle has a significant effect on its melting point. Cu nanoparticles have a lower melting point compare to bulk Cu. This is due to an increase in the self-diffusion coefficient as a consequence of the well-established size effect.^24,25 Thus, it is indisputable that Cu nanoparticles evaporate at a temperature of around 1000 K.

Conclusions

In summary, we have revealed the direct observation of the formation mechanism of a graphene nanoneedle structure during the I–V measurement for a Cu–CNN under a high current between 4.9 μA to 49.0 μA. During the I–V process, the amorphous carbon structure was transformed into a hollow graphitic structure, which was catalyzed by dispersed Cu nanoparticles. The Cu assisted graphitization process occurred prior to the evaporation of the metal nanoparticles. We were able to grow the graphene nanoneedle to around 800 nm in length across the cathode and anode in situ TEM. The graphitization process and Cu evaporation occur at the temperature 1073 K, which is lower than the evaporation temperature of bulk Cu. The observed graphene formation at the nanoscale by the in situ TEM process can be significant to understand the interaction between C atoms and Cu. This opens a new route to the application of graphene nanoneedle synthesized from Cu–CNN in interconnect applications.

References

1. A. K. Geim and K. S. Novoselov, Nat. Mater., 2007, 6, 183.
2. K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva and A. A. Firsov, Science, 2004, 306, 666.
3. C. Lee, X. Wei, J. Kysar and J. Hone, Science, 2008, 321, 385.
4. A. A. Balandin, S. Ghosh, W. Bao, I. Calizo, D. Teweldebrhan, F. Miao and C. N. Lau, Nano Lett., 2008, 8, 902.
5. Y. M. Lin, C. Dimitrakopoulos, K. A. Jenkins, D. B. Farmer, H. Y. Chiu, A. Grill and Ph. Avouris, Science, 2010, 327, 662.
6. M. Antonio and M. Giovanni, Appl. Sci., 2014, 4, 305.
7. D. Prezzi, ACS Nano, 2014, 8, 5765.
8. A. A. Puretzky, D. B. Geohegan, X. Fan and S. J. Pennycook, Appl. Phys. Lett., 2000, 76, 182.
9. Z. Xu, X. D. Bai, E. G. Wang and Z. L. Wang, Appl. Phys. Lett., 2005, 87, 163106.
10. R. Rao, D. Liptak, T. Cherukuri, B. I. Yakobson and B. Maruyama, Nat. Mater., 2012, 11, 213.
11. Y. Yaakob, M. Z. Mohd Yusop, C. Takahashi, M. S. Rosmi, G. Kalita and M. Tanemura, RSC Adv., 2015, 5, 5647.
12. C. Takahashi, Y. Yaakob, M. Z. Mohd Yusop, G. Kalita and M. Tanemura, Carbon, 2014, 75, 277.
13. M. S. Rosmi, M. Z. Mohd Yusop, G. Kalita, Y. Yaakob and M. Tanemura, Sci. Rep., 2014, 4, 7563.
14. M. Z. Mohd Yusop, P. Ghosh, Y. Yaakob, G. Kalita, M. Sasase, Y. Hayashi and M. Tanemura, ACS Nano, 2012, 6, 9567.
15. Y. Yaakob, M. Z. Yusop, C. Takahashi, G. Kalita, P. Ghosh and M. Tanemura, Jpn. J. Appl. Phys., 2013, 52, 11NL01.
16. C. H. Jin, J. Y. Wang, Q. Chen and L. M. Peng, J. Phys. Chem. B, 2006, 110, 5423.
17. K. Hideo, M. Yuhki, I. Satoshi, O. Yutaka, Y. Ichiro and T. Seiji, Nanoscale, 2009, 1, 344.
18. B. Bourlon, D. C. Glattli, B. Placais, J. M. Berroir, C. Miko, L. Forro and A. Bachtold, Phys. Rev. Lett., 2004, 92, 026804.
19. T. B. Y. Song, C. H. C. Chen, Y. Yang, B. Brion, D. Hsin-Sheng, L. Gang, T. King-Ning, H. Yu and Y. Yang, ACS Nano, 2014, 8(3), 2804.
20. H. A. Jones, L. Irving and M. J. Mackay, Phys. Rev., 1927, 30, 201.
21. R. B. McLellan, Scr. Metall., 1969, 3, 389.
22. S. Dorfman, D. Fuks and M. Suery, J. Mater. Sci., 1999, 34, 77.
23. V. I. Severin, A. V. Tseplyaeva, Y. A. Priselkov and L. P. Ryabtseva, High Temp., 1986, 24, 363.
24. K. Andreas, F. P. Schaper and H. Haoqing, J. Mater. Res., 2005, 20, 1844.
25. J. Mu, Z. W. Zhu, H. F. Zhang, H. M. Fu and A. M. Wang, J. Appl. Phys., 2012, 111, 043515.

---