A thin membrane or a net held on a grid with a certain pitch has been widely used for the characterization of nanomaterials by transmission electron microscopy (TEM) (the suspended part of grid is referred to as a mesh hereafter in this paper). In particular, suspended graphene and other two-dimensional materials are eagerly utilized for novel two-dimensional materials characterization15-17, even though a complicated and time-consuming process needs to be followed for commercial graphene grid processing18. Furthermore, a suspended graphene array on an appropriate pitch support can be used as a harmonic oscillator and pressure sensor19,20. In this study, we utilize amorphous carbon (a-C) membranes on TEM grids as a carbon source and plasma-excited metals generated by the pulsed arc plasma deposition (APD) technique as a catalyst for the formation of transmissible plasma-evolved graphene (TPEG). Physical vapour deposited catalysts favoured the formation of not only thin film coatings21,22 but also catalyst particles; therefore, they have been used for the preparation of catalyst particles for carbon nanotube (CNT) growth23,24, including the synthesis of single-walled CNTs (SWCNTs)25. Pulsed APD is a kind of physical vapour deposition in which a target element gets ionized with an applied pulsed bias voltage (Supplementary Fig. 1) and is then sputtered onto the substrate. For the formation of the TPEG herein, Co, Pt, Ir, W, and synchronous W and Co catalyst plasmas excited by pulsed APD with different bias voltages were irradiated onto suspended a-C for different sets of experiments (Fig. 1a, Supplementary Table 1). Pt, Co, and Ir are widely employed catalysts for the synthesis of SWCNTs due to their high catalytic reactivity to C, where sp2-C is formed from carbon feedstock23,24. Catalytically active W nanoparticles composited with C and Co (WC and Co6W7) have been utilized26,27 for the chiral-selective growth of SWCNTs. To probe the morphology of bimetallic CoW and its carbide catalysts employed for SWCNT growth, pulsed APD was used to deposit catalysts directly onto a TEM grid28. When employing a catalyst plasma with bias voltages that are too intense or a grid pitch size that is too large (Extended Data Fig. 10), we found significant damage to the a-C mesh, which is the bottleneck for the TPEG technique, and a systematic investigation is needed. An a-C film held onto a PMMA support was further utilized to investigate that this process occurred at a significantly low temperature, and the effect of the soft support was investigated and compared to that for Cu or Mo (illustrated in Fig. 1b).
Raman spectroscopy is an important tool to identify graphene quality29. The Raman spectrum of TPEG produced with Protocol P1_Co using a 50 mm pitch TEM grid showed that the Co plasma formed during pulsed APD with a bias voltage less than 100 V was not enough to form TPEG because the G band peak was not observed (Fig. 2a). The 2D peak induced by the second-order two-phonon process started to appear at 100 V in the Raman spectrum (Protocol P2_Co). These results indicate that nominal graphitization occurs under a bias voltage of 100 V or more. Upon increasing the bias voltage (140-250 V), a pronounced 2D peak appeared (Protocol P3-P6_Co), symbolizing the good quality of graphitic carbon29 (Fig. 2a, Extended Data Fig. 4a). The smallest full width at half maximum (FWHM) of the Raman 2D and G peaks and the largest intensity ratio of the 2D band to the G band (I2D/IG) for 170 V (Protocol P4_Co) confirmed the good quality of the TPEG at this condition (Fig. 2b). There was significant damage in the central part of the mesh for a Co plasma with a voltage greater than 170 V (Extended Data Fig. 10 (5,5), (5,6)), and qualitative analysis of the Raman spectra taken at the remaining part showed a comparatively low-quality TPEG for high bias voltages (Protocol P5-P6_Co). In addition, when employing a TEM grid with a pitch larger than 75 µm, more than 90 percent of the a-C mesh was destroyed even with a Co plasma with a voltage of 170 V (Extended Data Fig. 2b (Protocol P4_Co)). TEM analysis performed for the Co deposited at 170 V onto a 50 µm a-C mesh on a Mo grid showed a high-quality graphitic carbon with almost no dark contrast areas that correspond to Co particles (Fig. 2c). Selected area electron diffraction (SAED) for this region showed the formation of polycrystalline graphite with favourable {110} and {112} orientations30 (Fig. 2d). Bright-field scanning TEM (STEM) observations confirmed that they were composed of 3 to 12 layers of graphene (Fig. 2e). In very few places, nanoparticles with sizes larger than 10 nm were observed (Fig. 2f, Extended Data Fig. 1a) for the employed condition. Energy-dispersive X-ray spectroscopy (EDS) spectra and EDS mapping obtained for such an area (Extended Data Fig. 1b-f) showed that those particles were composed of Co covered with Mo particles. The reason for the formation of such particles is not clear, but the sputtered Co might have recombined with residual Mo on the a-C membrane because Mo comprised the a-C membrane support in the TEM grid. High-resolution TEM (HRTEM) analysis showed that the interlayer distance in the folded region of the 3.6 nm thick graphitic carbon was 0.34 nm (Fig. 2f, g), which confirmed the formation of good quality graphene. The fast Fourier transformation (FFT) image (inset to Fig. 2f) taken within the blue rectangle in Fig. 2f shows that the stacking orientation was 10130.
To probe the effect of the plasma exposure time, Protocol P2’_Co was employed, where the bias voltage was the same as that in Protocol P2_Co, but with 10 pulses. Optical microscopy (OM) is the easiest tool to identify graphene on SiO2/Si1, so OM observation was performed for the TPEG produced with Protocol P2’_Co by transferring it onto SiO2/Si after etching the Cu grid. Intensive care needed to be exercised while transferring the a-C membrane (thinner than 30 nm) onto the SiO2/Si. The dark bluish circular area in the OM image (Fig. 2h) represents the mesh part of the grid, and the distinct colour variation indicates the formation of TPEG in the mesh part. During the transfer onto SiO2/Si, crumpling of the TPEG was avoided for Raman mapping and atomic force microscopy (AFM) analysis. Raman mapping was performed within the red dotted rectangular area in Extended Data Fig. 3a. In the mesh part of the grid, the peak intensities for the G, 2D and D bands and their ratios (I2D/G and ID/G) were uniform (Extended Data Fig. 3b-e). The uniform colour corresponding to the mesh part of the grid in the Raman G peak mapping (Fig. 2i) indicates the homogeneous graphitic nature of the TPEG, and the 2D peak intensity was pronounced in the central part (Extended Data Fig. 3b). This indicates that with increasing Co plasma exposure time, TPEG was formed, even at 100 V. Similarly, Raman mapping was performed on TPEG produced with Protocol P3_Co, where the bias voltage was 140 V (3 pulses of Co onto 75 mm pitch Cu grid). After transferring the TPEG onto SiO2/Si, Raman mapping was performed and covered more than 80 percent of the mesh (red dotted rectangular area of Extended Data Fig. 4b). For the mesh part of the grid, the G, 2D and D band peak intensities and their ratios (I2D/G and ID/G) were uniform (Extended Data Fig. 4c-g), similar to that for the TPEG produced by Protocol P2’_Co (Extended Data Fig. 3b-e). A pronounced 2D peak was seen in the combined spectrum obtained from the red rectangular part of the mapped image (Extended Data Fig. 4h). These Raman mapping results indicate the formation of homogeneous graphene layers. In addition, X-ray photoelectron spectroscopy (XPS) of TPEG produced with Protocol P2’_Co showed a narrowed FWHM for the C1s peak in the mesh-dominant area (~70 percent of the suspended part), further confirming the transformation of a-C into sp2-C (Extended Data Fig. 2c, d). It should be noted here that the intensity of the Co 2p3/2 peak was significantly reduced in the mesh-dominant part (details are discussed later). AFM analysis shows that an almost 3-fold reduction in thickness occurred after the transformation of a-C into TPEG around the mesh part of the grid (Fig. 2j and Extended Data Fig. 2e-g), although a nominal change in the thickness of the a-C membrane was also observed for the part supported onto the grid (Extended Data Fig. 2e). These results indicate that the a-C membrane in the mesh part of the grid was efficiently transformed into graphene by catalytically active Co plasma.
In addition, Co, Pt, Ir and W were used as plasma catalysts to form TPEG; Co has a high C solubility, while elements, such as Pt, Ir, and W, have a lower solubility than Co31,32. In general, the desorption energy from the target material by plasma excitation depends on the element, which varies the intensity of plasmas generated even at an identical bias voltage. Therefore, the thickness of each element during the deposition was monitored by a quartz crystal oscillator. The obtained results showed that the deposition rate decreased with the melting point; the deposition rate was the lowest for W and highest for Co (Extended Data Fig. 5a). To compare the catalytic activity, identical sets of experiments were run with these elements to form TPEG. The Raman spectra (Extended Data Fig. 5b, d) show a comparable quality for the TPEG with Ir and Pt. Qualitative analysis of the Raman spectra showed that the quality of the TPEG produced with Pt and Ir plasmas with a voltage of 170 V was the best herein (Extended Data Fig. 5c, e), which is consistent with the result for TPEG produced with Co. In contrast, the W plasma did not produce TPEG even with a bias voltage of 180 V, although the W deposition rate was almost equal to that of Ir at 170 V (Extended Data Fig. 5a, 5f). Upon increasing the bias voltage to 210 V and above, both the G band and D band appeared in the Raman spectrum, indicating the low catalytic activity of W. Upon increasing the plasma exposure time, both Pt and Ir produced good-quality TPEG even at 100 V, and the crystallinity was comparable to that for TPEG produced with Co at the optimal bias voltage (Extended Data Fig. 6a, b). OM images (Extended Data Fig. 10) show a significant difference in the morphology (wrinkles and waviness) of the a-C mesh in the central part and towards the edges of pitch with an increase in the bias voltage (70 to 170 V for Co, Ir, and Pt; and 100 to 250 V for W), providing an overview of TPEG formation. The a-C mesh starts to be destroyed from the central part due to an extreme increase in the stress with a bias voltage that is too high for Ir, Pt and W, as seen for the TPEG produced from Co (Extended Data Fig. 10 (5,5), (5,6)), and an identical trend can be seen for larger pitch grids.
It was very interesting to investigate TPEG formation using two elements with drastic differences in their catalytic activity, and the use of a synchronous Co and W plasma was the best choice because they show remarkable differences in catalytic activity for graphitization and C solubility31,32. The bias voltage for Co was set to 70 V (instrumental lower limitation) and that for W was adjusted to make it equivalent to the voltage for other tested protocols (Supplementary Table 1). It is worth noting here that TPEG produced with Protocol P2_Co+W (140 V W and 70 V Co, each for 3 pulses) (Fig. 3a) was almost identical to that produced by a 140 V Co plasma only. This clearly indicates that the W plasma improved the catalytic activity of Co and vice versa. OM analysis further showed that after exposure to a 100 V W and 70 V Co plasma, the a-C mesh on a 100 µm pitched grid was not destroyed, unlike for TPEG produced with Protocol P4_Co on a 100 µm pitch grid (Extended Data Fig. 2b, 10) with a bias voltage of 170 V for Co. Upon increasing the bias voltage for W, a slight improvement in the TPEG occurred, and the W plasma at 180 V and Co plasma at 70 V were found to be best herein (Fig. 3a, b). Similarly, by increasing the plasma exposure time to 10 pulses for synchronous Co and W (Protocol P1’_Co+W), TPEG formation occurred even at low bias voltages (Extended Data Fig. 6c). TEM analysis of TPEG produced with Protocol P1’_Co+W showed graphitic features similar to those for TPEG produced with a 170 V Co plasma (Protocol P4_Co), but they were decorated with very dense nanoparticles (Fig. 3c). This observation was also consistent with TPEG produced with Protocol P4_Co+W that used a high bias voltage (240 V W and 70 V Co) (Extended Data Fig. 7a, b). HRTEM images taken within the blue and red rectangles in Fig. 3c show the formation of 3- to 12-layered graphene with an interlayer distance of 0.34 nm (Fig. 3c-f). Even after the transfer onto SiO2/Si, both the OM image and Raman mapping performed on the mesh part of the grid show TPEG with a uniform quality (Fig. 3h, i, Extended Data Fig. 8a-d). The average Raman spectra for TPEG obtained from the three different mapping regions show a pronounced 2D peak at the central part of the mesh and no 2D peak on the edge and outside part of the mesh (Fig. 3j). This clearly confirms that only the mesh part of the grid was transformed onto TPEG, and a-C supported onto the grid remained unchanged.
Furthermore, SWCNT nuclei that grew from different nanoparticles were observed in the TEM image (Fig. 3c) and are indicated by white arrows. Green and violet arrows point to areas containing other carbon structures that decorated the TPEG. This observation suggests that TPEG is not only appropriate for suspended graphene production but is also appropriate for the formation of three-dimensional (3D) hybrid structures composed of 0D, 1D and two-dimensional nanomaterials, which has potential applications ranging from energy storage to green energy production. SAED analysis showed the formation of polycrystalline graphite with preferred {110} and {112} orientations and WC with preferred 110 and 211 orientations30,33 (Fig. 3g). The interlayer distance of the lattice pattern (0.189 nm) measured for most of the particles revealed that those particles were WC with a 110 orientation33. There is a possibility that Co- and W-based particles could have formed Co6W7, Co6W7C and WC particles, which would have acted as catalysts for the growth of SWCNTs26-28, but EDS and electron energy loss spectroscopy (EELS) characterization of the nanoparticles showed no presence of Co with a significant amount of W (Extended Data Fig. 7d, e), confirming the formation of WC. The absence of Co in the mesh part of the grid after an exposure to a W plasma with 70 V Co is analogous to the absence of Co for TPEG produced with Co only (TEM and XPS observation of Protocol P2’_Co), and we suspected that a significant amount of Co was transmitted through the mesh in both cases.
To investigate the possible transmission of Co and TPEG formation for TPEG produced with Protocol P1’_Co+W, XPS analysis was performed on the transferred sample with a focus on the detection area (diameter~0.2 mm), similar to TPEG produced with Protocol P2’_Co. The XPS spectra obtained for the mesh-dominant area and grid peripheral area showed an almost identically intense peak of W 4f7/2 (Extended Data Fig. 8e, f). However, compared with that for the peripheral part, the mesh-dominant part showed a significantly less intense Co 2p3/2 peak and more intense C 1s peak, which indicated that a significant amount of Co was transmitted through the mesh part of the grid, thereby transforming a-C into TPEG. This observation is identical to the XPS observation for TPEG produced with Protocol P2’_Co (Extended Data Fig. 9b, c).
To explore the plasma transmissibility in the TPEG, XPS analysis was performed by two separate sets of experiments. The TEM grids were held onto SiO2/Si by maintaining a 0.5 mm gap during pulsed APD of Co, and they were exposed to a synchronous Co and W plasma with Protocol P2’_Co (100 V Co) and Protocol P1’_Co+W (100 V W and 70 V Co). OM analysis of the SiO2/Si surface just below the TEM grid kept area showed traces of circular spots with an optical contrast identical to that outside the edge of the grid (Fig. 4a, Extended Data Fig. 9a). XPS analysis with a focused detection area (diameter~0.2 mm) was performed for the edge and the mesh parts of the Cu grid and the mesh-traced parts on SiO2/Si of both samples. Relatively low intensity peaks in the Co 2p3/2 spectra from the mesh-traced part on SiO2/Si were observed for both experiment sets compared to those from other regions (Fig. 4b, Extended Data Fig. 9c). This indicates that relatively less Co plasma was recombined onto the SiO2/Si after being transmitted through the mesh. However, the almost flat peaks of W 4f7/2 and 4f5/2 for the spectra obtained from the mesh-traced part on SiO2/Si (Fig. 4b) indicate no transmission of W, which supports that WC nanoparticles were decorated on the TPEG obtained by identical sets of experiments. The C 1s peak in the spectra obtained from the mesh part of the grid decreased in width, supporting the successful transformation of a-C into TPEG in comparison to that for the other part.
The formation of only TPEG at the mesh part suggests that if a-C was supported on the materials softer than Cu or Mo, TPEG could be formed on them alike mesh on the TEM grid. To confirm this, 2-5 nm a-C deposited onto Cu by e-beam evaporation was transferred onto ~100 nm PMMA supported on SiO2/Si, where PMMA was regarded as a soft support (Fig. 4c). Then, TPEG was formed by Protocol P4’_Ir (170 V Ir for 10 pulses) and Protocol P4’’_Ir (170 V Ir for 5 pulses). OM and Raman analysis of TPEG produced with Protocol P4’_Ir and Protocol P4’’_Ir showed that the a-C film on PMMA was transformed into TPEG with a maximum domain size of ~20 µm (Fig. 4d-f, Supplementary Fig. 2). In the case of TPEG produced on PMMA, the intensity ratio of the D band to the G band was higher than that of most of the TPEG produced from suspended a-C. To investigate the origin of the large D peak in the Raman spectra, monolayer graphene was synthesized by CVD and then transferred onto SiO2/Si by a technique reported elsewhere34. After CVD graphene on SiO2/Si was annealed at 800°C under ultrahigh vacuum, it was exposed to an Ir plasma produced during pulsed APD. Raman analysis showed that the FWHM of the G band as well as the D peak intensity of the monolayered graphene significantly increased (Supplementary Fig. 3) after these treatments. This suggests that either lattice vacancies or non-crystalline regions formed on the graphene by plasma sputtering35,36. We would like to note here that the TPEG formation process was not due to either Joule heating37 or electromigration38, but it was realized during the transformation of a-C into graphene by plasma-excited elements using pulsed APD.
In summary, by tuning the catalytic activity of a pulsed APD plasma, plasma-activated Co, Pt and Ir can transmit through a suspended a-C membrane or that on a soft support. During transmission through the membrane, not only did sp2-C bonding form but also its crystallinity drastically improved, and graphitization occurred. On choosing a plasma combination that pairs a catalytically active and less active plasma, a new 3D hybrid structure composed of fullerene (0D), CNTs (1D) and graphene (two-dimensional) materials was also realized, which could be applicable for energy generation purposes. This TPEG formation process is both time- and energy-efficient because less than 10 sec of exposure to a pulsed APD plasma produced TPEG, and each experiment lasted no longer than 30 min. The sustainability of PMMA as a supporting layer during the TPEG formation process further confirmed that the processing temperature was far below the melting point of PMMA (160°C), which is close to room temperature. This process is an avenue for the transformation of various amorphous materials into their crystalline structure, similar to the transformation of a-C into sp2-C.