1. Introduction

Graphene-based nanomaterials have attracted a lot of attention due to their high surface area to volume ratio, remarkable electrical, mechanical and optical properties and their potential applicability in many emerging nanodevices in fields ranging from opto-electronics to biosensing [1–4]. However, the existing horizontal graphene structures currently used in such devices are bound to the substrate, which both interferes with the properties of the basal plane and limits the number of active edges. The presence of active edges is particularly important for certain applications, e.g., super-capacitors, fuel cells and sensors. Hence, it is crucial to be able to fabricate graphene in the form of vertically standing nanosheets (VGS) which have substrate-contact-free active basal planes along with open active graphitic edges [5–8]. Being able to exert a high degree of control over the dimensions and structural properties of VGS is therefore, a critical step required for the effective integration of graphene-based components into advanced nanodevice platforms.

To date, several growth methods including on arc discharge, chemical vapor deposition (CVD) and electrochemical based techniques [9,10] have been used for the production of graphene-based nanomaterials. Low-temperature plasma-enhanced CVD (PECVD) [11] is among the most successful of these techniques as it demonstrates an improved controllability over the VGS structures at relatively low temperatures, an important factor for device integration. However, the existing mechanisms presented in the literature do not explain the critical role played by various plasma parameters in the precise control of the structure and properties of VGS during growth [12], in particular on low melting temperature substrates.

Here, we report on the deterministic control over the dimensions and structure of vertical graphene nanosheets on catalyst-free silicon and glass substrates using a simple low-temperature plasma-based process. This high degree of control was achieved by changing the nitrogen concentration in the gas mixture during the growth process. The Raman spectra of the VGS were then taken to investigate some of the effect of nitrogen concentration on the graphitic structure. The optical properties were investigated through UV-Vis spectrophotometry, with a particular emphasis on revealing the UV-filtration properties of the VGS. Finally, we demonstrated the suitability of VGS as a biosensing platform, by analyzing the interaction of Bovine Serum Albumin (BSA) protein with VGS through both Raman spectroscopy and UV-Vis spectrophotometry.

2. Experimental work

The catalyst-free growth of VGS was carried out in an inductively-coupled plasma chemical vapor deposition (ICP-CVD) system (13.56 MHz, RF power 1.0 kW max) [13]. Thermally oxidized (500 nm oxide layer) silicon wafers and glass (2 x 2 cm²) were used as substrates for the deposition. After loading the substrates into the reactor chamber, the chamber was pumped down to 4.8 x 10⁻⁴ Pa and a gas mixture of CH₄/Ar/H₂ at a mass ratio of 1:2:1 was fed into it. The plasma was generated at a chamber pressure and RF power of 4.0 Pa and 1000 W, respectively. No external substrate heating source was used for the deposition. However, the substrate temperature increased up to a maximum of 450°C due to the plasma-heating effect [11] during the 9 minutes of deposition. To study the effect of nitrogen on the growth of the graphene nanosheets, nitrogen gas was added at a mass concentration of 20%, 33% and 50% to the gas mixture of CH₄/Ar/H₂ during the growth process, while the other deposition conditions were maintained unchanged.

The morphologies and structural properties of the as-deposited VGS were investigated using scanning electron microscopy (SEM) and confocal micro-Raman spectroscopy (Renishaw inVia Raman spectroscopy; Ar⁺ laser excitation 514 nm; at room temperature excitation, 1% power, exposure time of 10 s). The optical properties of VGS-networks as-deposited on glass substrates were studied using UV-Vis spectrophotometry (Varian Cary 5000 UV-Vis spectrophotometer).

To investigate the bio-sensitivity of the VGS, a 2% w/v solution of BSA (Sigma-Aldrich) in Dulbecco’s phosphate buffer saline (PBS) was prepared and dropped on the VGS in the following amounts: 150 µL was dropped on all 4 VGS samples grown on Si (for Raman measurements); 150 µL was dropped on VGS samples grown on glass (for UV-Vis transmission measurements) with N₂ = 0 and 20%; 75 µL was dropped on VGS grown on glass with N₂ = 33 and 50% (due to the reduced adhesion of the VGS on glass with N₂ increase). All samples were left overnight to dry in open air. After the samples were completely dry, the interaction of the BSA with the VGS was studied using both Raman spectroscopy and UV-Vis spectrophotometry (using the same instruments mentioned previously).

3. Results

The schematic in Fig. 1(a) illustrates the plasma-based deposition of VGS. Figures 1(b) and 1(c) are photographs of Ar/H₂/CH₄ and Ar/H₂/CH₄/N₂ plasmas, respectively, taken during the VGS growth process, which produced maze-like VGS networks (similar to our previous reports [12]) on both the silicon and the glass substrates. The different plasma colors (i.e., whitish with a faint purple tinge in Fig. 1(b) – no nitrogen, whereas Fig. 1(c) shows a pinkish plasma – N₂ = 20%) suggests that different ions/radicals in different excited states and in different concentrations (which in turn produce optical emission at different wavelengths and with different intensity) are present in the plasma. This different plasma composition resulted in the growth of VGS of different lengths as well as number density (as shown in Figs. 1(d) and 1(e)).

 

Fig. 1 (a) Schematic shows the plasma process for growth of VGS. (b, c) pictures show the distinct color of Ar/H₂/CH₄ plasma without (b) and with 20% nitrogen (c) during the growth process. (d, e) SEM image shows that as-deposited VGS-networks in presence of nitrogen (e) are denser than the VGS-networks produced without nitrogen (d).

Download Full Size | PDF

Further SEM investigations revealed that the as-deposited VGS were uniform in number density, and were interconnected to each other to create uniform VGS-networks all over the substrate in a nitrogen free plasma during the deposition process (Fig. 2(a) ). The average length of these nanosheets was found to be ~810 nm. Upon addition of nitrogen into the gas mixture (from 20 to 50%) during the growth process, the average length of VGS decreased from 810 (no N₂) to 250 nm (50% N₂) and their number density increased from 20 to 36 μm⁻² (Figs. 2(b)-2(d)). We also observed that the average size of voids formed between the nanosheets decreased (as indicated in Figs. 2(a) and 2(b) by V1 and V2). Figures 2(e) and 2(f) present histograms of the changing number density and the average length of as-deposited VGS with different nitrogen concentrations.

 

Fig. 2 (a,b,c,d) Scanning electron micrographs show VGS as-deposited at N₂ = 0% (a), N₂ = 20% (b), N₂ = 33% (c) and N₂ = 50% (d). “V1” and “V2” indicate the voids as formed between the graphene nanosheets (a,b). The circles show a domain of VGS-networks and the red arrows show the gap formation between the domains (c,d). (e,f) the histograms show the number density of VGS in 1 µm² (e) and the average length of VGS (f).

Download Full Size | PDF

It should also be noted that the VGS-networks remained uniform all over the substrate surface until the amount of nitrogen added to the gas mixture surpassed 20%. When a nitrogen concentration of greater than 20% was added, the interconnections between the nanosheets no longer remained uniform. For example, for N₂ = 33%, the VGS-networks are non-uniform over the substrate surface and, as shown by the encircled area in Fig. 2(c) small domains of VGS-networks began forming on the surface. A further increment in the nitrogen concentration (N₂ = 50%) during the growth process resulted in the formation of VGS-networks domains separated by increasingly larger gaps (as indicated in Figs. 2(c) and 2(d)), yet interconnected to each other, all over the substrate. However, the number density of graphene nanosheets in each domain is increased significantly. This reveals that the average diameter of the voids formed between the nanosheets [12] and the thickness of the nanosheets decreased with nitrogen concentration.

We further investigated the structural and optical properties of VGS using micro-Raman spectroscopy and UV-Vis spectrophotometry. There are four main Raman peaks, denoted as D, G, D’ and 2D peak, which are observed for all VGS samples [14]. The G-peak is activated due to the C-C stretching mode (E_2g) in the hexagonal lattice, whilst the D-peak is due to the presence of defects, in particular, the many graphitic edges of the graphene nanosheets. Therefore, the variation in defective states in the nanosheets or graphitic edges produces significant effects in the intensity ratio of D and G-peak, I_D/I_G, along with the full width half maximum (FWHM) of G and D peaks. The D’-peak is activated as a result of the symmetry breaking in the graphitic structure due to finite crystalline size, whereas the 2D-peak is a second order of the D-peak which does not require any defect for its activation. Thus the appearance of 2D-peak is considered to be a fingerprint of a crystalline graphitic structure. Figure 3(a) shows the normalized Raman spectra of as-grown VGS at different nitrogen concentrations. As the nitrogen concentration increases, the intensity of the G-peak (1578 cm⁻¹) decreases and the intensity of the D-peak (1350 cm⁻¹) increases progressively, along with an increased FWHM of G-peak. It is also noticed that the 2D-peak (2690 cm⁻¹) which appeared in the VGS that were deposited at N₂ = 0%, diminishes as the D’-peak (1615 cm⁻¹) increases for VGS deposited at higher nitrogen concentrations. This reveals that there is an increase in the number of edges in the VGS-networks as well as in the disorderness (i.e. a decrease in the amount of sp² carbons) between many VGS-networks. We also observed that the Si-peak (520 cm⁻¹) corresponding to the silicon substrate appeared for the VGS samples which were deposited at N₂ = 33 and 50%, which suggests that the coverage of VGS-networks is not uniform on the substrate.

 

Fig. 3 (a) Normalized Raman spectra of VGS deposited at N₂ = 0%, N₂ = 20%, N₂ = 33% and N₂ = 50% before the BSA treatment. (b) The UV-Vis transmission spectra of the VGS as-deposited at N₂ = 0%, N₂ = 20%, N₂ = 33% and N₂ = 50% before the BSA treatment. Note the sharp jump in transmission at 350 nm is due to a source changeover in the Cary 5000 UV-Vis spectrophotometer.

Download Full Size | PDF

UV-Vis transmission measurements were carried out for the VGS samples which were deposited on glass substrates. The transmittance spectra obtained from the VGS samples (Fig. 3(b)) suggest that the VGS-networks which were deposited at N₂ = 0% and 20%, have almost negligible (or zero) transmittance signal in the UV region (<400 nm). However, they allow the optical and NIR signal to transmit through. The transmittance in the optical region is 2% less than the VGS networks deposited at N₂ = 20%, however, it is 5% more than the VGS-networks deposited at N₂ = 33%. The opening of transmittance was further increased by 10% in the optical region when VGS-networks were deposited at N₂ = 50%, which also served to open the transmittance in the UV region (from 300 – 400 nm) to a maximum of 8%. Below 300 nm, the VGS networks proved to be completely opaque to UV. This reveals that simply by changing N₂ concentration, the transmittance characteristics of VGS-networks could be tuned for many applications.

The VGS-networks were also tested for suitability as a biosensing platform by analyzing their interaction with BSA via Raman spectroscopy and UV-Vis spectrophotometry. The Raman spectra of BSA treated VGS are shown in Fig. 4(a) . There is a dominant Raman peak at 2960 cm⁻¹ along with other peaks at 508, 751, 943, 1451 and 1657 cm⁻¹ which correspond to BSA protein [15]. These peaks are due to the presence of strong amide groups (І, III), aromatic amino acids groups, CH₂ and CH₃ scissoring modes and Raman active side chains in amino acid groups [15–17]. It should be noted that Raman activated peaks corresponding to graphene nanosheets (D, G and 2D peak) are almost suppressed for the VGS deposited at N₂ = 0%. However, these peaks could be observed for the VGS deposited at N₂ = 20, 33 and 50%. Thus, these results suggest that the BSA proteins have indeed attached to the edges of VGS-networks.

 

Fig. 4 (a,b) Raman spectrum after the BSA treatment which details is shown in (b). (c,d,e,f) The UV-Vis transmission spectra of the VGS as-deposited at N₂ = 0%, N₂ = 20%, N₂ = 33% and N₂ = 50%, before (curve 1) and after the BSA treatment (curve 2), the arrow mark shows the decreased transmittance. Please note the sharp jump in transmission at 350 nm is due to a source changeover in the Cary 5000 UV-Vis spectrophotometer

Download Full Size | PDF

The BSA protein attachment to VGS is further revealed from the transmittance measurements. Figures 4(b)-4(e) reveal that there is a change in transmission before and after the BSA treatment of VGS-networks. The highest change of 22.0% in transmittance in IR-region was observed for the VGS-networks deposited at N₂ = 50%, followed by 6.0, 2.5, 2.0% change for the VGS-networks deposited at N₂ = 33, 20, 0%, respectively. This confirms that there is a variable sensitivity of different VGS-networks for the analysis of BSA. However, it was also observed that there is an increase in the transmittance signal in the near UV-region (<400 nm) for the BSA treated VGS-networks which have almost negligible (or zero) transmittance before BSA treatment (Figs. 4(c)-4(e)).

4. Discussion

The main focus of this investigation was to study the effect of the nitrogen concentration on VGS growth. Previously, we fabricated VGS in the ICP-CVD system using Ar, CH₄ and H₂ gas mixtures only. Here, when nitrogen was introduced in the growth process, we observed a decrease in the individual nanosheets length and an increase in the VGS number density (Fig. 2). Moreover, from the Raman spectrum we confirmed that higher nitrogen gas concentrations lead to a thinning of the VGS layers. These observations reveal that the introduction of nitrogen gas produces nitrogen species in the plasma which leads to etching phenomena for the VGS.

There are two processes which compete with each other during the growth of VGS:

Raman spectroscopy revealed important structural information about the VGS. One of the significant points that was noticed was diminishing 2D peaks with respect to increased nitrogen concentrations. Further, the dispersion of G-peak became significantly larger with more N₂ which indicates increased disorder in the VGS-networks. Therefore, different BSA interactions with the VGS samples were also observed when analyzing Raman and UV-Vis spectra. The Raman peaks that originated from the VGS were diminished and the peaks corresponding to BSA dominated in the BSA-treated VGS which were grown at N₂ = 0%. This suggests that there was a strong interaction between the BSA and the highly-ordered graphitic nanosheets, which in turn resulted in the dense BSA coverage all over the vertical graphitic edges of VGS-networks. However, the interaction between the BSA and the nanosheets decreased when the graphitic disorderness increased in the VGS with increasing nitrogen concentration. Therefore, the BSA coverage on the VGS-networks also decreased and the Raman peaks originated from VGS again, appearing clearly for the VGS samples prepared at N₂ = 20, 33 and 50%.

In the UV-Vis transmission measurements, we have observed that the VGS samples prepared at N₂ = 20% have the lowest transmittance signal in the visible range. This is due to the highest number density of deposited VGS on the substrate whilst retaining uniformity of VGS-networks all over the substrate surface. However, the larger gap formation between the domains of highly packed VGS-networks due to increased nitrogen concentrations (for N₂ = 33 and 50%, Figs. 2(c) and 2(d)) leads towards a higher transmittance in the entire UV-Vis spectral region studied. This increase is particularly stronger in the 400-800 nm wavelength range. When the VGS samples are treated with the BSA, the voids between the graphene nanosheets and/or gaps formed between the domains of VGS-networks in the VGS samples prepared at N₂ = 20, 33 and 50% are filled by the BSA proteins, which resulted in a higher change (3, 7.5 and 22%, respectively) in the transmittance signal. This also suggests that the BSA interactions with the nanosheets are weak. The small change (of < 2%) in the transmittance signal from the VGS samples prepared at N₂ = 0% suggests that the BSA are mostly attached to the graphitic edges of the nanosheets rather than filling the voids formed between the nanosheets. Thus, the variable sensing properties of VGS for analysis of BSA protein were obtained using different nitrogen concentrations during the growth of the VGS.

5. Conclusion

We have demonstrated that variation of the nitrogen concentration in the plasma-assisted process is very effective for the deterministic control of the dimensions and structure of vertically oriented graphene nanosheets on catalyst-free silicon and glass substrates. The length and number density of the nanosheets could be controlled from 810 to 250 nm and from 20 to 36 μm⁻², respectively. The graphitic structure of VGS was also controlled, which resulted in their variable sensing properties, demonstrated using Bovine Serum Albumin protein solution. The VGS-networks were also found to act as UV-filters, where mostly optical signal is passed through the VGS. The amount of transmitted UV light (in the 200 to 400 nm spectral range) could be effectively controlled from 0 to 8% by a simple variation of nitrogen concentration in the gas mixture. Thus, by modifying the N₂ concentration, VGS-network based components can be tailor-made for various applications, with particular promise for optoelectronics, bio- and gas-sensors.