Epitaxial graphene gas sensors on SiC substrate with high sensitivity

Cui Yu; Qingbin Liu; Zezhao He; Xuedong Gao; Enxiu Wu; Jianchao Guo; Chuangjie Zhou; Zhihong Feng

doi:10.1088/1674-4926/41/3/032101

Abstract

2D material of graphene has inspired huge interest in fabricating of solid state gas sensors. In this work, epitaxial graphene, quasi-free-standing graphene, and CVD epitaxial graphene samples on SiC substrates are used to fabricate gas sensors. Defects are introduced into graphene using SF₆ plasma treatment to improve the performance of the gas sensors. The epitaxial graphene shows high sensitivity to NO₂ with response of 105.1% to 4 ppm NO₂ and detection limit of 1 ppb. The higher sensitivity of epitaxial graphene compared to quasi-free-standing graphene, and CVD epitaxial graphene was found to be related to the different doping types of the samples.

1. Introduction

Graphene is attractive for fabricating of solid state gas sensors due to its unique physical and chemical properties. It possesses large surface area per unit volume and zero charge carriers rest mass, and shows strong interaction with gas molecules due to physical or chemical adsorption^[1-4]. After exposure and adsorption of targeted gas molecules, the electrical conductivity of graphene changes because of a change in carrier concentration of graphene. The electrical conductivity of graphene increases with target chemical species act as donor, and decreases for that act as acceptor. This phenomenon can be measured easily by an electronic system.

Several graphene allotropes, such as exfoliated graphene flakes^{[1, 5, 6]}, CVD graphene^{[7, 8]} chemically reduced graphene oxide^{[9, 10]}, and epitaxial graphene (EG)^[11-14], have been successfully used as gas sensors for ammonia, nitrogen oxides, sulfur oxides, and so on. EG grown on single crystal SiC substrate typically shows promising electrical properties, such as good surface morphology. In particular, due to the semi-insulator property of the SiC substrate, graphene does not need transfer to another substrate for device fabrication. This can avoid the possible processing pollutants that influence the sensing performance, such as the polymethylmethacrylaat (PMMA) that is usually used in the transfer process. These make EG on SiC a promising platform for gas sensors and easy up-scaling. Lebedev et al. demonstrated that UHV-grown epitaxial graphene give high sensitivity (5 ppb) to NO₂^[14]. Novikov et al. used epitaxial graphene grown in Ar ambient to fabricate graphene gas sensor and NO₂ concentration of 0.2 ppb can be easily detected^[15]. Iezhokin et al. used EG and quasi-free-standing epitaxial graphene (QFSEG) samples for gas sensors. They found for NO₂ gas, the sensitivity of QFSEG shows a six-fold increase compared to EG^[13]. Samples show sensitivity better than 1 ppb. These results of graphene on SiC gas sensors are comparable to the current state-of-the-art solid state sensors^{[16, 17]}. The defect-engineer has been shown to be an important way to the sensitivity improvement of graphene gas sensors. Kumar et al. used Ar plasma treatment and graphene gas sensors showed improved response^[18]. Chung et al. used UV-ozone treatment to introduce defects in graphene^[19]. Lee et al. used reactive ion etching (RIE) to fabricate defective graphene and found that graphene gas sensors show ultrahigh sensitivity with 33% improvement and 614% improvement in NO₂ and NH₃ sensing, respectively^[7].

In this work, EG, QFSEG, and CVD grown epitaxial graphene (CVD-EG) on SiC substrates are used to fabricate graphene gas sensors. Defects are introduced into graphene using SF₆ plasma treatment with a conventional inductively coupled plasma (ICP) system to improve the performance of the graphene gas sensors.

2. Experiments

A 4H-SiC (0001) single-crystal wafer with chemical-mechanical polishing was first cut into pieces of 10 × 10 mm². Before graphene growth, the samples were cleaned with a conventional cleaning process, and then etched in hydrogen at 1550 °C for 6 min to remove the surface scratches. A regularly stepped surface formed on the SiC surface. Five graphene samples, named A–E were prepared in this work, as shown in Table 1. Samples A and B are EG grown by SiC sublimation method under Ar atmosphere at 1650 °C. More details about the graphene growth can be found in Ref. [20]. They show n type doping with sheet resistance R_s of 770 Ω/square, carrier mobility μ of 800 cm²/(V·s), and sheet density N_s of 1 × 10¹³ cm^–2. Samples C and D are QFSEG grown by SiC sublimation and H intercalation by annealing the monolayer EG in hydrogen at 900 °C^[21]. They show p type with sheet resistance R_s of 110 Ω/square, carrier mobility μ of 3200 cm²/(V·s), and sheet density N_s of 1.8 × 10¹³ cm^–2. Sample E is CVD-EG. For the CVD growth of graphene, propane was the source gas, argon and hydrogen were carrier gases. The growth pressure was 900 mbar. The ratio of hydrogen to argon was 1 : 3. The sample shows p type with sheet resistance R_s of 770 Ω/square, carrier mobility μ of 500 cm²/(V·s), and sheet density N_s of 1.6 × 10¹³ cm^–2. Samples B, C, and D were then treated by SF₆ plasma for 7, 10, and 7 s, respectively.

Table Infomation Is Not Enable

For EG sample grown on Si-face SiC substrate, a crucial feature is the existence of a buffer layer between epitaxial graphene and SiC surface. Unpaired electrons of the buffer layer lead to a series of partially occupied localized states, which makes epitaxial graphene n type doping^[22]. For QFSEG sample, hydrogen atoms intercalated into the interface of graphene and SiC. The topmost Si atoms which for epitaxial graphene are covalently bound to this buffer layer, and are now saturated by hydrogen bonds. The buffer layer is turned into a quasi-free-standing graphene monolayer with its typical linear bands. Epitaxial monolayer graphene turns into a decoupled bilayer. The QFSEG samples show heavily p-type doping due to the polarization of the SiC substrate^[23]. For CVD-EG sample, C₃H₈ and H₂ are used for the graphene growth. Due to the presence of H atoms, the CVD-EG in situ formed quasi-free-standing graphene in the growth process^{[24, 25]}.

Micro-Raman scattering measurements were performed at room temperature (RT) with a spectrometer at 514 nm. The carrier mobility, sheet density, and sheet resistance of all the graphene samples were measured at RT.

The resistance of the graphene gas sensors was measured by a four-probe method. The devices were measured in a chamber with atmospheric nitrogen environment at RT. Before measurement, the graphene samples were annealed in N₂ at 200 °C for 30 min to remove possible contamination of the graphene samples. The resistance of the graphene device was stable over several hours in a pure N₂ gas, which verify the inertness of graphene to nitrogen. The signal to noise ratio at base line is around 0.01. The test target gas was diluted by the N₂ carrier gas by a gas calibration system.

3. Results and discussion

The Raman spectra of the graphene samples A–E and the SiC substrate are shown in Fig. 1. Graphene contains two dominated Raman bands of G-band at ~1582 cm^–1 and 2D band at ~2679 cm^–1 (the second order of D-band)^[26]. D peak around 1335 cm^–1 is originated from the vacancy or disorder of carbon atoms, which can be used to reflect the number of defects in graphene. The Raman spectra of graphene on SiC substrate unavoidably contain apparent background peaks from the SiC substrate, as shown in Fig. 1. From Fig. 1, it can be found that samples B, C, and D shows obvious D peak, indicating that the SF₆ plasma treatment introduces some defects into the graphene lattice.

Figure 1.(Color online) Raman spectra of (a) graphene samples A and B and the SiC substrate, and (b) graphene samples C, D, and E.

The performance of the five graphene gas sensors was measured and their corresponding sensitivity was calculated by^[4]

${\rm{Response}} = \frac{{\left| {R - {R_0}} \right|}}{{{R_0}}} \times 100,$ (1)

where R₀ is the resistance prior of the device to exposure to the target gas and R is the resistance after exposure to the target gas molecules. Sensor responses for NO₂ and NH₃ gases, p- and n-type dopants to graphene, are represented in Fig. 2.

Figure 2.(Color online) Sensor responses for (a, b) NO₂ and (c, d) NH₃.

As shown in Fig. 2(a), all the graphene sensors show an obvious response to NO₂ gas. The sensitivity of the graphene samples shows nearly liner dependence with the NO₂ gas concentration at the lower concentration region (400 ppb to 1 ppm), as shown in Fig. 2(b) and Table 2. The response of EG (samples A and B) to NO₂ gas is much higher than the QFSEG (samples C and D) and CVD-EG (sample E). Compared with sample A, the defect-engineered sample B shows great improvement in the sensitivity to NO₂ gas. The detection limit was calculated by the sensor’s signal processing performance by the following equation^[27]

Table Infomation Is Not Enable

${\rm{DL\;(ppm)}} = 3\frac{{{\rm{rms}}}}{{{\rm{slope}}}},$ (2)

where rms is the noise (signal to noise ratio at base line). The DL to NO₂ gas for samples A and B reached 1 and 2 ppb, respectively, indicating the high sensitivity of the EG samples to NO₂ gas. The DL to NO₂ gas for samples C, D and E were 60, 70, and 50 ppb, respectively.

QFSEG and CVD-EG samples (samples C, D, and E) show comparable response and DL to NO₂ gas. The EG samples (samples A and B) show much higher sensitivity to NO₂ gas than the QFSEG and CVD-EG samples. This may be due to the different doping types of the samples: samples A and B are n type doping; samples C, D, and E are p type doping; and, NO₂ gas is a well-known p-type dopant to graphene sample.

As illustrated in Fig. 3, a simple model was used to explain the different responses of graphene samples with different doping types (samples A, C/D, and E). The Fermi energies of EG and QFSEG and CVD-EG are different (Fig. 3(a)). The EG samples are n type (N_s = 1 × 10¹³ cm^–2) due to the Fermi level pinning effect of the buffer layer^[15]. After hydrogen intercalation in QFSEG, samples show heavily p-type doping (N_s = 1.8 × 10¹³ cm^–2) due to the polarization of the SiC substrate^[23]. The CVD-EG is also heavily p-type doping (N_s = 1.6 × 10¹³ cm^–2)^[28]. The density states of graphene are proportional to the energy with linear relationship of E ∝ N_s. Fig. 3(b) shows a schematic illustration of the relationship of resistance R and sheet density N_s of graphene^[29]. For the n-type EG sample, the initial resistance point is on the left-hand side of the resistance peak, as shown in Fig. 3(b) by the red circle. Adsorbing of NO₂ molecules causes hole doping of the EG system, leading to shift of the resistance of the EG sample closer to the charge neutral point (CNP) point (N_s = 0). This will lead to a rapid increase in the resistance (sharp peak region). For the QFSEG and CVD-EG samples, the samples are p type with the initial point in the right side of the resistance peak. Adsorbing of NO₂ molecules also causes hole doping of the graphene, leading to the resistance shift away from the CNP point (N_s = 0). The change of resistance due to surface doping is smaller compared to the EG samples.

Figure 3.(Color online) Schematic illustration of (a) the Fermi level shift and (b) resistance due to NO₂ doping for n-type EG and p-type QFSEG and CVD-EG.

Compared with sample A, the defect-engineered sample B shows a great improvement in the sensitivity to NO₂ gas. The response of the graphene sensors to NH₃ is not the same with NO₂. Only samples A and C show responses to NH₃, as shown in Fig. 2(c). Other samples show no obvious response to NH₃. These results are consistent with the reported results and first-principles calculations^{[19, 30]}. Theoretical studies have proven that compared to pristine graphene, adsorption of gas molecules is stronger on defective graphene than pristine graphene. Therefore, the introduction of defects into graphene remarkably improves the sensitivity of graphene gas sensor^[30]. Zhang et al. found that gas molecules (NO₂, NH₃) and defective graphene show stronger interactions than with pristine graphene by DFT calculations. They found the adsorption energy of NO₂ gas molecule on defective graphene is –3.04 eV, and it is –0.48 eV for that of pristine graphene. Meanwhile, the charge transfer between NO ₂ gas molecules and defective graphene (–0.38e) is also larger than pristine graphene (–0.19e)^[30]. For NH₃ gas, the adsorption energy of NH₃ gas molecule on defective graphene (–0.24 eV) is higher than that on pristine graphene (–0.11 eV). However, the charge transfer between NH₃ gas molecules and defective graphene (–0.02e) is the same with that of pristine graphene (–0.02e). The calculated charge transfer is relatively small for NH₃ gas than that for NO₂ gas. The small charge transfer of graphene with NH₃ gas is in agreement with the weak response of the graphene gas sensor to NH₃ observed in the experiments.

4. Conclusions

Epitaxial graphene, quasi-free-standing graphene, and CVD epitaxial graphene samples on SiC substrates are used to fabricate gas sensors. The epitaxial graphene shows the highest sensitivity to NO₂ gas. This can be explained by the doping type and resistance-carrier density relation. The defect-engineered sample shows improved sensitivity to NO₂ with response of 105.1% to 4 ppm NO₂, which is consistent with the DFT calculations that both of the adsorption energies and charge transfer increased. The samples show weak response to NH₃ due to the small charge transfer. Our results show the application potential of epitaxial graphene on SiC to NO₂ gas detection.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (61674131 and 61306006).

References

[1] F Schedin, A K Geim, S V Morozov et al. Detection of individual gas molecules adsorbed on graphene. Nat Mater, 6, 652(2007).

[2] A K Geim, K S Novoselov. The rise of graphene. Nat Mater, 6, 183(2007).

[3] K S Novoselov, V I Falko, L Colombo et al. A roadmap for graphene. Nature, 490, 192(2012).

[4] S S Varghese, S Lonkar, K K Singh et al. Recent advances in graphene based gas sensors. Sens Actuators B, 218, 160(2015).

[5] Y Dan, Y Lu, N J Kybert et al. Intrinsic response of graphene vapor sensors. Nano Lett, 9, 1472(2009).

[6] S Rumyantsev, G Liu, M S Shur et al. Selective gas sensing with a single pristine graphene transistor. Nano Lett, 12, 2294(2012).

[7] G Lee, G Yang, A Cho et al. Defect-engineered graphene chemical sensors with ultrahigh sensitivity. Phys Chem Chem Phys, 18, 14198(2016).

[8] A K Singh, M A Uddin, J T Tolson et al. Electrically tunable molecular doping of graphene. Appl Phys Lett, 102, 043101(2013).

[9] K Toda, R Furue, S Hayami. Recent progress in applications of graphene oxide for gas sensing: A review. Anal Chim Acta, 878, 43(2015).

[10] W Li, X Geng, Y Guo et al. Reduced graphene oxide electrically contacted graphene sensor for highly sensitive nitric oxide detection. ACS Nano, 5, 6955(2011).

[11] M W K Nomani, R Shishir, M Qazi et al. Highly sensitive and selective detection of NO₂ using epitaxial graphene on 6H-SiC. Sens Actuators B, 150, 301(2010).

[12] R Pearce, T Iakimov, M Andersson et al. Epitaxially grown graphene based gas sensors for ultra sensitive NO₂ detection. Sens Actuators B, 155, 451(2011).

[13] I Iezhokin, P Offermans, S H Brongersma et al. High sensitive quasi freestanding epitaxial graphene gas sensor on 6H-SiC. Appl Phys Lett, 103, 053514(2013).

[14] A A Lebedev, S P Lebedev, S N Novikov et al. Supersensitive graphene-based gas sensor. Tech Phys, 61, 3, 453(2016).

[15] S Novikov, N Lebedeva, A Satrapinski. Graphene based sensor for environmental monitoring of NO₂. J Sen, 2015, 7(2015).

[16] K Wetchakun, T Samerjai, N Tamaekong et al. Semiconducting metal oxides as sensors for environmentally hazardous gases. Sens Actuators B, 160, 580(2011).

[17] T Zhang, S Mubeen, N V Myung et al. Recent progress in carbon nanotube-based gas sensors. Nanotechnology, 19, 332001(2008).

[18] B Kumar, K Min, M Bashirzadeh et al. The role of external defects in chemical sensing of graphene field-effect transistors. Nano Lett, 13, 1962(2013).

[19] M G Chung, D H Kim, H M Lee et al. Graphene-based composite materials for chemical sensor application. Sens Actuators B, 166/16, 172(2012).

[20] C Yu, J Li, Q B Liu et al. Buffer layer induced band gap and surface low energy optical phonon scattering in epitaxial graphene on SiC (0001). Appl Phys Lett, 102, 013107(2013).

[21] C Yu, Q B Liu, J Li et al. Preparation and electrical transport properties of quasi free standing bilayer graphene on SiC (0001) substrate by H intercalation. Appl Phys Lett, 105, 183105(2014).

[22] O Pankratov, S Hensel, M Bockstedte. Electron spectrum of epitaxial graphene monolayers. Phys Rev B, 82, 121416(2010).

[23] J Ristein, S Mammadov, T Seyller. Origin of doping in quasi-free-standing graphene on silicon carbide. Phys Rev Lett, 108, 246104(2012).

[24] T Ciuk, W Strupinski. Charge carrier concentration and offset voltage in quasi-free-standingmonolayer chemical vapor deposition graphene on SiC. Carbon, 93, 1042(2015).

[25] T Ciuk, P Caban, W Strupinski. Statistics of epitaxial graphene for Hall effect sensors. Carbon, 101, 431e438(2016).

[26] A C Ferrari. Raman spectroscopy of graphene and graphite: Disorder, electron–phonon coupling, doping and nonadiabatic effects. Solid State Commun, 143, 47(2007).

[27] Y R Choi, Y G Yoon, K S Choi et al. Role of oxygen functional groups in graphene oxide for reversible room-temperature NO₂ sensing. Carbon, 91, 178(2015).

[28] M Portail, A Michon, S Vezian et al. Growth mode and electric properties of graphene and graphitic phase grown by argon-propane assisted CVD on 3C-SiC/Si and 6H-SiC. J Cryst Growth, 349, 27(2012).

[29] D Waldmann, J Jobst, F Speck et al. Bottom-gated epitaxial graphene. Nat Mater, 10, 357(2011).

[30] Y H Zhang, Y B Chen, K G Zhou et al. Improving gas sensing properties of graphene by introducing dopants and defects: a first-principles study. Nanotechnology, 20, 185504(2009).