Decorated CNT based on porous silicon for hydrogen gas sensing at room temperature

Abstract

A new triple-component sensor for detection of H₂ was developed based on porous silicon and CNTs. An increase in deposition of CVD catalysis was shown to promote a high and fast response. Also, it was shown that the composite system exhibited good selectivity.

---

Environmental impacts must be taken into account in efforts to identify renewable and clean energy resources. Hydrogen is an excellent candidate, which satisfies many of the required criteria. However, hydrogen is indiscriminable among other gases and also flammable; it can explode at a concentration of only 4%.¹ Therefore, to address safety concerns during applications of hydrogen, it has become necessary to develop hydrogen gas sensing techniques. Many metal composites, such as gold and platinum ones, have been evaluated for this purpose.^2–4

Scientists have investigated use of porous silicon surface for sensing of different gas molecules, such as H₂O, C₂H₅OH, CH₃OH, C₃H₇OH, CO_x, NO_x, NH₃, PH₃, O₂, HCl, SO₂, H₂S, and H₂,⁵ and structures incorporating porous silicon have been found useful in gas sensing systems.⁶ Also, carbonaceous nanomaterials such as graphene, carbon nanofiber, and carbon nanotube (CNTs) have been examined as potential hydrogen gas sensors,^7–9 with the latter attracting a large number of studies because of its appropriate hydrogen sensing potential.¹⁰ The sensing performance of either porous silicon or CNTs did not meet the high standards required; however, use of composites or functionalized structures has resulted in improved hydrogen sensing.^1,4,11

Several methods have been developed to improve the sensitivity of either porous silicon- or CNT-based hydrogen gas sensors. Metal-decorated porous silicon has been identified as a sensor of various materials, such as Mg and ZnO.^12–15 Sanger et al. prepared Pd/Mg thin films by DC magnetron sputtering. Proper reversibility of change in resistance was recognized. Moreover, a short recovery time (∼1 min at 100 °C) during hydrogenation/dehydrogenation was observed. Also, metal-decorated CNTs have been prepared, and showed significant improvement in hydrogen gas sensing.^16–18

As prepared, the treated arrays described above are useful for hydrogen sensing. However, combining structures of porous silicon and CNTs may offer significant improvement. In this communication, we propose fabrication of a multi-component system as a sensitive and selective H₂ gas sensor, at low cost and following a simple procedure. We used electrochemical anodization (EA) to prepare porous silicon; the quality of this process has been considered in a previous study.¹⁹ Results are reported for a triplet hydrogen gas sensor based on decorated CNTs grown over porous silicon. As far as we know, there are no previous reports describing incorporation of decorated CNTs over porous substrate. Furthermore, we assessed the contribution of decorated arrays to the sensing behaviour.

Experimental details (S1), experimental setup (S2), and details of measurements (S3) are given in the ESI.†

The microstructure and morphology of samples were analyzed using FESEM (Fig. 1 shows the top view images; cross-section images can be found in our accepted paper¹⁹). Fig. 1(a) and (b) illustrate Pd catalysts over porous silicon substrate for C1 and C2, respectively. A greater amount of Pd nanoparticles were deposited over C2, compared with C1. This could be attributed to the duration of the electroless process. Also, the Pup corn-like arrays on Fig. 1(b) translate over layer growth of Pd nanoparticles. The Pd grains were distributed in the range of 7–18 nm on both samples. Fig. 1(e) and (f) show the grown CNTs over C1 and C2, respectively; the distributed pattern of CNTs was demonstrated in a large area (Fig. 1(c) and (d)). Because of strong deposition of catalyst, especially in the case of C2, the CVD (Chemical Vapor Deposition) operation simulated the growth of Pd nanoparticles along the nanotubes (tip growth). From Fig. 1(f), it could be stated that intensified tip growth is seen in the case of C2. This may be ascribed to the potential substrate, which was provided by high deposition of Pd nanoparticles. Fig. 1(g) and (h) show the functionalized nanotubes for C1 and C2, respectively. The functionalization occurred through introduction of several defects on the nanotube sidewalls; these defects could be a hot centre for reduction of Pd²⁺. Regardless of the distribution of nanotubes, growth of Pd nanoparticles was seen on the walls for both C1 and C2. The chemical composition of the fabricated device was measured using energy dispersive X-ray spectroscopy (EDS) characterizations (results are given in S4†).

Fig. 1 FESEM analysis for C1 and C2. (a) and (b) Before CVD; (c)–(f) after CVD; (g) and (h) post treatment.

As we used palladium nanoparticles as a major sensitive component to H², especially for C2, it was expected that the device would be responsive to test gases at low temperatures and low concentration. The studies demonstrated that the proposed devices are capable of detecting hydrogen molecules at temperature and concentration as low as 18 °C and 0.2%, respectively.

The sensing character of a hydrogen sensor is usually evaluated by sensitivity and response time. The response (S) is defined by the ratio of maximum resistance against resistance measured without hydrogen for a gas mixture with a fixed concentration of hydrogen, and is expressed as

The response and recovery times are defined as the time required to reach 90% and 60% of the total change in the resistance on the supply or removal of hydrogen.

S5† demonstrates a typical frequent response of the triple-component sensor to 2% H₂ at room temperature. Release of hydrogen gas increases the initial resistance; with enhancement continuing to saturation approaching maximum. As the hydrogen flow stops, the resistance decreases towards the initial value. The stability behaviour of the proposed device is shown in illustration S5†.

Fig. 2 demonstrates the sensor response to proposed concentrations, with the response magnitude decreasing as hydrogen concentration decreases. The great slope at concentrations lower than 4000 ppm may be the real response. Indeed, at concentrations higher than the threshold (the breakage point), responses are reduced; this could be attributed to a hydrogen storage property of the systems. Studies have demonstrated the key role of Pd (nano) particles in detection as well as storage of hydrogen gas.²² Generally, hydrogen gas, when exposed to Pd, can be detected and its trapping in the lattice of Pd particles can be followed. Our group has proved this claim and the results will be published soon.

Fig. 2 Gas response in terms of hydrogen concentration for C1 and C2.

However, the reduction of slope in Fig. 2 may be attributed to the nature of the sensor mechanism. According to the catalytic activity of Pd, a possible mechanism of detection for the proposed system is:

Using the Langmuir isotherm concept, interaction of hydrogen molecules can be described as follows:^23,24

Adsorption

r_a = k_aP(1 − θ)²

Desorption

r_d = k_dθ²

where k_a and k_d are the adsorption and desorption constants, respectively; P is the hydrogen pressure; and θ is the fraction of occupied sites for hydrogen adsorption. At low concentrations of hydrogen, the sites are only slightly occupied by the hydrogen atoms.

Using the above equations, the maximum adsorption and minimum desorption rates can be derived, showing that the response of the dilute mixture depends on hydrogen concentration, linearly. As the hydrogen concentration increases, the surface occupancy rises quickly; consequently, the response declines as (1 − θ)². This phenomenon results in saturation level, i.e. there are minimum sites available for hydrogen to be adsorbed. The examined sensors demonstrate relatively fast response times, attributed to fast kinetics of adsorption and desorption of hydrogen molecules at a wide range of concentrations; while the recovery times are longer than expected, which results from both working at room temperature and also the hydrogen storage capacity of our triplet structure. As in comparison with the literature these systems are capable of a fast response at relatively low temperatures, there is no evidence of saturation. Moreover, we tested the C2 sample under the aforementioned conditions and at a hydrogen concentration of 40 000 ppm (4 vol%). A high response of 167.14 was demonstrated, with response and recovery times of 3 and 21 seconds, respectively. This indicates high capability of the proposed system, which was not saturated, even at critical (flammable) concentration. Further evidence is supplied in the ESI (S6†). Generally, the storage capacity of such structure should be considered whenever exposes to hydrogen species.

We considered the resistance change origins and mechanisms. The electroless duration can determine the density of Pd nanoparticles over substrate, either porous silicon or CNTs substrate. From Fig. 2, it can be seen that the sensor of C2 is more efficient at sensing hydrogen gas, while no recognizable change in resistance was defined for C1.

Dissociated hydrogen molecules occurred at the nanoparticle surfaces, and these could have an effect on the CNT/Pd interface. Recently, it was shown that hydrogen atoms form covalent bonds with graphene, and the resulting structure will demonstrate higher work function compared with pure graphene. Also, enhancement of the separation distance of graphene and Pd may also lead to broadness of the Fermi-level.²⁵ As CNTs are a rolled form of graphene, this is true in the case of CNTs. Subsequently, the free carrier concentrations may increase the conductance of the interface, as supported by previous work.²⁶ Following injection of the hydrogen mixture, composition of the Pd nanoparticles, whether tip grown initial catalysts or CNT-decorated ones, changed to palladium hydride (further details are given in S7†), which possesses lower work function in comparison with pure Pd.^27,28 The lower work function associated with palladium hydride is beneficial to the transfer of more electrons from the Pd nanoparticles to the CNTs, which leads to trapping of the p-type carriers in the CNTs; this can be recognized as increment of resistance. However, there is another reason for enhancement of resistance (S8†). The grown CNTs also have an effect on the response. Ghasempour et al.²⁹ prepared a Pd/MWCNT system and proved that both reversible and irreversible interaction of hydrogen can affect the response. They demonstrated that as hydrogen is exposed to the hybrid system, the slight structure of C=O may reduce to C–O, O–H, and, chemically, C–H. Therefore, both hydrogen chemisorption and diffusion in the interior of MWCNT (in the slight unblocked tips) should be considered. Overall, results in the case of irreversible behaviour revealed a sensible reduction in the response, and consequently, response time. The irreversibility introduces a stronger effect on the recovery time. The measurements demonstrated that this phenomenon is far-reaching in the case of lower concentrations, i.e. the deviation of normal absorption/desorption of hydrogen species leads to higher value of recovery times. In fact, because of fresh and/or unblocked sites, the irreversible adsorption is more decisive.

The porous silicon has a key role in the obtained response, as the sensitivity of porous silicon has been reported in the literature.³⁰ This role may be defined through contribution to superior growth of CNT, compared with that on a smooth surface. It has been demonstrated that porous silicon, as a substrate, strongly optimizes the condition of CNT growth,²¹ because of the large surface area offered by the porous substrate. Moreover, porous silicon may contribute to align-growth of CNTs. It has been proved that the large surface area as well as aligned distribution of CNTs promote the performance of gas sensors.³¹

Jung et al.³¹ derived thermodynamics relations that predict the interaction of gas molecules and solid surface. According to their study, thermal annealing is favourable for desorption, consequently improving recovery of the sensors.²² We investigated the effect of temperature in the as-prepared samples through repetition of experiments at 50 °C for hydrogen concentrations of 0.6, 1.5, and 2 vol%. There was a 5% improvement in response, and also response time decreased by a factor of 3%. Therefore, use of higher temperature could be considered for CNTs.

We carried out current–voltage characterizations for the proposed sensor. Fig. 3 shows the current–voltage measurements in air and hydrogen concentration of 1.5%; the sensor can be considered to exhibit ohmic behaviour. However, as mentioned before, Rahimi et al. demonstrated that distributed Pd nanoparticles over porous silicon exhibit diode behaviour.²⁰ They reported that the I–V response may be in agreement for air and hydrogen containing gas at low voltages, whereas a great offset could be measured at high voltages. Also, our group proved that a great rectification could be achieved at voltages higher than 1 V.¹⁹ Fig. 3 shows that ohmic behaviour was seen on exposure to both air and hydrogen; however, a minor offset emerges at a voltage of ∼2.3 V. This may be attributed to the presence of nanotubes, i.e. the CNTs and decorated Pd could offer an equivalent circuit. This may introduce a current pathway opposed to that defined through porous silicon and local deposited Pd. However, as illustrated in Fig. 3, there is a threshold. At voltages above this value (∼2.3 V), diode behaviour may be more dominant.

Fig. 3 Typical I–V curve for as-prepared sample in air (line) and 1.5% H₂ (dashed).

Selectivity is important for a gas sensor, i.e. the capability to sense (recognize) a given gaseous species among a mixture. Fig. 4 demonstrates the selectivity of the competitive sensor exposed to different gases (H₂, CH₄, and CO) at 1.5 vol%. The best selectivity refers to H₂ (about two times higher than other gases), while the responses to other gases are substantially low.

Fig. 4 Selectivity of as-synthesized sample.

The higher selectivity of CH₄ in comparison with CO could be attributed to the high content of hydrogen. Selectivity of the sensor also may be explained through several parameters such as electron affinity, catalytic efficiency, and adsorption property. Clearly, for the gas molecules with low electron affinity, the energy needed for gas sensing reaction is small; thereby, high sensitivity can be achieved. Moreover, high catalytic efficiency of the surface structure as well as a high amount of gas adsorption over the sensing array may have roles in detection of target gas. The aforementioned items (electron affinity, catalytic efficiency, and adsorption property) are provided for hydrogen gas, while CH₄ and CO could not meet the criteria.

The proposed system is in progress and will be presented in future studies. According to the results which are comparable with previous similar research, we worked on improving the response, both amplitude and time, by means of chemical techniques and devised components.

In summary, a composite system based on porous silicon was fabricated as a hydrogen gas sensor. Silicon wafer was etched using an EA process. Next, a porous substrate was treated with Pd deposition using an electroless technique. The treated substrate was employed for growth of CNT using a CVD technique. The final sample was exposed to hydrogen gas, and revealed high and fast responses. Also, it was demonstrated that increasing the duration of the electroless process results in efficient performance. Selectivity of the competitive sample was examined, showing good results for hydrogen sensing.

Notes and references

1. J. Wang, S. Rathi, B. Singh, I. Lee, H.-I. Joh and G.-H. Kim, ACS Appl. Mater. Interfaces, 2015, 7, 13768–13775.
2. J. S. Lee, S. G. Kim, S. Cho and J. Jang, Nanoscale, 2015, 7, 20665–20673.
3. B. Devadas, S. Cheemalapati, S.-M. Chen and M. Rajkumar, RSC Adv., 2014, 4, 45895–45902.
4. P. Rai, S. M. Majhi, Y.-T. Yu and J.-H. Lee, RSC Adv., 2015, 5, 76229–76248.
5. S. Ozdemir and J. L. Gole, Curr. Opin. Solid State Mater. Sci., 2007, 11, 92–100.
6. H.-G. Kim and K.-W. Lee, Sens. Actuators, B, 2015, 219, 10–16.
7. Z. Wang, S. Liu, T. Jiang, X. Xu, J. Zhang, C. An and C. Wang, RSC Adv., 2015, 5, 64582–64587.
8. W. Yuan, Q. Zhou, Y. Li and G. Shi, Nanoscale, 2015, 7, 16361–16365.
9. L. Zhang, X. Wang, Y. Zhao, Z. Zhu and H. Fong, Mater. Lett., 2012, 68, 133–136.
10. S. JooáPark and O. SeokáKwon, Chem. Commun., 2013, 49, 4673–4675.
11. M. Zhang, L. L. Brooks, N. Chartuprayoon, W. Bosze, Y.-H. Choa and N. V. Myung, ACS Appl. Mater. Interfaces, 2013, 6, 319–326.
12. A. A. Ensafi, M. M. Abarghoui and B. Rezaei, Sens. Actuators, B, 2014, 196, 398–405.
13. A. Sanger, A. Kumar, S. Chauhan, Y. K. Gautam and R. Chandra, Sens. Actuators, B, 2015, 213, 252–260.
14. D. Yan, M. Hu, S. Li, J. Liang, Y. Wu and S. Ma, Electrochim. Acta, 2014, 115, 297–305.
15. G. Barillaro, in Handbook of Porous Silicon, Springer, 2014, pp. 845–856.
16. Q. Liu, J. Tian, W. Cui, P. Jiang, N. Cheng, A. M. Asiri and X. Sun, Angew. Chem., 2014, 126, 6828–6832.
17. S. M. Seo, T. J. Kang, J. H. Cheon, Y. H. Kim and Y. J. Park, Sens. Actuators, B, 2014, 204, 716–722.
18. J. García-Aguilar, I. Miguel-García, Á. Berenguer-Murcia and D. Cazorla-Amorós, Carbon, 2014, 66, 599–611.
19. H. Ghorbani Shiraz and F. Razi Astaraie, et al., Silicon DOI:10.1007/s12633-016-9417-9.
20. F. Rahimi and F. Razi, J. Phys. D: Appl. Phys., 2004, 38, 36.
21. A. Nikfarjam, F. Razi and S. Z. Mortazavi, Sens. Actuators, A, 2010, 162, 24–28.
22. J. Zou, B. Zdyrko, I. Luzinov, C. L. Raston and K. S. Iyer, Chem. Commun., 2012, 48, 1033–1035.
23. M. G. Chung, D.-H. Kim, D. K. Seo, T. Kim, H. U. Im, H. M. Lee, J.-B. Yoo, S.-H. Hong, T. J. Kang and Y. H. Kim, Sens. Actuators, B, 2012, 169, 387–392.
24. R. D. Martínez-Orozco, R. Antaño-López and V. Rodríguez-González, New J. Chem., 2015, 39, 8044–8054.
25. B. H. Chu, C. Lo, J. Nicolosi, C. Chang, V. Chen, W. Strupinski, S. Pearton and F. Ren, Sens. Actuators, B, 2011, 157, 500–503.
26. M. Shafiei, P. G. Spizzirri, R. Arsat, J. Yu, J. du Plessis, S. Dubin, R. B. Kaner, K. Kalantar-Zadeh and W. Wlodarski, J. Phys. Chem. C, 2010, 114, 13796–13801.
27. M. G. Chung, D.-H. Kim, D. K. Seo, T. Kim, H. U. Im, H. M. Lee, J.-B. Yoo, S.-H. Hong, T. J. Kang and Y. H. Kim, Sens. Actuators, B, 2012, 169, 387–392.
28. D. Gupta, D. Dutta, M. Kumar, P. Barman, C. Sarkar, S. Basu and S. Hazra, Sens. Actuators, B, 2014, 196, 215–222.
29. R. Ghasempour, S. Mortazavi and F. Rahimi, Int. J. Hydrogen Energy, 2010, 35, 4445–4449.
30. F. Geobaldo, B. Onida, P. Rivolo, S. Borini, L. Boarino, A. Rossi, G. Amato and E. Garrone, Chem. Commun., 2001, 2196–2197.
31. D. Jung, M. Han and G. S. Lee, Sens. Actuators, A, 2014, 211, 51–54.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra03541h

---