Open Access
Add to BibSonomy
Abstract
Silicon has been studied as a room-temperature material for electrical-based gas sensing but the sensing performance after surface passivation or natural aging is unacceptable. In the present work, we report that for a gas sensor based on the femtosecond-laser structured silicon hyperdoped with sulfur, the gas sensing performance after long-term aging can be significantly enhanced by using a photovoltaic sensing mechanism. After sensor aging, the recorded response/recovery time is 478/2550 s in response to 50 ppm NH3. In comparison, by using the new mechanism, the response/recovery time is much decreased and the shortest is recorded as 292/930 s. Moreover, the relative gas response could be increased by nearly 2 orders of magnitude. Even at a dryer environment where the gas adsorption/desorption process could take hours long, a much enhanced and rapid response is available in the same way. The enhanced sensing performance could be controlled by the bias voltage or by the light density. The results show that for the aged silicon surface, it can also be a potential gas sensing material through different working principles.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Silicon has been studied as a room-temperature material for electrical-based gas sensing, in the forms like planar silicon [1], porous silicon [2], structured silicon [3], thin-film silicon [4] and even silicon nanowires [5]. It is promising first of all for fabrication of micromachining gas sensor platforms [2]. One of the major setbacks, however, that prevents its practical use is the issue of the long-term stability. As a chemically active material, the unpassivated silicon surface would experience a continuous oxidization in air, which means its surface properties would evolve with storage time [3,6]. Therefore, gas sensing parameters are not reproducible and especially, the sensitivity decreases while the response/recovery time increases significantly [6]. Even though temporal stability of silicon surface could be achieved through controlled passivation like oxidation [7] or carbonization [8], the cost is a longer response/recovery time compared to that of an untreated surface, which is hardly to be deployed in real applications.
Ammonia is a useful chemical gas in industrial, agricultural and medical applications, but it is also physically damaging [9]. The permissible exposure limit of ammonia in work place is set 50 ppm by the Occupational Safety and Health Administration, thus many occasions call for the need for ammonia detection. Recently we study the femtosecond-laser structured silicon hyperdoped with sulfur as a gas sensing material at room temperature. As it is particularly selective to ammonia, the ammonia sensing performance is thoroughly investigated [6]. This material has microstructures morphologically and is rich in surface defects [10], so naturally it can be used for a conductometric gas sensor. Out of expectation, the results show that the response is extremely low. However, considering the excellent optoelectronic properties of the silicon hyperdoped with sulfur [11], we find light can largely facilitate the gas sensing performance for this material. Therefore, a specific lighting method is introduced, called the asymmetric light illumination with which the sensor works under a photoelectron-based photovoltaic mechanism. The above results mainly apply to the freshly prepared material, and as for the aged one, primary studies also show a positive result by using the light. In the present work, we report in detail that for a gas sensor based on the femtosecond-laser structured silicon after substantial nature aging, with low response for ammonia and nearly unacceptable response/recovery time, the gas sensing performance can be significantly enhanced and even controllable based on such photoelectron-based photovoltaic mechanism.
2. Experimental
The structured silicon was prepared on double-polished high-resistivity (3–5 kΩ cm) phosphorus-doped Si (111) wafer. The silicon surface was irradiated by a 1 kHz train of 190 fs and 515 nm laser pulses under a SF6 atmosphere. The fluence of the focused beam was about 4.7 kJ/m2 and a progressive scanning of the laser on the surface created a structured area of 3×8 mm2. This led to the incorporation of sulfur in high concentrations above 1019 cm−3 [12]. Subsequently, the silicon was annealed at 873 K for 15 min. Aluminum electrodes were thermally evaporated onto the structured region for ohmic contacts. A more detailed experimental process was published elsewhere [6]. As shown in Fig. 1, the sensing surface of the sensor is covered by arrayed microstructures that increase the surface area of silicon. The microstructures consist of a crystalline silicon core covered with a disordered layer less than 1 µm thick and annealing could improve the crystallinity of the layer [12,13]. The sensor was stored in laboratory environment for natural aging. During the aging process, the surface morphologies did not show visible changes under the scanning electron microscope (SEM) because only a thin layer (∼7 nm) of silicon dioxides could be grown on the surface [14]. The interior structure under the surface would not be affected by the aging.
Gas sensing measurement was performed by using a homemade static volumetric system in which a specific volume of standard gas could be injected by a gas syringe though a rubber plug into an openable and sealed acrylic chamber. For using the light purposely, a white-light LED was used as the light source and a light-proof tape was mounted above the sensor to shield half of the sensing surface from the light source, resulting in the asymmetric light illumination. The external bias source was provided and the output current signal was measured by using a (Keysight B2902A) source-and-measurement unit. The time step for measuring the response transients is 1 s and the instrumental uncertainty of the current is 0.1 nA. All the measurements were taken at room temperature.
3. Results and discussion
3.1 Aging of sensing parameters as a conductometric gas sensor
The sensing properties are first investigated under the principle of a conductometric gas sensor, and are characterized by the sensor’s response to 50 ppm NH3 in a dark environment. In order to compare the sensing parameters before and after aging, gas sensing measurements are performed at different states of the sensor. Figure 2 shows response transients of the sensor in the fresh state (1 d after fabrication) and in the aged state (44 d after fabrication). The sensor shows a typical n-type response, i.e., the conductance increases in response to a reducing gas like NH3. The sensor response (R) is defined as the change of conductance (ΔG) with respect to the base conductance in air (G0), i.e.,
Response time (T90) is defined as the time taken from G0 to attain G0 + 90%ΔG and recovery time (T10) is the time taken from G0 + ΔG to regain G0 + 10%ΔG. Figure 2 shows an increased sensor response from 4.1% [Fig. 2(a)] to 8.3% [Fig. 2(b)] after 44 d natural aging mainly arising from a decrease of baseline conductance. The more obvious change is that the response/recovery time is prolonged by at least an order of magnitude, from 27/221 s to 478/2550 s. It means that for the aged sensor, it should take at least 100 mins for the baseline to fully recover to the initial state. Originated from the natural oxidation of the silicon surface, the aged performance can hardly meet the real working requirements.3.2 Sensing parameters under the asymmetric light illumination
To enhance the sensing performance, light is used as a driving source for the sensor. Under the asymmetric light illumination, a lateral photovoltaic effect appears and results in a zero-bias photocurrent output [inset of Fig. 3(a)]. This photocurrent is in direct proportion to the light density [6] and it is measurable to reflect the sensor’s response to NH3. Figure 3(a) shows response transients of the aged sensor (45 d after fabrication) under different light densities. From Fig. 3(a), we obtain the gas response under a similar definition to Eq. (1), i.e.,
The relation between the gas response R and the basecurrent (reflecting different light densities) is outlined in Fig. 3(b) and is well fitted by a power function y = 11.52x−0.63 (with Adj. R2 = 0.995). From the relation, the gas response tends to increase at lower light densities. The maximum measured gas response is 705%, or ∼84.9 times the response without the light (Fig. 2). For the fresh sensor, a similar power function relation between the gas response and the light density could be obtained (see supplementary material of [6]). However, we observe significantly higher gas response in general on the aged sensor due to lower base current after aging.
Fig. 3. (a) Response transients of the aged sensor under the asymmetric light illumination of different light densities. Inset shows measurement circuit. (b) Gas response versus basecurrent plot with the fitting curve. (c) Response time versus basecurrent plot. (d) Recovery time versus basecurrent plot.
Moreover, response/recovery time under different light densities is shown in Figs. 3(c) and 3(d), respectively. The response/recovery time tends to increase at lower light densities. But even with the weakest light used (at ∼5 mW power consumption), the response/recovery time (403/1552 s) is obviously shorter than that without the light [Fig. 2(b)]. Increasing the light density leads to an even more rapid response. The shortest response/recovery time is measured to be only 292/930 s. Therefore, it can be asserted that by solely using the asymmetric light illumination, the gas response and response/recovery time could be greatly improved for the aged sensor. The reason for the enhanced sensing performance is due to a different working mechanism under the light from that without the light. Without the light, the sensor works by the change of surface conductance. Under the asymmetric light illumination, the sensor works under a photovoltaic self-powered sensing mechanism and the transducer function of the sensor is based on photoelectrons [6]. Also, there is a balance between an increased gas response and shortened response/recovery time under different light densities, which means there is a freedom to regulate the light density to meet different working scenarios.
It should be noted that the natural aging on silicon surface is extremely slow but gradual. The response/recovery time without the light will further increase and can be difficult to measure. The asymmetric light illumination could still enhance the sensing parameters to some extent. However, the response/recovery time would further increase even in the asymmetrical light illumination. Although our method could not fully resolve the long-term stability issues of silicon-based sensors, it is indicated that light could increase the life span of the sensor to more than 30 days, which is the minimum requirement for a practical sensor [14].
3.3 Sensing parameters under different humidities
Although the relation between the gas response and the basecurrent can be well fitted, the response/recovery time exhibits a non-monotonic dependence to the basecurrent [Figs. 3(b)–3(d)]. This is ascribed to the effects of humidity. During the measurement of the results in Fig. 3, the relative humidity (29–34% RH) and temperature (25.2–25.7 °C) is also recorded for each set of data. We find a strong correlation between the RH and the measured response/recovery time, as visually compared in Fig. 4. This indicates that the response/recovery time is a joint result of both light density and the RH.
Fig. 4. Correlation between the RH and the measured response/recovery time at different light densities.
To further investigate the effects of the RH on the sensor parameters, response transients were measured at different RHs both for conductometric gas sensing and for photovoltaic gas sensing. The statistical results are summarized in Table 1 for simplicity. On the one hand, as RH increase, the gas response slightly decreases under the conductometric sensing mechanism but increases under the photovoltaic sensing mechanism. This implies that the enhancement effect of light on gas response is more favorable at more humid atmosphere. On the other hand, the response/recovery time is two-fold increased at a dryer atmosphere (24 ± 1%RH), especially working as a conductometric sensor, so the repeated use of the sensor in a short term seems impossible. However, more gratifying is that by using the new principle under the asymmetric light illumination, response/recovery time can be significantly shortened at different RHs.
Table 1. Effects of the RH on the sensor parameters.
3.4 Effects of bias voltage on sensing mechanism
The different sensing mechanisms with and without the light find expression in the varied gas sensing parameters. Under the principle of a conductometric gas sensor, the adsorption of the ammonia gas increases the surface conductance, while under the principle of the photovoltaic gas sensing, the adsorption of the ammonia gas leads to an increase of the photoelectrons. By integrating the two working principles, a regulated sensing performance is expected.
Figure 5 shows response transients of the aged sensor under the asymmetric light illumination (of a fixed light density) and under different bias voltages simultaneously. The change of the sensing performance depends on the relative amplitude of the bias voltage |VB| to the photovoltage |Vph| (∼1 mV). When the |VB| is much greater than the |Vph|, the sensor shows similar response transients to that in the dark [Fig. 5(a) and Fig. 2(b)]. But when the |VB| is approximately equal to the |Vph|, the sensor shows fast and repeatable responses on the measured time scale [Fig. 5(b)].
Fig. 5. Response transients of the aged sensor under the asymmetric light illumination and different bias voltages (VB). (a)|VB| ≫ |Vph|. (b)|VB| ∼ |Vph|.
Table 2 summarizes the sensor parameters at different bias voltages with that in the dark for comparison. Overall, when |VB| ≫ |Vph|, the gas response is relatively low and response/recovery time is fairly long. However, when |VB| ∼ |Vph|, an enhanced sensing performance is exhibited with improved gas response and more rapid response/recovery time. The results indicate that under the joint drive of the asymmetric light illumination and the bias voltage, there is a competition between the two mechanisms. Even under the asymmetric light illumination, the sensing mechanism is dominated by the change of surface conductance at a large bias voltage. But when the bias voltage is much smaller, the proportion of photoelectrons becomes dominant and the enhanced response is attributed to the principle of the photovoltaic gas sensing.
Table 2. Sensor parameters at different bias voltages.
It is indicated from the above results that both the response and response/recovery time could be regulated by carefully choosing the bias voltage (Fig. 5 and Table 2) or by solely adjusting the light density (Fig. 3). With the combination of the two driving sources, one could regulate the sensing performance to meet actual requirements.
In addition, as previously described, the gas sensor based on the hyperdoped silicon requires the usage of fs-laser irradiation on silicon surface. There are two potential ways to reduce the time cost for mass manufacturing such a sensor. One is to develop newer technology to integrate the light source and the sensing material in order to reduce the required size of the hyperdoped silicon (with development costs). The other is to improve the fabrication efficiency of the hyperdoped silicon, for example, using ns-pulsed laser, higher laser intensity, faster scanning speed, bigger spot size, etc. (with equipment costs). However, the asymmetric light illumination scheme may also be integrated with other established sensing materials to achieve better sensing performance and more cost-efficient manufacture.
4. Conclusion
In summary, the femtosecond-laser structured silicon is fabricated as an ammonia gas sensor. After experiencing a substantial natural aging, the sensor exhibits an unacceptable sensing performance under the principle of the conductometric gas sensing. However, when the sensor works under the asymmetric light illumination, an improved gas response and more rapid response/recovery time are measured under the principle of the photovoltaic gas sensing. The results show that for the aged silicon surface, it can also be a potential gas sensing material through different working principles. Moreover, under the joint drive of the asymmetric light illumination and the bias voltage, there is a competition between the two mechanisms. With the balance of the two driving sources, one could regulate the sensing performance to meet actual requirements.
Funding
National Natural Science Foundation of China (61675045); National Basic Research Program of China (973 Program) (2012CB934200); Specialized Research Fund for the Doctoral Program of Higher Education of China (20130071110018).
Disclosures
The authors declare no conflicts of interest.
References
1. L. T. T. Tuyen, D. X. Vinh, P. H. Khoi, and G. Gerlach, “Highly sensitive NOx gas sensor based on a Au/n-Si Schottky diode,” Sens. Actuators, B 84(2-3), 226–230 (2002). [CrossRef]
2. G. Korotcenkov and B. K. Cho, “Porous semiconductors: advanced material for gas sensor applications,” Crit. Rev. Solid State Mater. Sci. 35(1), 1–37 (2010). [CrossRef]
3. X.-L. Liu, S.-W. Zhu, H.-B. Sun, Y. Hu, S.-X. Ma, X.-J. Ning, L. Zhao, and J. Zhuang, “‘Infinite sensitivity’ of black silicon ammonia sensor achieved by optical and electric dual drives,” ACS Appl. Mater. Interfaces 10(5), 5061–5071 (2018). [CrossRef]
4. A. Helwig, G. Müller, G. Sberveglieri, and G. Faglia, “Gas sensing properties of hydrogenated amorphous silicon films,” IEEE Sens. J. 7(11), 1506–1512 (2007). [CrossRef]
5. J. Yun, C. Y. Jin, J. H. Ahn, S. Jeon, and I. Park, “A self-heated silicon nanowire array: Selective surface modification with catalytic nanoparticles by nanoscale Joule heating and its gas sensing applications,” Nanoscale 5(15), 6851–6856 (2013). [CrossRef]
6. X.-L. Liu, S.-X. Ma, S.-W. Zhu, Y. Zhao, X.-J. Ning, L. Zhao, and J. Zhuang, “Light stimulated and regulated gas sensing ability for ammonia using sulfur-hyperdoped silicon,” Sens. Actuators, B 291(C), 345–353 (2019). [CrossRef]
7. N. K. Ali, M. R. Hashim, and A. A. Aziz, “Effects of surface passivation in porous silicon as H2 gas sensor,” Solid-State Electron. 52(7), 1071–1074 (2008). [CrossRef]
8. J. Salonen, M. Kaasalainen, O.-P. Rauhala, L. Lassila, M. Hakamies, T. Jalkanen, R. Hahn, P. Schmuki, and E. Makila, “Thermal carbonization of porous silicon: The current status and recent applications,” ECS Trans. 69(2), 167–176 (2015). [CrossRef]
9. B. Timmer, W. Olthuis, and A. van den Berg, “Ammonia sensors and their applications—a review,” Sens. Actuators, B 107(2), 666–677 (2005). [CrossRef]
10. X. Liu, P. R. Coxon, M. Peters, B. Hoex, J. M. Cole, and D. J. Fray, “Black silicon: fabrication methods, properties and solar energy applications,” Energy Environ. Sci. 7(10), 3223–3263 (2014). [CrossRef]
11. J. E. Carey, C. H. Crouch, M. Shen, and E. Mazur, “Visible and near-infrared responsivity of femtosecond-laser microstructured silicon photodiodes,” Opt. Lett. 30(14), 1773–1775 (2005). [CrossRef]
12. C. H. Crouch, J. E. Carey, J. M. Warrender, M. J. Aziz, E. Mazur, and F. Y. Génin, “Comparison of structure and properties of femtosecond and nanosecond laser-structured silicon,” Appl. Phys. Lett. 84(11), 1850–1852 (2004). [CrossRef]
13. C. H. Crouch, J. E. Carey, M. Shen, E. Mazur, and F. Y. Génin, “Infrared absorption by sulfur-doped silicon formed by femtosecond laser irradiation,” Appl. Phys. A 79(7), 1635–1641 (2004). [CrossRef]
14. G. Korotcenkov, Handbook of Gas Sensor Materials: Properties, Advantages and Shortcomings for Applications (Springer, 2014).
