Rheotaxially Grown and Vacuum Oxidized SnOx Nanolayers for NO2 Sensing Characteristics at ppb Level and Room Temperature

This work presents, for the very first time, very promising nitrogen dioxide (NO2) sensing characteristics of SnOx nanolayers obtained by the innovative and unique rheotaxial growth and vacuum oxidation (RGVO) processing technique. The NO2 gas sensing experiments were performed using the novel surface photovoltage gas sensing device. The measured detection limit at room temperature (RT) is as low as 10 ppb NO2 in synthetic air, whereas the detection limit calculated on the basis of signal to noise ratio is around 6 ppb NO2. For the complementary study of surface chemistry of RGVO SnOx nanolayers, including nonstoichiometry, presence of carbon contamination and surface bondings, the X-ray photoelectron spectroscopy (XPS) method was applied. The SnOx RGVO samples reveal nonstoichiometry because the relative concentration [O]/[Sn] equals 0.94 for the as deposited sample and increases upon subsequent air exposure and NO2 sensing. Moreover, carbon contamination has been recognized after exposing the RGVO SnOx nanolayers to the air and during the NO2 detection.


Introduction
Nitrogen dioxide NO 2 can be considered as one of the main pollutants of the environment induced by industrial development in modern society. It can be harmful to living organisms not only by breathing in its vapors leading directly to serious illnesses of the airways, especially dangerous in the case of people suffering from asthma [1][2][3], but also by its indirect destructive impact on environment, including among others formation of acid rain and near ground level ozone [4,5]. In the case of NO 2 , exposure to a concentration larger than 1 ppm can lead to serious illnesses of the human respiratory system or aggravation of existing afflictions such as, among others, bronchitis, emphysema and lung insufficiency as well as worsen the medical condition of the circulatory system [6].
Having these dramatic consequences in mind, the issue of monitoring NO 2 concentration, especially in urban areas, has recently became highly important. According to the current regulations of the European Parliament on ambient air quality and cleaner air in Europe [7] the limit value of NO 2 concentration for an exposure of no longer than 1 h not exceeded more than 18 times a calendar year is 200 µg/m 3 , whereas the limit value for NO 2 concentration in the case of the constant exposure in the calendar year cannot exceed 40 µg/m 3 .
The issue of monitoring NO 2 can also have diagnostic applications, since it appears that the increase of its level in exhaled breath of people having asthma foreshadows an asthma attack and can also be used to identify respiratory system infections [8,9].

Materials and Methods
SnO x nanolayers were obtained using the rheotaxial growth and vacuum oxidation (RGVO) technique, the technique being our unique modification of rheotaxial growth and thermal oxidation (RGTO) method [32], which was described recently in detail in [28,29]. The samples with the thickness of 20 nm, controlled with quartz microbalance, were deposited at the Si (100) substrate by evaporation of Sn from the ceramic source under vacuum conditions related to 10 −4 mbar of oxygen partial pressure. Additionally, these nanolayers were oxidized in situ at 400 • C with an oxygen partial pressure of 10 −2 mbar for 2 h in order to increase their stoichiometry.
X-ray photoelectron spectroscopy (XPS) measurements were performed using a SPECS XPS spectrometer operating with Al Kα lamp (XR-50 source) and PHOIBOS-100 hemispherical analyzer. The XPS spectra of our RGVO SnO x nanolayers registered in the various modes (survey, windows and lines) have been additionally calibrated with respect to reference binding energies (BE) using both the XPS Au4f peak at 84.5 eV, as well as the XPS C1s peak at 284.5 eV of residual C contamination, being always at the surface of all our samples under investigation.
The gas sensing experiments of our RGVO SnO x nanolayers towards NO 2 were performed using a novel type surface photovoltage gas sensor device operated at room temperature and described in detail in [30,31]. All the gas sensing measurements were performed with the total gas flow rate of 50 mL/min with the relative NO 2 gas concentration in the synthetic air ranging from 10-500 ppb. Figure 1 demonstrates the variation of amplitude of the surface photovoltage (SPV) signal of the RGVO SnO x nanolayers after exposure to sequential relative concentration of NO 2 in synthetic air in the range of 10-500 ppb.

Materials and Methods
SnOx nanolayers were obtained using the rheotaxial growth and vacuum oxidation (RGVO) technique, the technique being our unique modification of rheotaxial growth and thermal oxidation (RGTO) method [32], which was described recently in detail in [28,29]. The samples with the thickness of 20 nm, controlled with quartz microbalance, were deposited at the Si (100) substrate by evaporation of Sn from the ceramic source under vacuum conditions related to 10 −4 mbar of oxygen partial pressure. Additionally, these nanolayers were oxidized in situ at 400 °C with an oxygen partial pressure of 10 −2 mbar for 2 h in order to increase their stoichiometry.
X-ray photoelectron spectroscopy (XPS) measurements were performed using a SPECS XPS spectrometer operating with Al Kα lamp (XR-50 source) and PHOIBOS-100 hemispherical analyzer. The XPS spectra of our RGVO SnOx nanolayers registered in the various modes (survey, windows and lines) have been additionally calibrated with respect to reference binding energies (BE) using both the XPS Au4f peak at 84.5 eV, as well as the XPS C1s peak at 284.5 eV of residual C contamination, being always at the surface of all our samples under investigation.
The gas sensing experiments of our RGVO SnOx nanolayers towards NO2 were performed using a novel type surface photovoltage gas sensor device operated at room temperature and described in detail in [30,31]. All the gas sensing measurements were performed with the total gas flow rate of 50 mL/min with the relative NO2 gas concentration in the synthetic air ranging from 10-500 ppb. Figure 1 demonstrates the variation of amplitude of the surface photovoltage (SPV) signal of the RGVO SnOx nanolayers after exposure to sequential relative concentration of NO2 in synthetic air in the range of 10-500 ppb.

Results and Discussion
As it can be concluded, exposing the SnOx RGVO nanolayer to NO2 induces a significant increase in the SPV value. As the gas flow of NO2 drops to zero, the baseline of the SPV signal is recovered. The time of response, tresp, defined as the period required to reach 90% of the final signal change, is in the order of a dozen minutes within the whole NO2 concentration range, for example in the case of the interaction with 500 ppb NO2 it equals (14 ± 2) min, 250 ppb: (15 ± 2) min, 40 ppb: (14 ± 2) min. The time of recovery is still rather long. However, it can be improved in the future by applying additional procedures to accelerate surface regeneration, such as infra-red illumination.  As it can be concluded, exposing the SnO x RGVO nanolayer to NO 2 induces a significant increase in the SPV value. As the gas flow of NO 2 drops to zero, the baseline of the SPV signal is recovered. The time of response, t resp , defined as the period required to reach 90% of the final signal change, is in the order of a dozen minutes within the whole NO 2 concentration range, for example in the case of the interaction with 500 ppb NO 2 it equals (14 ± 2) min, 250 ppb: (15 ± 2) min, 40 ppb: (14 ± 2) min. The time of recovery is still rather long. However, it can be improved in the future by applying additional procedures to accelerate surface regeneration, such as infra-red illumination.
In the case of the exposure to 20 and 10 ppb NO 2 (see Figure 1), the measurement was repeated in order to examine the short-time stability of the sensor response, which appears to present a satisfyingly good level, as the variation of amplitude of the SPV gas sensor signal for both 20 and 10 ppb NO 2 obtained in the second step reaches the same value as the corresponding measurement performed previously. Moreover, as one can see, the variation of amplitude of the SPV gas sensor signal decreases as the RGVO SnO x sample faces lower concentrations of NO 2 (see Figure 2). In the case of the exposure to 20 and 10 ppb NO2 (see Figure 1), the measurement was repeated in order to examine the short-time stability of the sensor response, which appears to present a satisfyingly good level, as the variation of amplitude of the SPV gas sensor signal for both 20 and 10 ppb NO2 obtained in the second step reaches the same value as the corresponding measurement performed previously. Moreover, as one can see, the variation of amplitude of the SPV gas sensor signal decreases as the RGVO SnOx sample faces lower concentrations of NO2 (see Figure 2). As it can be seen in Figures 1 and 2 the detection limit for the RGVO SnOx nanolayer is lower than 10 ppb NO2. In order to discuss theoretically the smallest amount of NO2 that could be detected, the signal to noise ratio was taken into account according to the International Union of Pure and Applied Chemistry recommendations [33], which specify that for reliable measurement the signal to noise has to be larger than 3. The procedure applied within this study, previously proposed by Li et al. [34] and successfully applied in literature [21,[35][36][37] is given by Equation (1): where rmsnoise denotes root mean square deviation between experimental data within baseline region, yi, and values fitted using polynomial function, y; slope corresponds to a coefficient in linear function: = + used for fitting the sensor response, ΔSPV, as a function of the gas concentration (see Figure 2); whereas, N denotes the number of data points taken into account for fitting-in this work N = 10.
On the basis of the procedure described above, it appears that the theoretical detection limit for NO2 recognition in the case of a SnOx RGVO nanostructure is around 6 ppb. Undoubtedly, this result is unique for pure SnOx sensing material working at room temperature.
In order to examine long term stability of RGVO SnOx nanolayers, some of the gas sensing experiments were repeated. Figure 3 presented below demonstrates the response towards 120 ppb of NO2 obtained initially and after six days. As it can be concluded, in the case of 120 ppb NO2 the As it can be seen in Figures 1 and 2 the detection limit for the RGVO SnO x nanolayer is lower than 10 ppb NO 2 . In order to discuss theoretically the smallest amount of NO 2 that could be detected, the signal to noise ratio was taken into account according to the International Union of Pure and Applied Chemistry recommendations [33], which specify that for reliable measurement the signal to noise has to be larger than 3. The procedure applied within this study, previously proposed by Li et al. [34] and successfully applied in literature [21,[35][36][37] is given by Equation (1): where rms noise denotes root mean square deviation between experimental data within baseline region, y i , and values fitted using polynomial function, y; slope corresponds to a coefficient in linear function: y = ax + b used for fitting the sensor response, ∆SPV, as a function of the gas concentration (see Figure 2); whereas, N denotes the number of data points taken into account for fitting-in this work N = 10.
On the basis of the procedure described above, it appears that the theoretical detection limit for NO 2 recognition in the case of a SnO x RGVO nanostructure is around 6 ppb. Undoubtedly, this result is unique for pure SnO x sensing material working at room temperature.
In order to examine long term stability of RGVO SnO x nanolayers, some of the gas sensing experiments were repeated. Figure 3 presented below demonstrates the response towards 120 ppb of Sensors 2020, 20, 1323 5 of 12 NO 2 obtained initially and after six days. As it can be concluded, in the case of 120 ppb NO 2 the variation of amplitude of surface photovoltage (SPV) reaches a value in the range of 32-35 mV, which means that in both cases the sensor response is stable and repeatable. However, the recovery is noticeably faster for the measurement repeated after 6 days. This can be attributed to the real life, dynamic conditions that can affect the RGVO SnO x nanolayers' recovery. In the case of very low concentrations of NO 2 in the ppb range, the sensor becomes significantly sensitive to the surrounding temperature and humidity. From this point of view, as it was mentioned above, one can consider applying some additional procedures, e.g., illuminating with IR radiation in order to speed up regeneration.
Sensors 2020, 20, 1323 5 of 12 variation of amplitude of surface photovoltage (SPV) reaches a value in the range of 32-35 mV, which means that in both cases the sensor response is stable and repeatable. However, the recovery is noticeably faster for the measurement repeated after 6 days. This can be attributed to the real life, dynamic conditions that can affect the RGVO SnOx nanolayers' recovery. In the case of very low concentrations of NO2 in the ppb range, the sensor becomes significantly sensitive to the surrounding temperature and humidity. From this point of view, as it was mentioned above, one can consider applying some additional procedures, e.g., illuminating with IR radiation in order to speed up regeneration. The issue of surface reactions in the case of SnOx based gas-sensing material interacting with reducing and oxidizing gases together with the theoretical description linking the measured characteristics with the change in electron work function have been discussed in detail in [38][39][40][41].
The main limitation of work function variation defined as the contact potential difference, CPD, (measured using mainly the Kelvin probe approach) for the potential gas sensing application is its relatively poor sensitivity, related to the observed low values of the signal to noise ratio, as reviewed by Korotcenkov et al. [42]. This can be improved using the surface photovoltage effect SPV [43] which consists in measuring the variation of surface potential ΔVS, upon illumination at the well-defined constant radiation intensity Io, related to the charge redistribution appearing after photon induced electron-hole pair generation. The variation of SPV can be defined as: In our case the gas sensing mechanism is governed by the separation of charge carriers based on significant differences in their mobilities, which lead to the relatively large variation of electric potential within the space charge layer (SCL) observed finally as the variation of surface potential ΔVS. The fundamentals of the SPV technique were also briefly mentioned in our previous papers [30,31].
From our experience it appears that the observed ΔSPV values can even be in the range of hundreds mV. The variation of ΔSPV can be easily interpreted on the basis of the interaction of gas molecules with the surface of our gas sensing material (RGVO SnOx). It is commonly known that at low temperatures (below 150 °C) oxygen ions adsorb at the surface of metal oxide semiconductors in a form of O2 − [44] according to the equation: The issue of surface reactions in the case of SnO x based gas-sensing material interacting with reducing and oxidizing gases together with the theoretical description linking the measured characteristics with the change in electron work function have been discussed in detail in [38][39][40][41].
The main limitation of work function variation defined as the contact potential difference, CPD, (measured using mainly the Kelvin probe approach) for the potential gas sensing application is its relatively poor sensitivity, related to the observed low values of the signal to noise ratio, as reviewed by Korotcenkov et al. [42]. This can be improved using the surface photovoltage effect SPV [43] which consists in measuring the variation of surface potential ∆V S , upon illumination at the well-defined constant radiation intensity Io, related to the charge redistribution appearing after photon induced electron-hole pair generation. The variation of SPV can be defined as: In our case the gas sensing mechanism is governed by the separation of charge carriers based on significant differences in their mobilities, which lead to the relatively large variation of electric potential within the space charge layer (SCL) observed finally as the variation of surface potential ∆V S . The fundamentals of the SPV technique were also briefly mentioned in our previous papers [30,31].
From our experience it appears that the observed ∆SPV values can even be in the range of hundreds mV. The variation of ∆SPV can be easily interpreted on the basis of the interaction of gas molecules with the surface of our gas sensing material (RGVO SnO x ). It is commonly known that at low temperatures (below 150 • C) oxygen ions adsorb at the surface of metal oxide semiconductors in a form of O 2 − [44] according to the equation: Sensors 2020, 20, 1323 6 of 12 leading in the case of SnO 2 to the upward band bending due to the electrons' trapping and depletion layer's formation. In addition to the above, it is also generally accepted that oxidizing gases like NO 2 adsorb at the surface of metal oxide semiconductor materials in an ionic form: (4) according to Cho et al. [45] adsorption of NO 2 competes with that of oxygen, which is given by Equation'(3).
Both O 2 − as well as NO 2 − presence leads to the final surface charge density. The adsorption of NO 2 − which is considerably stronger than that of O 2 − , according to [45], results in the further increase of the surface potential barrier qV S . In the case of semiconductors, the work function Φ is given as: where (E C -E F ) b denotes the difference between the energy of conduction band and the Fermi level in the bulk, χ is an electron affinity. In general, all the given components can change upon the interaction between the semiconductor surface and the gas phase. However, in our case both (E C -E F ) b as well as the χ parameters in Equation (5) can be treated as constant, as no bulk changes take place. What is crucial is that in our case the gas sensing mechanism is governed by significant changes in ∆V S promoted by the illumination, as described above.
In order to study surface chemical properties of RGVO SnO x nanolayers, crucial for their gas sensing characteristics, X-ray photoelectron spectroscopy was applied. Figure 4 presents the XPS survey spectra for the as deposited sample, after air exposure as well as after subsequent NO 2 sensing experiments.
Sensors 2020, 20, 1323 6 of 12 leading in the case of SnO2 to the upward band bending due to the electrons' trapping and depletion layer's formation.
In addition to the above, it is also generally accepted that oxidizing gases like NO2 adsorb at the surface of metal oxide semiconductor materials in an ionic form: according to Cho et al. [45] adsorption of NO2 competes with that of oxygen, which is given by Equation (3).
Both O2 − as well as NO2 -presence leads to the final surface charge density. The adsorption of NO2 − which is considerably stronger than that of O2 − , according to [45], results in the further increase of the surface potential barrier qVS.
In the case of semiconductors, the work function Φ is given as: where (EC -EF)b denotes the difference between the energy of conduction band and the Fermi level in the bulk, χ is an electron affinity. In general, all the given components can change upon the interaction between the semiconductor surface and the gas phase. However, in our case both (EC -EF)b as well as the χ parameters in Equation (5) can be treated as constant, as no bulk changes take place. What is crucial is that in our case the gas sensing mechanism is governed by significant changes in ΔVS promoted by the illumination, as described above.
In order to study surface chemical properties of RGVO SnOx nanolayers, crucial for their gas sensing characteristics, X-ray photoelectron spectroscopy was applied. Figure 4 presents the XPS survey spectra for the as deposited sample, after air exposure as well as after subsequent NO2 sensing experiments.  As can be clearly seen, the contribution from Sn and O is observed for the pristine and for both the air as well as NO 2 exposed nanolayer. In the case of the sample which underwent NO 2 detection, an evident carbon presence at the surface can be observed. Having in mind that carbon undesired contamination is crucial for subsequent gas sensing characteristics, the detailed XPS analysis of C1s spectral windows was applied (see Figure 5). As it can be concluded on the basis of Figure 5a, the RGVO SnO x nanolayers elaboration procedure applied within this work does not trigger unwanted carbon contamination, as on the basis of the signal to noise ratio the contribution from C1s in this case is not observed. This fact undoubtedly can be interpreted as a great advantage of rheotaxial growth and the vacuum oxidation method. In the case of the RGVO SnO x nanolayer which underwent air exposure (Figure 5b), a small contribution of carbon on the surface can be recognized and attributed to CO and CO 2 adsorbed from the surrounding atmosphere [46,47]. For the sample after NO 2 gas sensing experiments ( Figure 5c) the amount of C increases. The deconvolution procedure of C1s spectral line shows that carbon present on the surface in this case comes from CO and CO 2 (C-O component) as well as hydroxyl groups originating from dissociated water vapor [46,47].
Sensors 2020, 20, 1323 7 of 12 As can be clearly seen, the contribution from Sn and O is observed for the pristine and for both the air as well as NO2 exposed nanolayer. In the case of the sample which underwent NO2 detection, an evident carbon presence at the surface can be observed. Having in mind that carbon undesired contamination is crucial for subsequent gas sensing characteristics, the detailed XPS analysis of C1s spectral windows was applied (see Figure 5). As it can be concluded on the basis of Figure 5a, the RGVO SnOx nanolayers elaboration procedure applied within this work does not trigger unwanted carbon contamination, as on the basis of the signal to noise ratio the contribution from C1s in this case is not observed. This fact undoubtedly can be interpreted as a great advantage of rheotaxial growth and the vacuum oxidation method. In the case of the RGVO SnOx nanolayer which underwent air exposure (Figure 5b), a small contribution of carbon on the surface can be recognized and attributed to CO and CO2 adsorbed from the surrounding atmosphere [46,47]. For the sample after NO2 gas sensing experiments ( Figure 5c) the amount of C increases. The deconvolution procedure of C1s spectral line shows that carbon present on the surface in this case comes from CO and CO2 (C-O component) as well as hydroxyl groups originating from dissociated water vapor [46,47]. In the second step for the more precise and quantitative analysis of XPS results, O1s -Sn3d as well as C1s spectral windows were used in order to calculate the relative concentration of the main components: [O]/[Sn] and [C]/[Sn], based on the atomic sensitivity factor (ASF) approach [48] and the procedure described in detail in our previous papers [49,50]. The results of this analysis are given in Table 1. As can be seen, the relative concentration [C]/[Sn] for the air exposed sample is as low as In the second step for the more precise and quantitative analysis of XPS results, O1s -Sn3d as well as C1s spectral windows were used in order to calculate the relative concentration of the main components: [O]/[Sn] and [C]/[Sn], based on the atomic sensitivity factor (ASF) approach [48] and the procedure described in detail in our previous papers [49,50]. The results of this analysis are given in Table 1. As can be seen, the relative concentration [C]/[Sn] for the air exposed sample is as low as 0.08, Sensors 2020, 20, 1323 8 of 12 whereas in the case of SnO x it equals 3.96 after NO 2 sensing. Perhaps this is related to the fact that small NO 2 molecules promote hydroxyl group adsorption at the surface of our samples. In the subsequent step the XPS O1s and Sn3d 5/2 spectral lines were decomposed as can be seen in Figure 6. In the case of the as deposited sample, the predominant contribution of Sn 2+ is observed both for the Sn3d 5/2 as well as O1s decomposed lines (see Figure 6a,b). For O1s core line it appears that the two components related to O-Sn 2+ (at 530.4 eV) and O-Sn 4+ (at 531.0 eV) can be recognized. As far as the Sn3d 5/2 line is discussed, one can conclude that the pristine sample contains also a small amount of metallic tin Sn 0 .
Exposing the RGVO SnO x nanolayers to the air leads to some modifications in the chemical properties of their surface, being still a mixture of SnO 2 and SnO as the contribution from the latter one decreases. Based on the deconvolution of Sn3d 5/2 line (see Figure 6c) the contribution of Sn 2+ (at 486.6 eV), Sn 4+ (at 487.0 eV) and Sn 0 (at 484.3 eV) can be recognized. As for the deconvolution of O1s (see Figure 6d), the two constituents are observed, i.e., O-Sn 2+ (at 530.8 eV) and O-Sn 4+ (at 531.4 eV). These results remain in good agreement with our previous paper [29].
In the case of the sample which underwent NO 2 exposure, the O1s spectral line (Figure 6f) can be decomposed into three components attributed to O-Sn 2+ (at 530.8 eV), O-Sn 4+ (at 532.1 eV) and strong carbon contamination O=C or C-OH (at 534.7 eV). Decomposition of the Sn3d 5/2 spectral line ( Figure 6e) still reveals the impact of both Sn 2+ (at 486.4 eV) and Sn 4+ (at 487.0 eV). However, there is no contribution from metallic tin Sn 0 .

Conclusions
Within this study the novel RGVO SnOx nanolayers were examined for possible NO2 detection at room temperature using the surface photovoltage effect. This gas sensing material is very promising in terms of gas detection because the experimental detection limit at room temperature is as low as 10 ppb NO2. In turn, the theoretical detection limit calculated on the basis of signal to noise ratio equals 6 ppb NO2. This means that our novel RGVO technique enables the obtaining of the promising gas sensor material, being a mixture of tin oxide SnO and tin dioxide SnO2, without undesired carbon contamination. This is very promising in terms of improving NO2 gas sensing characteristics. However, as it is generally, the C unwanted surface species usually obstruct the interaction between the semiconductor active surface and the gas under detection. In our experiments

Conclusions
Within this study the novel RGVO SnO x nanolayers were examined for possible NO 2 detection at room temperature using the surface photovoltage effect. This gas sensing material is very promising in terms of gas detection because the experimental detection limit at room temperature is as low as 10 ppb NO 2 . In turn, the theoretical detection limit calculated on the basis of signal to noise ratio equals 6 ppb NO 2 . This means that our novel RGVO technique enables the obtaining of the promising gas sensor material, being a mixture of tin oxide SnO and tin dioxide SnO 2 , without undesired carbon contamination. This is very promising in terms of improving NO 2 gas sensing characteristics. However, as it is generally, the C unwanted surface species usually obstruct the interaction between the