Selective Detection of H2 Gas in Gas Mixtures Using NiO‐Shelled Pd‐Decorated ZnO Nanowires

Hydrogen (H2) gas is a green fuel, but its leakage during storage and transportation can lead to disasters due to its explosive nature. Here, a sensitive and selective H2 gas sensor is developed that can detect H2 in H2/CO and H2/NO2 gas mixtures. First, ZnO nanowires (NWs) are grown using vapor–liquid–solid growth. This is followed by atomic layer deposition‐mediated growth of Pd nanoparticles (NPs) on the ZnO NWs and uniform deposition of a thin NiO shell layer (12 nm in thickness) over the Pd‐decorated ZnO NWs. Characterization of the synthesized samples by different methods confirms the desired chemical composition, morphology, and phases. H2 gas sensing studies reveals the highly sensitive and selective response of the optimized gas sensor to H2 at 200 °C. In the presence of H2/CO and H2/NO2 gas mixtures, the NiO‐shelled Pd‐decorated ZnO NW sensor displays good selectivity for H2, but not the Pd‐decorated ZnO NW gas one. The NiO‐shelled Pd‐decorated ZnO NW gas sensor efficiently detected H2 also in the presence of 40% relative humidity and displays good stability even after 1 month. The present results can open the doors to the fabrication of highly selective H2 gas sensors using the described rationale design.


Introduction
Hydrogen (H 2 ) gas is considered a next-generation green fuel: it is renewable, releases a significant amount of heat during combustion, and does not produce toxic emissions.Currently, H 2 gas able to reliably detect H 2 gas in a mixture of H 2 /CO or H 2 /NO 2 gases. [3]hemiresistive gas sensors with metal oxides are particularly interesting due to their high response, high-stability, fast dynamics, ease of fabrication, and operation, and low-price.However, metal oxide gas sensors operate at high temperatures (150-450°C) and often display poor selectivity. [4]In general, the sensing performance of n-type (e.g.SnO 2 and ZnO) gas sensors is better than that of p-type metal oxide gas sensors, mostly linked to their high electron mobility. [5]Among n-type metal oxides, ZnO is a promising sensor material due to its stability, high electron mobility, desirable band gap (E g = 3.6 eV), ease of synthesis, availability and low price.However, like other metal oxide gas sensors, it exhibits poor selectivity in the pristine form because it shows similar responses to most gases. [6]Various approaches have been investigated to improve the selectivity of ZnO and other metal oxide gas sensors: metal doping, [7] noble metal decoration, [8] n-n heterojunction formation, [9] n-p heterojunction formation, [10] and use of filters. [11]oble metal decoration can increase the sensing performance, especially in term of selectivity, through two mechanisms.The first is related to the catalytic effect of noble metals whereby the incoming gas molecules become easily adsorbed on the noble metal and are then dissociated into smaller ions that can move to the surface of neighboring metal oxide.For example, it is wellaccepted that Pd nanoparticles (NPs) can dissociate H 2 molecules to atomic H and increase their reactions with adsorbed oxygen ions. [12]The second mechanism is related to the formation of heterojunctions between noble metals and metal oxides, leading to extensive resistance modulation. [13]1a,d,14] However, the noble metal nanoparticles (NPs) used to decorate the surface of metal oxides may be easily poisoned in the presence of certain gases.1d,15] This may affect their catalytic activity and limit the metal NP role as promising agents to enhance the gas sensing performance.Furthermore, as gas sensors operate at high temperatures, ultrafine noble metals might be oxidized in air, thus decreasing the sensing performance, and noble metals might agglomerate on the sensor surface. [16]Therefore, it is essential to find a way to protect the noble metals from direct exposure to air in order to enhance the sensing properties.The deposition of a thin metal oxide layer over the decorated NPs could be a promising protective strategy, which has been less studied for ternary systems.In this study, we used NiO as deposition layer over the Pd-decorated ZnO NWs.In addition, to prevent Pd poisoning, p-type NiO is interesting because it displays very good structural stability, good oxygen adsorption, and strong catalytic activity. [5]In the absence of Pd decoration of the ZnO surface, NiO is in direct contact with ZnO.Intimate contacts between n-ZnO and p-NiO can provide numerous heterojunctions with potential barriers to flow of charges in air.Upon exposure to the target gas, the barrier height changes, leading to the sensor resistance modulation. [17]Previous studies described enhanced performance of ZnO-NiO heterojunctions for gas sensing applications compared with their pristine counterparts. [18]Moreover, as NiO is deposited as a continuous layer, a core-shell structure is expected where ZnO and NiO are in direct contact.This should maximize the contact area between components, leading to significant resistance modulation. [19]On the other hand, when Pd NPs are placed between two layers of ZnO and NiO, ZnO-Pd-NiO heterojunctions are formed. [20]Modulation of the barriers formed between these materials can strongly influence the sensing mechanism.
Atomic layer deposition (ALD) is a reliable deposition technique and it is perfectly suitable for coating core materials with conformal and uniform thin layers. [21]ALD allows controlling the Pd NP thickness at the sub-nanometer scale and also their relatively narrow size distribution. [22]NiO and Pd have already been grown by ALD. [23]Pd/Ni catalysts deposited on nanoporous Al 2 O 3 display good activity in formic acid electrooxidation. [24]oreover, Pd deposition on Ni/NiO nanofibers shows electrochemical hydrogen and oxygen evolution reactions. [25]More recently, NiO layer deposition on porous In 2 O 3 has been used to boost NO 2 detection. [26]ven though there are some papers related to ZnO─NiO, Pd─ZnO, and Pd─NiO gas sensors, their combinations in form of NiO-shelled Pd-decorated ZnO nanowires have not been reported in literature.Furthermore, detection of H 2 gas in a gas mixture is rarely reported in literature.Therefore, in the present study combination of NiO─Pd and ZnO in form of NiO-shelled Pd-decorated ZnO nanowires was used for successful sensing of H 2 gas in gas mixtures.In this work, first ZnO NWs were easily synthesized by vapor-liquid-solid (VLS) growth as previously described. [27]Then, Pd NPs were decorated over the synthesized ZnO NWs using ALD.This was followed by deposition of a NiO shell (12 nm in thickness) over the Pd-decorated ZnO NWs.As expected, this work demonstrated the beneficial effect of Pd NPs on the ZnO NW sensitivity.Although the NiO shell weakened the raw response of the device, it significantly increased its selectivity when H 2 was mixed with CO and NO 2 .These properties demonstrate the interest of such gas sensor architecture for use in realistic conditions.

Morphological and Chemical Studies
Figure 1a shows the ZnO NWs grown over the tri-layered interdigital electrode (TIE) substrate and Figure 1b a scanning electron microscopy (SEM) image of ZnO NWs.As ZnO NW size depends on the initial size of the as-formed Au-droplets, ZnO NWs with different sizes were synthesized, but with a narrow size distribution.The diameter of a ZnO NW was > 70 nm (Figure 1c). Figure 1d shows a single ZnO NW uniformly and abundantly covered by Pd NPs after ALD.
High-resolution transmission electron microscopy (TEM) (Figure 2a) showed that bare ZnO NWs had a crystalline structure, with spacings between the parallel fringes of 0.519 and 0.273 nm, attributed to the {0001} and {01-10} ZnO planes, respectively.TEM analysis of the NiO shell over a Pd-decorated ZnO NW (Figure 2b) showed a uniform coating by the NiO shell thanks to ALD deposition.The spotty selected-area electron diffraction (SAED) pattern of bare ZnO (Figure 2c) showed the single crystalline nature of bare ZnO NWs.In the SAED pattern of Ni-shelled Pd-decorated ZnO NWs (Figure 2d), the spotty pattern related to ZnO and the ring patterns related to NiO and Pd with a polycrystalline nature could be seen.The weak intensity of the circles may be explained by the thin layer thickness or the presence of small imbedded crystallites.The white circles corresponded to Pd NP diffraction.A thorough analysis revealed that the circles were composed of small spots coming from each monocrystalline NP.TEM images of ZnO NWs after deposition of Pd NPs and NiO thin film (Figure 2e,f) showed that the NiO shell completely covered the Pd-decorated ZnO NWs, which had a diameter between 60 and 120 nm.In addition, Moiré fringes (white arrows in Figure 2f) demonstrated the superposition of lattice planes.
Energy-dispersive X-ray spectroscopy (EDS) chemical analysis of a single NiO shelled Pd-decorated ZnO NW during SEM (Figure 3a) showed the presence of the peaks related to Zn, Ni, Pd, and O in addition to those related to the Si substrate and the usual carbon contamination (Figure 3b).EDS mapping confirmed the presence of Zn, Ni, and Pd elements (Figure 3c-e).Oxygen mapping is not shown because oxygen was detected on the entire scrutinized area.Zn and Ni were uniformly distributed, confirming the full coverage of the ZnO NW by the NiO shell.Distinct Pd NPs could not be observed in Figure 3e due to the high particle density and because the Pd signal had to pass through the NiO film.
The chemical states of NiO-shelled Pd-decorated ZnO NWs were analyzed by X-ray photoelectron spectroscopy (XPS) (Figure 4).XPS peaks and some Auger electron transitions of the constituting elements (Zn, Ni, Pd, and O) were detected on the survey spectrum (Figure 4a) and the peaks related to Si (substrate) and C (adventitious carbon).The Zn 2p spectrum  (Figure 4b) included a single contribution (Zn─O), with a BE of 1022.1 eV (Zn 2p 3/2 ) and a spin-orbit-splitting (SOS) of 23.2 eV.This corresponds to ZnO in the literature. [28]Figure 4c shows Ni 2p 3/2 contribution and its satellite peaks.For clarity, the 2p 1/2 area is not shown, but its position relative to 2p 3/2 corresponded to the expected SOS for NiO (18.4 eV). [28]As previously reported, [29] the 2p 3/2 peak was fitted with two contributions (BE = 854.1 and 855.9 eV), both assigned to NiO and hydroxylated Ni (the latter located mainly at the surface).Additional contributions were measured at BE = 861.0 and 863.5 eV, and were attributed to the satellite peaks.The Pd 3d 5/2 and 3d 3/2 peaks (Figure 4d) had the expected SOS = 5.2 eV. [28,30]The Pd 3d 5/2 and 3d 3/2 peaks exhibited two contributions: i) BE = 335.2for Pd 3d 5/2 and BE = 340.4eV for 3d 1/2 (low energy contribution), assigned to metallic palladium (Pd 0 ) and ii) BE = 336.6 for Pd 3d 5/2 and BE = 341.8eV for 3d 1/2 (high energy contribution), attributed to Pd 2+ [30,31] and corresponding to Pd bonded to the O of NiO.The O 1s peak (Figure 4e) also included two contributions: i) BE = 529.6 eV ascribed to O -II bonded to a metal, such as Ni. [32]owever, as Zn was detected, a small contribution can also arise from oxygen bonded to Zn from the ZnO NWs underneath the NiO layer [33] ; and ii) BE = 531.9eV that displayed the largest intensity was attributed to several contributions, particularly O bonded to Ni 2[32] and Pd 2 . [30]This component was also related to superficial contamination, [34] such as hydroxyl groups.Lastly, the C 1s region (Figure 4f) corresponded to adventitious carbon impurities and presented three contributions corresponding to C─C (BE = 284.8eV), C─O (BE = 286 eV), and C═O (BE = 288.5 eV). [34,35]

Gas Sensing Studies
First, the fabricated gas sensors (ZnO NWs, Pd-decorated ZnO NWs, and NiO-shelled Pd-decorated ZnO NWs) were exposed to various concentrations of H 2 gas (1, 10, 100, and 500 ppm) at different temperatures in order to identify their optimal sensing temperature.The dynamic resistance curves (Figure 5a-c) showed that all sensors had an n-type behavior, in which their resistances decreased upon injection of H 2 gas with a reducing nature.This indicates that the behavior of these sensors is mostly governed by ZnO.Moreover, the gas sensor responses were strongly influenced by the temperature (Figure 5d-f).Specifically, with the pristine ZnO NW gas sensor, no sensing signal was detected at 25 and 50°C.At 100 °C, it showed a very low response to 100 and 500 ppm H 2 gas.The response gradually increased with the increasing temperatures and the maximum response was observed at 250 °C.However, this response was not high (e.g.6.1 to 500 ppm H 2 gas at 250 °C).Conversely, the Pd-decorated ZnO NW sensor showed a low response to H 2 gas even at room temperature.Its response progressively increased up to 200 °C (41.01 to 500 ppm H 2 gas) and was strongly decreased to 250 °C.The NiO-shelled Pd-decorated ZnO NW gas sensor showed a similar response profile: the response increased up to 200 °C (16.12 to 500 ppm H 2 gas) and then decreased.The low response at low temperatures was due to the fact that there was not enough energy for H 2 to overcome the adsorption barrier.With temperatures above the optimal temperature, the desorption rate was higher than the adsorption rate and the gas response decreased again.At the optimal temperature, the adsorption was maximum and the highest response to H 2 gas was observed.Based on the above findings, the optimal sensing temperature was 200 °C for Pd-decorated ZnO NWs and NiO-shelled Pd-decorated ZnO NWs and the other tests were performed at this temperature.Due to the low response of the pristine ZnO NW gas sensor, it was not used for the next experiments.
In the next step, the selectivity for H 2 of Pd-decorated ZnO NW and NiO-shelled Pd-decorated ZnO NW gas sensors was studied by calculating the dynamic resistance curves in the presence of increasing concentrations of H 2 gas (1, 10, 100, and 500 ppm H 2 gas) alone or mixed with 500 ppm CO gas or NO 2 gas at 200 °C.Upon exposure of Pd-decorated ZnO NWs to pure H 2 , which is a reducing atmosphere, resistance decreased (Figure 6a).This demonstrated the n-type behavior of the sensor.Indeed, ZnO is an n-type metal oxide semiconductor due to the presence of oxygen vacancies.The response values were 2.14 (1 ppm H 2 gas), 4.61 (10 pm H 2 gas), 13.36 (100 ppm H 2 gas) and 41.01 (500 ppm H 2 gas).When H 2 gas at different concentrations was mixed with 500 ppm CO gas, the response values were 1.08, 1.17, 2.57, and 12.11 for 1, 10, 100, and 500 ppm H 2 gas, respectively.When H 2 gas at different concentrations (1, 10, 100, and 500 ppm) was mixed with 500 ppm NO 2 gas, the response values were 1, 1.05, 1.75, and 6.39, respectively.These data highlighted the significant response decrease in the presence of CO and NO 2 gases.When NiO-shelled Pd-decorated ZnO NWs were tested in the same conditions (Figure 6b), the sensor resistance decreased upon exposure to H 2 alone, indicating again the sensor n-type behavior.It also revealed that the sensing behavior was governed by ZnO n-type behavior, despite the presence of NiO, a p-type semiconductor.The response values were 1.84 (1 ppm H 2 gas), 2.32 (10 ppm H 2 gas), 8.29 (100 ppm H 2 gas), and 16.12 (500 ppm H 2 gas).When 1, 10, 100, and 500 ppm H 2 gas were mixed with 500 ppm CO gas and with 500 ppm NO 2 gas, the response values were 1.15, 1.31, 2.6, and 12.41 and 1.19, 1.39, 2.95, and 13.19,   respectively.NO 2 is an oxidizing compound that leads to the gas sensor resistance increase.However, with both Pd-decorated ZnO NWs and NiO-shelled Pd-decorated ZnO NWs, the sensor resistance decreased also when exposed to H 2 mixed with NO 2 gas.This indicated that the sensors are more sensitive to H 2 than NO 2 .
To better understand the sensing behavior of the two sensors toward the target gases, their responses to pure H 2 gas and mixed with 500 ppm CO or 500 ppm NO 2 gases were compared.Overall, the response of the Pd-decorated ZnO NW sensor (Figure 7a) significantly decreased upon exposure to the gas mixtures.For example, the response values were 41.01 to 500 ppm pure H 2 and 12.11 and 6.39 to 500 ppm H 2 + 500 ppm CO and 500 ppm H 2 + 500 ppm NO 2 gases.Conversely, the response of NiO-shelled Pddecorated ZnO NWs did not significantly change when exposed to gas mixtures (Figure 7b).Indeed, the response values were 16.12 (500 ppm H 2 gas), 12.41 (500 ppm H 2 + 500 ppm CO) and 13.19 (500 ppm H 2 + 500 ppm NO 2 ).Then, comparison of the responses of Pd-decorated ZnO NW and NiO shelled Pd-decorated ZnO NW gas sensors to various concentrations of H 2 mixed with 500 ppm NO 2 at 200 °C (Figure 7c) showed that the response of NiO-shelled Pd-decorated ZnO NWs was always higher than that of Pd-decorated ZnO NWs.For example, the response values of NiO-shelled Pd-decorated ZnO NWs and Pd-decorated ZnO NWs to 500 ppm H 2 and 500 ppm NO 2 were 13.19 and 6.39.A simi-lar trend was observed for all H 2 concentrations in the presence of 500 ppm CO (Figure 7d).For example, the response values of NiO-shelled Pd-decorated ZnO NWs and Pd-decorated ZnO NWs to the 500 ppm H 2 + 500 ppm CO mixture were 12.41 and 12.11, respectively.This showed again the superiority of the NiO-shelled Pd-decorated ZnO NW system compared with Pd-decorated ZnO NWs because it displayed similar responses to pure H 2 and to H 2 mixed with CO and NO 2 gases.
To assess repeatability, the two gas sensors were used for five sequential cycles (100 ppm H 2 and mixtures of 100 ppm H 2 with 500 ppm CO or 500 ppm NO 2 gas at 200 °C).The resistance values of the two sensors did not significantly change during the five gas sensing cycles (Figure 8a,c).To precisely monitor the sensing behaviors, the responses of Pd-decorated ZnO NWs and NiO-shelled Pd-decorated ZnO NWs were calculated in function of the cycle number (Figure 8b,d).The Pd-decorated sensor response slightly fluctuated during the five cycles.Conversely, the NiO-shelled Pd-decorated ZnO NW response remained almost constant during the five sequential sensing cycles.
In the next step, the responses of the two gas sensors to 100 ppm H 2 and 100 ppm H 2 mixed with 500 ppm CO or with 500 ppm NO 2 gas were measured in the presence of 0% and 40% relative humidity (RH) at 200 °C.The dynamic resistance curves of the Pd-decorated ZnO NW gas sensor (Figure 9a-c) showed that the initial resistance decreased in the presence of humidity.Indeed, the resistance of n-type semiconductors is affected by water vapors.The following reactions can take place between water molecules and adsorbed oxygen on the ZnO NW surface, leading to the release of electrons back to the sensor surface and consequently to the decrease of the electrical resistance (Equations (1) and ( 2)) [36] : Similarly, for all tested gases, the response was slightly decreased in the presence of 40% RH (Figure 9d for Pd-decorated ZnO NWs).This was explained by the fact that in a moist atmosphere, upon adsorption of water molecules at the sensor surface, the number of free sites on the sensor surface drops, thus decreasing the adsorption of target gases and consequently the sensing response.
Similarly, the initial resistance of the NiO-shelled Pd-decorated ZnO NW gas sensor decreased in the presence of 40% RH (Figure 10a-c) and the response was slightly decreased in the presence of 40% RH (Figure 10d) for the same reasons described for the Pd-decorated ZnO NW sensor.
Lastly, analysis of the response of Pd-decorated ZnO NWs (Figure S1a-c, Supporting Information) and NiO-shelled Pddecorated ZnO NWs (Figure S1d-f, Supporting Information) to pure 100 ppm H 2 gas, 100 ppm H 2 + 500 ppm CO and 100 ppm H 2 + 500 ppm NO 2 at 200 °C after 15 and 30 days showed their long-term stability.Comparison (Figure 11a,b) of the responses of Pd-decorated ZnO NWs and NiO-shelled Pd-decorated ZnO NWs to target gases at 200 °C immediately after fabrication and after 15 and 30 days showed that although the responses slightly decreased likely due to adsorption of water molecules or of some particles from the laboratory atmosphere, they were still high even after 30 days.This good stability is important for practical applications.

Gas Sensing Mechanism
The general sensing mechanism of chemiresistive gas sensors is based on the resistance change in the presence of the target gases. [37]When pristine ZnO NWs are exposed to clean air, oxygen from air will be adsorbed on their surface.Then, due to its electrophilic nature, this oxygen takes electrons from the sensor surface to which they were adsorbed as ionic species [38] : The oxygen species present on the sensor surface are influenced by the surface condition and temperature.The dominant species are O − 2 and O − for temperatures < 150 °C and for 150 < temperatures < 300 °C, respectively.As an electron depletion layer (EDL) is formed at the ZnO NW surface, the conductivity is confined to the ZnO NW inner sides that form a conduction channel.Due to EDL formation and the conduction channel narrowing inside ZnO NWs, the conductivity decreases relative to the vacuum condition when oxygen is not adsorbed on the sensor surface.When added, H 2 gas is adsorbed on the ZnO NW surface where it reacts with the already adsorbed oxygen species as described in Equation ( 6) [39] : This leads to the liberation of electrons on ZnO NW surface, followed by the EDL narrowing and the conduction channel expansion inside ZnO NWs.This contributes to the changes in the sensor resistance and the appearance of the sensing signal.Moreover, the ZnO NW networked nature promotes the formation of homojunctions in air, in contact areas between ZnO-ZnO NWs.As a result, potential barriers are formed that hinder the electron flow from one ZnO NW to another.13b] Yet, the pristine ZnO NW gas sensor showed poor performance compared with other gas sensors.
To explain the improved performance of the Pd-decorated ZnO NW gas sensor, additional Pd NP-related mechanisms must be considered.Indeed, Pd is a good catalyst for splitting O 2 or H 2 molecules. [40]Pd can dissociate incoming oxygen species on its surface and the dissociated oxygen atoms can be adsorbed on the surface of neighboring ZnO in a socalled spill-over effect, leading to adsorption of more oxygen ions on the ZnO surface. [41]This increases the Pd-decorated ZnO baseline resistance compared with pristine ZnO NWs.In the presence of Pd NPs, H 2 is catalytically dissociated into H atoms that can move to neighboring ZnO surfaces (spill-over effect) [42] :  Then, H atoms can react freely with the previously adsorbed surface ions as follows [43] : Hence, Pd NPs can increase the dissociation rate and the subsequent reactions of H atoms with oxygen ions, thus improving the gas sensor response.
Moreover, because of the different work function values of Pd (5.3 eV) and ZnO (4.5 eV), [41] when they are in intimate contact, electrons from ZnO move to Pd NPs to equalize the Fermi levels.Therefore, band bending occurs and Schottky barriers form between the Pd and ZnO interfaces, leading to the formation of potential barriers to the electron flow from ZnO to Pd.In addition, due to electron transfer from ZnO to Pd, the EDL width on ZnO NWs increases, leading to a significant enlargement of the base resistance, as shown in Figure 5a,b for pristine and Pddecorated ZnO NW gas sensors.Then, following exposure to H 2 gas and release of electrons back to the sensor surface, the height of the Schottky barriers changes, leading to a significant decrease of the electrical resistance.It is also possible that after H 2 dissociation on the Pd NP surface, some H atoms diffuse into the Pd lattice.13b] Therefore, Pd can be partially converted to PdH x (Pd + x H → PdH x ) [44] that exhibits a lower work function and higher resistance compared with metallic Pd.Upon conversion to PdH x in the presence of H 2 gas, the height of the initially formed Schottky barriers significantly changes, contributing to the sensor signal.Furthermore, as revealed by the XPS analysis, Pd is partially oxidized into PdO because during the deposition of the NiO shell, the oxygen source was ozone that is considered a strong oxidizing agent.Therefore, when exposed to H 2 , a reducing gas, PdO should be reduced to Pd with a different resistance that contributes to the resistance change: All these contributions explain the higher response of the Pddecorated ZnO sensor to H 2 gas compared with pristine ZnO NWs.
The NiO-shelled Pd-decorated ZnO NW gas sensor displayed a lower response to H 2 gas compared with Pd-decorated ZnO NWs, but a better selectivity toward H 2 gas when mixed with NO 2 and CO gases.This can be mainly attributed to the fact that due to the thin continuous NiO shell deposited over the Pd-decorated ZnO NWs, Pd NPs are directly in contact with air.Therefore, Pd-related sensing enhancement is limited in the NiO-shelled Pd-decorated ZnO NW gas sensor, leading to a lower response compared with Pd-decorated ZnO NWs.Furthermore, NiO response to gases is intrinsically lower than that of ZnO due to its p-type nature. [45]Therefore, after the formation of a thin NiO shell over ZnO, NiO is exposed to H 2 instead of bare ZnO.
As the NiO-shelled Pd-decorated ZnO NW gas sensor displayed an n-type it can be inferred that the sensing properties are still governed by ZnO (n-type semiconducting behavior) and not by p-NiO.Indeed, due to the limited thickness of the NiO shell layer, electrical transport is not fully localized on this layer, but is observed in the NiO shell and also in the ZnO core (i.e.smearing effect). [46]Therefore, during exposure to air, the whole NiO shell is converted to a hole accumulation layer (HAL) and some ZnO NWs also are engaged in the sensing reactions.
This raises the question of why the NiO-shelled Pd-decorated ZnO NWs showed a lower baseline resistance compared with the Pd-decorated ZnO NW gas sensor.In Pd-decorated ZnO NWs both adsorbed oxygen species and Pd NPs lead to extraction of electrons from ZnO, resulting in a high baseline resistance.In NiO-shelled Pd-decorated ZnO NWs, electrons move from ZnO to both Pd and NiO, leading to EDL expansion in ZnO NWs and consequently to increased resistance, as reported by Nakate et al. [20] Moreover, some of the electrons that move from NiO to Pd are harvested by adsorbed oxygen species.These two effects cause HAL expansion on NiO and decrease resistance.Therefore, in NiO-shelled Pd-decorated ZnO NWs, both NiO and Pd lead to electron extraction from ZnO; however, the presence of a thin p-NiO shell, the full removal of electrons by adsorbed oxygen species and Pd NPs, and the full coverage of HAL on NiO lead to a slight reduction of resistance compared with Pd-decorated ZnO NWs gas.
In the NiO-shelled Pd-decorated ZnO NW gas sensor, three heterojunctions should be considered: NiO─ZnO, NiO─Pd Schottky, and Pd─ZnO Schottky heterojunctions.Specifically, as NiO has a work function value of 5-5.3 eV, [47] at the ZnO and NiO interface, electrons move from ZnO to NiO to equalize the Fermi levels, resulting in band bending and heterojunction formation.Upon exposure to H 2 gas, the released electrons cause a reduction of the potential barrier height between ZnO and NiO, resulting in the modulation of the sensor resistance.Furthermore, as Pd work function value is higher than those of NiO and ZnO, in the areas where ZnO and NiO are in contact with Pd, electrons move from NiO and ZnO to Pd, resulting in the generation of potential barriers.Upon exposure to H 2 gas, electrons are released back to the sensor surface and the height of these potential barriers changes, contributing to the sensing signal.Figure 12 schematically describes the energy band levels of Pd─ZnO─NiO before and after contact with air and H 2 .Now, it is possible to explain the sensing behavior of the optimized gas sensor in the presence of mixed gases.Upon exposure to a gas mixture (H 2 and 500 ppm CO), its response decreases compared to the response to pure H 2 .Like H 2 , CO gas is a reducing molecule.When the sensor is exposed to CO gas, the following reaction is likely to occur [48] : Consequently, as observed with H 2 gas, electrons move back to the sensor surface, resulting in the conductivity increase and resistance decrease.As only one electron is liberated upon CO reaction with the adsorbed oxygen species (like with H 2 ), the response decrease upon exposure to CO gas is not linked to electron-donating effects.The response decrease in the presence of CO can be related to i) the sensing temperature and ii) the larger size of CO molecules relative to H 2 .Indeed, as each gas has its own unique characteristics, depending on the surface condition and gas type, the maximum adsorption of a gas at the sensor surface occurs at a specific temperature, which may differ for different gases.Our results suggest that 200 °C is not the optimal sensing temperature for CO gas.Moreover, CO molecules have a larger kinetic diameter (3.70 Å) [49] than H 2 molecules (2.89 Å). [50] Assuming the same adsorption sites on the sensor surface, CO molecules can occupy more adsorption sites due to their larger size when they are mixed with H 2 gas.This leads to a reduction of sites for H 2 gas adsorption and consequently to a lower gas sensor response in the presence of H 2 + CO gas mixtures.
Concerning the slight decrease of the response in the presence of H 2 + NO 2 gas mixtures, NO 2 can react with oxygen species adsorbed to the sensor surface or directly collect electrons from the sensor surface, leading to the following reactions [51] : According to these reactions, the collection of electrons by the adsorbed NO 2 results in a resistance increase.However, in the presence of H 2 + NO 2 gas mixtures, the overall resistance decreased, reflecting the fact that the gas sensor is more sensitive to H 2 than to NO 2 .As previously discussed, the reduced response to such mixtures can be partly explained by the larger size of NO 2 , compared with H 2 , and the sensing temperature.
To confirm the lower response of the optimized sensor to CO and NO 2 gases, we also measured the response of NiO-shelled Pd-decorated ZnO NWs to different concentrations of CO and NO 2 gases at 200 °C (Figure S2a,b, Supporting Information, respectively).Comparison of the obtained results (Figure S2c, Supporting Information) showed that the response to CO and NO 2 was low and similar for both gases.

Conclusion
Pristine ZnO NW, Pd-decorated ZnO NW and NiO shelled Pddecorated ZnO NW gas sensors were fabricated for H 2 sensing studies.Using ALD, Pd NPs and a NiO shell were coated over the as-grown ZnO NWs.The morphology, crystalline nature, phase, and chemical composition studies demonstrated the synthesis of materials with the expected features.The gas sensing studies showed the pristine ZnO NW gas sensor poor performance toward H 2 gas compared with the other two gas sensors.At 200 °C, the responses of the Pd-decorated ZnO NW and NiO-shelled Pddecorated ZnO NW gas sensors to 500 ppm H 2 gas were 41.01 and 16.12.When gas mixtures were used, the NiO-shelled Pddecorated ZnO NW gas sensor demonstrated superior selectivity to H 2 compared with the Pd-decorated ZnO NW sensor: 13.19 versus 6.39 to 500 ppm H 2 + 500 ppm NO 2 and 12.41 versus 12.11 to 500 ppm H 2 + 500 ppm CO, respectively.These results highlight the enhanced selectivity of the NiO-shelled Pddecorated ZnO NW gas sensor to H 2 gas in the presence of H 2 + CO and H 2 + NO 2 gas mixtures.The enhanced performance of the optimized sensor was related to the formation of NiO-ZnO heterojunctions, NiO-Pd Schottky junctions and Pd-ZnO Schottky junctions, Pd catalytic effect and the H 2 gas nature (small kinetic diameter).This study demonstrated that it is possible to develop a highly selective H 2 gas sensor in realistic conditions.

Experimental Section
Growth of ZnO NWs: Networked ZnO NWs were directly grown on a patterned electrode substrate using VLS growth with an Au-catalyst as previously described. [27,52]Briefly, TIEs were sputter-deposited on SiO 2 (200 nm thick)-covered Si (100) substrates with the following sequence: Ti (50 nm), Pt (200 nm), and Au (3 nm).Then, the substrate with TIEs was placed into a quartz tube furnace at a fixed distance from a ceramic boat containing metallic Zn powder (99.9%) placed at the furnace center.The temperature increase to 900 °C, in an atmosphere containing flowing N 2 (300 sccm) and O 2 (10 sccm) gases, led to the melting of the Au catalyst and to the generation of Zn vapors that were moved toward the substrate by the carrier gas (N 2 ).Zn diffused into the Au droplets and ZnO directional growth at the substrate/droplets interface began when the droplets became supersaturated with Zn atoms.The networked ZnO NWs were grown on the TIE for 30 min, and they became entangled in the areas between the electrodes.
Pd NP Deposition by ALD: Pd NPs were deposited on ZnO NWs in a home-built ALD reactor [53] at 220 °C using Pd(hfac) 2 (95% from Strem Chemicals) heated to 70 °C and formalin (37% formaldehyde in water with 10-15% methanol; Sigma-Aldrich) as precursor and co-reactant, respectively, as described earlier. [54]Each ALD cycle was as follows: Pd pulse for 5 s, exposure for 15 s and Ar purge for 10 s, followed by formalin pulse for 1 s, exposure for 15 s and Ar purge for 60 s.N Pd = 100 cycles were used to obtain the desired Pd NPs.
NiO Layer Deposition by ALD: NiO was deposited on Pd-decorated ZnO NWs and on Si (100).For deposition on Si, the wafers were cut into 1 × 1 cm 2 pieces and were degreased by sonication in acetone, isopropanol, methanol, and ultra-pure water, followed by dipping in 5% hydrofluoric acid solution for 30 s.32a] Ni(CpEt) 2 (99.99%,Strem Chemicals) and O 3 generated from O 2 were used as Ni precursor and O source, respectively.Ar (≥ 99.999%, from Linde Electronics) served as vector gas.Ni(CpEt) 2 was maintained at 94 °C and the chamber temperature was fixed at T NiO = 250 °C.ALD sequence consisted of pulsing and purging successively Ni(CpEt) 2 and O 3 for defined times with an additional exposure step after each pulse to uniformly cover the ZnO NWs.The ALD sequence (pulse/exposure/purge) was set as follows: 2/15/34 s for Ni(CpEt) 2 and 0.3/12/34 s for O 3 .32a] Characterization: Morphology was investigated by SEM and TEM, using a JSM 7900F (JEOL Ltd) and a JEM 2010 (JEOL Ltd) microscope, respectively.The crystal structure was studied by SAED performed with TEM.The different elements were localized by EDS (Quantax FlatQuad, Bruker).The chemical composition was assessed by XPS (Kratos Analytical, UK) with a monochromatic Al K  source (1486.6 eV).The binding energy (BE) was corrected using the C 1s peak at 284.8 eV.Curves were fitted with CASAXPS, version 2.3.25, a Shirley background subtraction routine and Lorentzian/Gaussian components.NiO thickness was measured in situ with a M2000V spectroscopic ellipsometer (J. A. Woollam Inc).
Gas Sensing Measurements: The method used for gas sensing measurements was previously described. [55]The gas sensors under study were exposed to different gases at various temperatures using a homemade sensing system.The target gas concentration was monitored by adjusting the target gas-to-dry air ratio using accurate mass flow controller devices (total flow rate = 500 sccm).During measurements, the gas sensor resistance changes in air (R a ) and in the presence of the target gas (R g ) were automatically recorded.The response was defined as R = R a /R g .The sensor response time was defined as the time needed to reach 90% of the final resistance after exposure to the target gas and the recovery time was the time needed to recover 90% of the initial resistance after removal of the gas.

Figure 1 .
Figure 1.SEM images of a) VLS-grown ZnO NWs on the substrate, b) ZnO NW network, c) a single ZnO NW, and d) a single ZnO NW coated by Pd NPs.

Figure 2 .
Figure 2. Representative TEM photographs of a) a bare single ZnO NW and b) a NiO-shelled Pd-decorated ZnO NW.SAED patterns of a c) bare and d) Ni-shelled Pd-decorated ZnO NW. e) Low and f) high magnification TEM photographs of NiO-shelled Pd-decorated ZnO NWs.The box in (e) corresponds to the high-magnification area in (f).

Figure 3 .
Figure 3. EDS spectrum and mapping of a single sensing NW. a) SEM image of a typical NiO-shelled Pd-decorated ZnO NW. b) EDS spectrum corresponding to the boxed area in (a).EDS mapping of c) Zn d) Ni and e) Pd in the NiO shelled Pd-decorated ZnO NW.

Figure 5 .
Figure 5. a) Dynamic resistance curves of (a) pristine ZnO NWs, b) Pd-decorated ZnO NWs, and c) NiO-shelled Pd-decorated ZnO NW gas sensors in the presence of 1, 10, 100, and 500 ppm H 2 gas at the indicated temperatures.Sensor response in function of the temperature for d) pristine ZnO NWs, e) Pd-decorated ZnO NWs, and f) NiO-shelled Pd-decorated ZnO NW gas sensors at different H 2 gas concentrations.

Figure 6 .
Figure 6.Dynamic resistance curves of a) Pd-decorated ZnO NW and b) NiO-shelled Pd-decorated ZnO NW gas sensors to 1, 10, 100, and 500 ppm H 2 gas and to mixtures of different concentrations of H 2 and 500 ppm CO or 500 ppm NO 2 at 200 °C.

Figure 7 .
Figure 7. Responses of a) Pd-decorated ZnO NW and b) NiO-shelled Pd-decorated ZnO NW sensors to 1, 10, 100, and 500 ppm H 2 gas and to H 2 (different concentrations) mixed with 500 ppm CO or 500 ppm NO 2 at 200 °C.Comparison of the responses of Pd-decorated ZnO NWs and NiO-shelled Pd-decorated ZnO NWs to H 2 (different concentrations) mixed with c) 500 ppm NO 2 gas and d) 500 ppm CO gas at 200 °C.

Figure 8 .
Figure 8. Repeatability tests of a) Pd-decorated ZnO NWs and c) NiO-shelled Pd-decorated ZnO NWs during five sequential cycles of exposure to 100 ppm H 2 alone and to100 ppm H 2 with 500 ppm CO or 500 ppm NO 2 at 200 °C.Responses in function of the cycle number of b) Pd-decorated ZnO NWs and d) NiO-shelled Pd-decorated ZnO NWs.

Figure 9 .
Figure 9. Dynamic resistance of the Pd-decorated ZnO NW gas sensor in the presence of a) 100 ppm H 2 gas, of b) 100 ppm H 2 + 500 ppm CO, and of c) 100 ppm H 2 + 500 ppm NO 2 in a dry (0% of RH) or moist atmosphere (40% RH) at 200 °C.d) Comparison of Pd-decorated ZnO NW responses to the target gases in the presence of 0 and 40% RH at 200 °C.

Figure 10 .
Figure 10.Dynamic resistance of the NiO-shelled Pd-decorated ZnO NW gas sensor in the presence of a) 100 ppm H 2 , of b) 100 ppm H 2 + 500 ppm CO, and of c) 100 ppm H 2 + 500 ppm NO 2 in a dry (0% RH) or moist atmosphere (40% RH) at 200 °C.d) Comparison of the responses to the target gases in the presence of 0 and 40% RH at 200 °C.

Figure 11 .
Figure 11.Long-term stability of the response of a) Pd-decorated ZnO NWs and b) NiO-shelled Pd-decorated ZnO NWs to the indicated gases at 200 °C just after fabrication (Fresh) and after 15 and 30 days.

Figure 12 .
Figure 12.Energy band levels of Pd─ZnO─NiO a) before contact and after contact b) in air and c) in H 2 .