Substrate critical effect on the structural and H2 Gas sensing characteristics of solution-processed Zn0.075Cu0.025O films

In this study, we report the synthesis of Zn0.075Cu0.025O films by chemical bath deposition to determine the effect of substrate (glass slide or ZnO seed layer) on the structural and H2 gas sensing properties of the produced films. The crystal phase, structural topography, surface morphology, and functional groups of the as-synthesized films as well as H2 gas sensing properties were investigated. Although both films have a hexagonal wurtzite structure, ZnO seed layer-based Zn0.075Cu0.025O film is more crystalline than glass slide-based Zn0.075Cu0.025O films. ZnO seed layer-based Zn0.075Cu0.025O films exhibited much more nanorod and fewer nanosphere forms compared to glass slide-based Zn0.075Cu0.025O films. EDX analysis and Raman spectra of both samples confirmed the presence of defects in Cu: ZnO samples. ZnO seed layer-based sensors showed higher response (140%) and lower operating temperature (80 °C) compared to glass slide-based sensors (87% response and 140 °C operating temperature). The most important thing to note here is that the fabricated sensors exhibited a high response at room temperature. The responses at room temperature were found as 46% and 23% for the ZnO seed layer-based and glass slide-based sensors, respectively. Sensors operating at room temperature are especially important for commercial applications.


Introduction
Although H 2 is a non-toxic gas, some of the hazards associated with hydrogen usage can be given as follows; respiratory distress and suffocation, phase changes, component failures, ignition, and combustion. In most cases, a combination of these dangers can also occur. Precautions must be taken because the primary risk associated with any form of hydrogen is the possible fire or explosion by accidentally producing a flammable or detonable mixture [1]. Therefore, fuel cell, battery and electrochemical research laboratories, oil and gas refining, manufacturing processes that use or generate hydrogen as well as oil and gas refining factories must use real-time, selective, and sensitive H 2 gas sensors. Previous studies involving H 2 gas sensors have been carried out using various metal oxides such as SnO 2 , ZnO, and NiO. Among them, ZnO material has been widely employed to detect a series of inorganic gases and organic vapors due to 3.37 eV wide bandgap energy, 60 meV binding energy, good chemical stability at high temperature, and high electron mobility [2]. Researchers have extensively used ZnO material for gas sensing applications because of matching good sensor performance [3]. It is primarily aimed to improve the gas detection properties of ZnO material by using different strategies. Zhou et al have prepared a ZnO sensor and investigated its NO 2 gas sensing performances under UV-activated sensing performance at room temperature (25°C) [4]. Wang et al have studied a NO 2 sensor featuring composite film of few-layer black phosphorus (BP) nanosheets and zinc oxide nanowires serving as the sensing layer. BP-ZnO sensor exhibited 74% sensing performance compared to undoped one (37.7% for 50 ppb NO 2 gas) [5]. Zhou et al have studied cuprous oxide grown onto reduced graphene oxide nanosheets prepared to serve as the sensing Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
layer. Compared to the partial recovery at 25°C, the sensor showed a response of 20% toward 1 ppm H 2 S at 40°C [6].
Creating new active centers on the surface or altering the surface/volume ratio of materials are the main strategies of these studies to decrease operating temperatures and detectable threshold of H 2 gas concentration down to low ppm levels [7]. With the use of dopant materials, the control of impurities in wide bandgap semiconductors can lead to surprising changes in the charge carrier properties [8]. Therefore, the doping process with metals of compatible ionic radius and high electrical conductivity is often preferred. Copper is a proper dopant metal for ZnO semiconductor due to its high electrical conductivity (58.7×10 6 S m −1 ), similar ionic radius (r Cu 2+ =0.72 Å and r Zn 2+ =0.74 Å) and tendency of altering oxygen vacancies with Cu 2+ doping. Researches on the operation of Cu:ZnO films as a gas sensor in recent years can be given as follows; Brahma et al have investigated room temperature acetone sensors using Cu:ZnO thin films [9]. Alev et al have reported that 3% Cu-doped ZnO nanorods showed the highest sensor response to ethanol which is 5.5 times higher than that of pure ZnO with increased surface oxygen species and lowered activation energy by Cu doping [10]. When Zn 0.90 Cu 0.10 O samples were investigated for the detection of NO gas, improved sensing performance was observed at an operating temperature of 55°C as was shown in our previous study [11]. By sol-gel process used with different types of precursors, Gómez-Pozos et al have reported that the maximum sensitivity was exhibited in Cu-doped ZnO films prepared using copper chloride and deposited with eight immersions [12]. Bearing that H 2 S production is usually accompanied by H 2 , Shewale et al have shown that the selective performance was good in Cu:ZnO/rGO structures that exhibited high sensitivity towards H 2 S gas [13]. ZnO seed layer acts as a nucleation center and so it provides a more stable structure and larger nucleation sites. Therefore, particle growth can be controlled with an easy coating on ZnO seed layer. In addition, using ZnO seed layer can reduce the lattice mismatch between the growing layer and substrate [14]. Gonzalez-Garnica et al have shown that ZnO nanowires with seed layers have much more sensitivity towards CO 2 gas [15]. In the present work, a systematic study was carried out to determine the possible effects of the type of substrate on the structural, morphological, and H 2 gas sensing properties of Zn 0.075 Cu 0.025 O films synthesized by chemical bath deposition.

Experimental
Firstly, the used microscope glasses were cleaned in an ultrasonic mixer with ethanol, then acetone and finally pure water and dried at room temperature. The cleaned glasses were kept in the conditioned solution for 15 min to allow ZnO to adhere to the glass surface more easily. The ratio of ammonia (NH 3 )/trimethylamine (TEA)/ pure water in this conditioning solution was set as 4:2:25. As a zinc (Zn) solution source, 99.99% purity Zn(CH 3 COO) 2 .2H 2 O (zinc acetate dihydrate) was used. First, 0.1 M and 100 ml of zinc aqueous solution was prepared. 10 ml of ammonium hydroxide (NH 4 OH) (26%) solution was added dropwise and the solution was mixed with a magnetic stirrer.
Ammonia was added to make the zinc source alkaline, and as ammonia was added, solution color changed from milky white to colorless, and the pH value of this solution was set to 10. Then, pure water was added to this solution until it reached 500 ml and a (NH 4 ) 2 ZnO 2 bath was obtained. When the solution was examined the next day, it was observed that Zn(OH) 2 formed a white precipitate at the bottom. 20 ml of this prepared solution and 60 ml of distilled water were taken and the solution mixed in the beaker was placed on a magnetic stirrer at 50±5°C (pH 9.8). The films were taken from the conditioned solution, immersed in this mixture and left for 5 min The films were left to dry for one day and then dried in an oven at 100°C the next day; thus, weakly adherent structures were removed from the surface of the obtained ZnO seed layer.
After that, Cu-doped ZnO thin film was deposited on the synthesized ZnO seed layer. First, Zn(Ac) 2 2H 2 O (Sigma Aldrich), copper powder (Sigma Aldrich) and NH 3 (Sigma Aldrich) were used as the source materials and complex agents. Initially, zinc source was dissolved in distilled water and stirred using magnetic stirrer. Copper source (at.% of 0.025) was added into the zinc-aqueous solution. Ammonia was added drop by drop to maintain the pH value at 10.4. The cleaned micro glass and ZnO seed layer [16] was used as substrates and they were immersed vertically with the help of a substrate holder. The solution was stirred by magnetic stirrer at 70±5°C in order to get the uniform coating on the substrate. After 20 min of deposition, the coated substrates were taken out from the solutions and cleaned with distilled water and was put on hold for a day. The coated substrates are kept in the furnace and annealed at 500°C for 1 h.
The surface morphology and topography of the samples were analyzed using scanning electron microscopy (SEM, JEOL JSM-7100F) and atomic force microscopy (AFM, Witec alpha 300 RA) with non-contact mode. Elemental analysis was determined by energy dispersive x-ray spectrometer (OXFORD Instruments X-Max) attached to SEM. X-ray diffraction (XRD, Rigaku SmartLab) patterns of the samples were identified in the 2θ range of 20°-80°using CuK α radiation (λ = 1.5406 Å), whilst Raman spectrum (Termo DXR) was employed to characterize the functional groups of the samples for 780 nm laser line. Gas measurements were taken in a fully computer-controlled gas detection system. The sensor cell consists of three parts: gas valves providing gas inlet and outlet, sample holder, and BNC connectors for necessary connections to Keithley 2400 device. Temperature-dependent measurements are made with the Lakeshore 325 temperature controller and gas measurements are made using the Keithley 2400 device. First, the sensors were fabricated and placed in the sensor cell. The sensor cell was then kept at a constant temperature using a temperature controller. With the help of gas flow controllers, target gas flow was provided to the sensor surface and measurements were taken. The flow rate of dry air entering the test cell was fixed at 500 cm 3 min −1 during the measurements. Air flow rate was kept constant under the same conditions in order to observe the behavior of different target gas concentrations. Sensor response degraded with humidity, in as much as water molecules and target gas compete in adsorption on the sensing surface which then leads to a decrease in sensitivity [17,18]. Therefore, in this study, the humidity rate was kept constant at 25% to eliminate such effects.
The sensor response (S) was calculated using the following equation [19]; where R a is the resistance in dry air and R g is the resistance upon exposure to the target gas. Response and recovery times are defined as the time interval when the resistance reaches 90% of the total change upon sending the target gas into the sensor cell and upon evacuating the target gas from the sensor cell, respectively [20].

Elemental analysis
In figure 1, elemental analysis results of the samples are shown. The presence of atomic percentage (at.%) of Zn, O and Cu has been confirmed, which implies the formation of Cu:ZnO system. It is seen that the atomic percentage of copper in the structure is higher with the seed layer usage. It underlines that the atomic ratio of Zn/O for all samples, along with any other defects in the samples, as well as the presence of defects caused by oxygen vacancies in the samples, is the atomic ratio of Zn/O for all samples [21].

Structural characterization
X-ray patterns of the samples are illustrated in figure 2. All the patterns of both samples correspond to hexagonal wurtzite phase of ZnO with JCPDS card . No secondary phase was observed in the samples due to the fact that Cu doping ratio was quite low and so copper cluster peaks were not detected. Similar copper doping optimization was realized by Roguai et al [22]. Copper doping ratio arrangement is important because initiation of copper doping can lead to an increase in the free energy of the system and a decrease in structural stability [23].
On the other hand, by introducing new phases above a certain doping ratio, the doping process can be eliminated, as explained by Sayed et al [24]. In the films where glass slide was used as a substrate, the x-ray diffraction intensities and FWHM remain quite weak compared to the films with ZnO seed layer substrate. The higher diffraction intensities of the ZnO seed layer-based samples means that the Zn 0.075 Cu 0.025 O/ZnO layer film has an excellent crystal quality [25]. In addition, peaks other than the characteristic peaks in the range 30°− 40°could not be seen in the glass slide-based samples because there is a mismatch between the glass slide and the coated Cu:ZnO layer. This suggests that the ZnO seed layer is more ideal for film coating because it allows ZnO crystal growth with Cu doping, although it is intended for nanorod or nanowire growth [26,27]. The slight shift in the original ZnO peaks' positions demonstrates that doping is successful by replacing the Zn 2+ ions with Cu 2+ ions in both of the samples. Table 1 gives the structural parameters of the samples.
For the hexagonal wurtzite, the lattice parameters (a and c) are computed by [28]; The calculated lattice constants are given in table 1. The c/a ratio is 1.633 in a stoichiometric wurtzite structure [29]. As seen, c/a ratio of all samples (table 2) is smaller than ideal which might be linked to the presence of oxygen vacancies V O and extended defects [30]. The average crystallite sizes of the films were estimated using Debye Scherrer's formula, also the microstrain and dislocation density was computed by below equations [31][32][33]; The dislocation density is a measure of the number of dislocations in a unit volume of crystalline material. One of the main parameters for determining crystallinity and its enhancement is dislocation density. It refers to the displacement of atoms relative to the reference lattice positions, and this can be associated with the heat generation and lattice vibration in the molecular bond. If the structure has a small dislocation density, its crystalline structure is better than the other ones. However, a larger dislocation density implies a higher hardness [34]. When table 1 is examined, it is seen that ZnO seed layer-based film has smaller dislocation density and microstrain values. According to Deepa et al, the lower the lattice strain and defect density are, the more the carrier density and surface oxygen adsorption can happen [35].
Continuing with the XRD analysis, the parameters given in table 2 were calculated with the help of the following equations. Equations  distribution and random orientation of particles is noticeable although hexagonal particle formation is more pronounced. A similar study was conducted by Katı for Co:ZnO and explained that an increase in surface roughness has been observed due to a deterioration in the crystal structure, increase of defect intensities and deterioration in surface distribution [46]. As shown in figure 3, particle sizes decreased with ZnO seed layer usage. However, the very high average height values obtained by AFM analysis (431.4 and 128 nm) can be explained as follows, as elaborated by Farhad et al; the crystallite sizes obtained from XRD can be more accurate, since AFM analysis gives the crystallite size which can consist of more than one crystal and possible including of cluster forms [47].
The AFM analysis makes the synthesis of the samples repeatable and generalizes the results of the analysis from each part of the sample to the whole [48,49]. Table 3 depicts some structural parameters from AFM analysis of Cu:ZnO films deposited on different substrates.
Under ×50,000 magnification ratio, SEM images of the synthesized Cu:ZnO films are shown in figures 4(a) and (b). From these surface images, high density particle formations and no-crack forms were determined. Although there are granule structures of different sizes, the crystallite sizes of the granular forms are consistent with the results obtained from the Debye-Scherrer equation, which is measured for samples using glass substrates. Nanorod, decreasing nano-granular forms and big granular forms have been observed in ZnO seed layer-based samples. This morphological mixing and degradation by changing the substrate type to the ZnO seed layer can be explained by the increase in nucleation centers due to ZnO seed layer which can act as a nucleation center for Cu: ZnO crystal growth, as explained by Boukaous et al [50]. Increasing nanorod forms can be attributed to the lattice mismatch that might have occurred between the substrate and the coated Cu:ZnO layer [51]. Figure 5 represents the Raman spectrum of the synthesized Zn 0.075 Cu 0.025 O films in the spectral range between 250-1250 cm −1 .

Vibrational analysis
For a perfect ZnO crystal, optical phonon modes can be defined by the following description;  According to figure 5, seven typical vibrational peaks emerged in both samples at 290, 334, 432, 468, 576, 630 and 1098 cm −1 . The peak intensities of Raman modes of the films coated on ZnO seed layer are clearly higher than those of the films coated on glass slide. The peak at 334 cm −1 corresponds to optical phonon overtone with A1 symmetry [52]. The characteristic E2 (high) mode of wurtzite ZnO phase is observed at 432 cm −1 , especially in films coated on ZnO seed layer [21]. The presence of defects such as oxygen vacancies or interstitials zinc was manifested by peak of E1(TO) mode located at 468 cm −1 , indicating ZnO large crystal [53]. The peaks that are observed at 576 cm −1 and 630 cm −1 (Ribut et al have shown that these peaks were observed for ZnO/PET sample) can be assigned to A1(LO) phonon mode [24]. Abdelouhab et al reported that this peak was observed at 535 cm −1 , indicating that copper doping can cause a Raman shift due to ZnO phonon localization change by point defects and surface impurities [54]. Yin et al have shown that the vibration absorption peak at 576 cm −1 is closely related to the oxygen vacancies of ZnO [55]. The sharp peak observed at 1098 cm −1 corresponds to CH vibrations and is probably caused by indestructible aqueous solution [56]. In both samples, a slight shift from the original ZnO phonon mode locations were detected due to surface roughness and defects with copper doping effect. From Raman spectrum, E1(LO) phonon mode peak shift is in well agreement with EDX results to prove the presence of defects.   were calculated as 96.15 kΩ and 71.53 kΩ for glass slide-and ZnO seed layer-based samples, respectively. It can be inferred from other literature studies that the presence of ZnO seed layer increases current conduction and changes the current-voltage graphs [57][58][59].

H 2 gas sensing properties
To investigate the optimum operating temperature of the sensors, the gas chamber was subjected to an operating temperature from 30°C to 150°C and the gas response was calculated. Figure 7 shows the H 2 gas sensing properties (500 ppm) of the sensors as a function of operating temperature. The most notable thing is that both sensors exhibited high responses at room temperature. The responses at room temperature were found as 46% and 23% for the ZnO seed layer-based sensor and glass slide-based sensor, respectively. Sensors operating at room temperature are very important for commercial applications.
Mustaffa et al have reported that they have grown ZnO thin films by magnetron sputtering technique and optimized their producing parameters. They obtained acceptable responses for hydrogen gas at room temperature [60]. Hassan et al have reported that ZnO nanorods were produced with the microwave-assisted chemical bath deposition method. They used Polyvinylalcohol (PVA)-Zn(OH) 2 nanocomposite as a seed layer and they have carried out the hydrogen gas sensing measurements at room temperature and reported that using a seed layer made the sensors work at low operating temperatures [61]. The optimal operating temperature was found at 80°C and 140°C for the ZnO seed layer-based sensor and glass slide-based sensor. After this temperature, responses began to decrease. The energy of the oxygen atoms sent to the surface increases with increasing temperature. However, after a certain critical temperature, the increase in energy values becomes too much, so that oxygen atoms can no longer hold onto the surface [62]. This causes instability and prevents the desired reaction with the target gas and thus, a decrease in sensitivities starts to be observed. Therefore, each sensor material has a unique operating temperature. In addition, the improved gas response of the Cu-doped ZnO samples might be due to the effects of the Cu ions on the ZnO surface properties, defects, and the large oxygen vacancy concentration [62].
The glass-based sample always has an uptrend but the ZnO seed-based sample has a maximum at 80°C. This can be associated with the seed layer effect. ZnO structures were grown on different substrates and so the difference in trend can be attributed to the nature, surface topography and structure of substrates as well as the lattice mismatch with the synthesized nanostructures. These are particular significant parameters that can be factors to control the morphology and nature of the synthesized ZnO structures [63,64].
The dynamic gas sensing measurements were carried out at 80°C for both sensors, as shown in figure 8. The responses increased with increasing gas concentrations. The responses of the Zn 0.075 Cu 0.025 O films on the glass slide were calculated as 19%, 50%, 58%, 68%, 79% and 87% for 50 ppm, 100 ppm, 150 ppm, 200 ppm, 250 ppm and 500 ppm, respectively. It can be seen from the measurements that the sensors showed different responses for each gas concentration, the sensitivity increased with increasing gas concentration and the sensor resistance returned to its initial value when the gas entering the test cell was turned off. The responses of the Zn 0.075 Cu 0.025 O films on the ZnO seed layer were calculated as 58%, 68%, 79%, 85%, 99% and 140% for 50 ppm, 100 ppm, 150 ppm, 200 ppm, 250 ppm and 500 ppm, respectively. There is an exchange of electrons from the lower ZnO seed layer to the upper layer, thus increasing the number of free electrons in the structure which results in an n-n heterostructure. Studies in this form are also available in the literature [61].
It is seen that higher gas responses were obtained in the sample grown on the ZnO seed layer. These results are consistent with the XRD analysis. Referring to table 1, the sample grown on the ZnO seed layer has a lower lattice strain and defect density which can lead to more carrier density and surface oxygen adsorption [35]. In addition, the volume of nanoparticles was higher for the ZnO seed layer-based sample, as shown in table 2. As the number of nanoparticles increases, the amount of gas adsorbed to the surface will increase.
Also, the doping process leads to increases in gas responses. When ZnO structure was doped with Cu, some of the Zn 2+ ions in the ZnO structure get replaced by Cu 2+ ions. As a result, Cu-Zn complex defects occur which may lead to more electrons in the conduction band [65]. The doping of Cu can occur more increasing charged oxygen vacancies because of the formation of more surface-absorbed oxygen [66]. With the increment of active adsorption regions, an increment in adsorbed gas molecules can be detected [9]. However, the distribution of Cu in ZnO thin films is not homogeneous, which may also help to form defects such as Cu interstitial and oxygen vacancies. Thus, the Zn 0.075 Cu 0.025 O films on the ZnO seed layer and glass slide, can both of them generate oxygen vacancy defects easily with the Cu doping process. It can be found in the literature that Cu doping leads to corrections in the sensor parameters. Nimbalkar et al have reported that copper-doped zinc oxide thin films were successfully prepared by sol-gel spin coating technique and their H 2 S gas sensing properties were investigated [67]. The Cu-doped ZnO thin film can stand as a prospective candidate for the detection of H 2 S gas at low concentration (5-50 ppm). Chow et al have reported that Cu-ZnO sensors were produced by  hydrothermal synthesis method. They have found that Cu-ZnO sensors showed room-temperature sensitivity to H 2 gas with low power consumption [68]. Figure 9 depicts the gas responses versus gas concentrations at 80°C where almost a linear increase was observed for both sensors.
Response and recovery times are other important parameters for gas sensors. Figures 10 and 11 give the response and recovery times of both sensors where measurements were taken at the optimum operating temperature. Based on the obtained results, it has been seen that the thin films produced can be used as sensor materials in commercial applications. The response and recovery times for 50 ppm H 2 gas were found respectively as 7 s and 12 s for ZnO seed layer-based sensor as well as 9 s and 18 s for glass slide-based sensor. Apart from the sensor responses, both sensors have very fast response and recovery times. Tables 4 and 5 give the H 2 gas sensing analysis of the synthesized Zn 0.075 Cu 0.025 O films for ZnO seed layer and glass slide-based sensors, respectively. As can be seen from the tables, the sensors have very high responses at low operating temperature. In addition, the sensors have response and recovery times in the order of seconds. Based on these results, it can be said that the produced sensors can be used commercially easily.
Generally, the gas sensing mechanism is a chemical reaction-based detection system [69]. The gas sensing mechanism can occur as two main reactions on the MOS sensor surface. In the first reaction, oxygen molecules in the air are chemically adsorbed on the sensor surface and the oxygen molecules transfer electrons in the conduction band towards it [70]. As a result of this transfer, oxygen molecules easily attach to the sensor surface and form an electron consumption region on the surface [70]. The second reaction takes place between the target gas and the oxygen molecules chemically adsorbed to the surface. An interaction of H 2 gas with molecular oxygen can take place as is given by the below equations [61]; Meanwhile, from the sensor response study, it is concluded that the synthesized Zn 0.075 Cu 0.025 O films on the ZnO seed layer are more preferable, indicating that the ZnO seed layer has noticeable positive effect on the gas sensing performance of the synthesized films.
Also, the effect of crystallite size is one of the important factors for gas sensing applications. When interpreted together with the volume of nanoparticles value, it was observed that the sample with ZnO seed layer had higher responses. Rothschild et al have reported that the response values they found in their study changed in proportion to the crystallite size D. When almost all the electrons are trapped at the surface, the carrier concentration of the surface state density reaches a critical value that shows a condition of fully depleted grains [71]. Savu et al have studied that tin oxide films were grown by two different methods, RF magnetron sputtering and doctor blade techniques, respectively. Because of the lower initial resistance and bigger crystallite size, the doctor-bladed samples can be shown more good sensing performance [72].

Conclusion
In this work, a trend change in H 2 gas sensitivity has been revealed and discussed by substrate type (glass slide or ZnO seed layer) of Cu:ZnO films. X-ray patterns, EDX analysis and Raman spectrum of both samples confirm the incorporation of Cu into the ZnO host matrix and the presence of localized defects by Cu doping. The most notable thing is that both sensors exhibited high responses at room temperature. The responses at room temperature was found as 46% and 23% for the ZnO seed layer-based sensor and glass slide-based sensor, respectively. Sensors operating at room temperature are very important for commercial applications. Apart from the sensor responses, both sensors showed very fast response and recovery times, though the sensor based on ZnO seed layer was more preferable. This signfies that the ZnO seed layer can have noticeable positive effects on the gas sensing performance of the synthesized films.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).