CuO-Ga2O3 Thin Films as a Gas-Sensitive Material for Acetone Detection

The p-n heterostructures of CuO-Ga2O3 obtained by magnetron sputtering technology in a fully reactive mode (deposition in pure oxygen) were tested under exposure to low acetone concentrations. After deposition, the films were annealed at previously confirmed conditions (400 °C/4 h/synthetic air) and further investigated by utilization of X-ray diffraction (XRD), X-ray reflectivity (XRR), energy-dispersive X-ray spectroscopy (EDS). The gas-sensing behavior was tested in the air/acetone atmosphere in the range of 0.1–1.25 ppm, as well as at various relative humidity (RH) levels (10–85%). The highest responses were obtained for samples based on the CuO-Ga2O3 (4% at. Ga).


Gas-Sensitive Layer Deposition
The gas-sensitive layers were based on copper oxide, which was previously confirmed as good gas-sensitive material [16,17,100], as well as on the copper oxide-gallium oxide composition with various Ga 2 O 3 contents. The films were deposited by the magnetron sputtering system with a glancing angle deposition technique and mosaic sputtering, where pure Ga (99.99999% purity) metal were placed on the pure Cu (99.9999%) magnetron target. The deposition conditions were previously fixed and are briefly presented in the Table 2.

Gas-Sensing Measurements
The gas-sensing measurement system consisted of a few elements: The quartz-tube oven with an internal heater and temperature control unit, gas-dosing lines equipped with mass flow controllers 1179B (MKS Instruments, Andover, MA, USA), and the resistance measurement unit, which consisted of an electrometer (34401A HP, Keysight, MA, USA) and target gas canisters with 5 ppm of acetone (Air Products, Hersham, UK). The gas-sensing measurement system was previously described in detail [101] The gas-sensing measurements were performed at various temperatures and 50% relative humidity (RH) level. Figure 1 shows the measurement system. The gas sensor response (S) was defined as the resistance ratio S = R gas /R air , where R gas and R air are electrical resistances in gas and air, respectively.

EDS, XRD and XRR
The chemical composition of the samples was studied with scanning electron microscopy (Vega Tescan 3, Vienna, Austria) equipped with an X-ray spectroscopy (EDS) spectrometer (Bruker XFlash 610M, Vienna, Austria). For each sample, the EDS spectroscopy measurements were done for several arbitrary points, which were chosen on the surface of the sample. Additionally, maps with a scan size of around 10 × 10 μm were performed to study elements distribution. The acquisition time for

EDS, XRD and XRR
The chemical composition of the samples was studied with scanning electron microscopy (Vega Tescan 3, Vienna, Austria) equipped with an X-ray spectroscopy (EDS) spectrometer (Bruker XFlash 610M, Vienna, Austria). For each sample, the EDS spectroscopy measurements were done for several arbitrary points, which were chosen on the surface of the sample. Additionally, maps with a scan size of around 10 × 10 µm were performed to study elements distribution. The acquisition time for single-point measurement and map were 5 min and 1 h, respectively.
The X-ray diffraction (XRD) and X-ray reflectivity (XRR) measurements were performed with an X'PertPro PANalitycal diffractometer equipped with an X-ray source with a Cu anode operating at 40 kV and 30 mA. The XRR data were collected for the ω angle between 0.2-2.0 degrees for 12 h for a sample to ensure sufficient statistics. The XRD θ/2θ patterns were gathered for a 2θ angle between 30 degrees and 90 degrees for 12 h for each sample. Additionally, in order to reduce signal from the Si wafer, a scattering vector was tilted by an angle of 3 degrees with respect to the diffraction plane. Details about measurement conditions have been described by the authors of [102,103]. The analysis of the obtained diffraction patterns was performed with FullProf software [104].

SEM
The EDS chemical composition analysis showed the presence of Si, O, Cu, and Ga elements in all samples, together with signals from C and N arising from sample contamination from the air. From the point of view of this study, the relative composition of copper and gallium was the most important factor. Hence, in this study, EDS analysis was restricted to those elements only. Table 3 shows the relative atomic compositions for Cu and Ga elements. Table 3. Atomic concentration of copper and gallium from X-ray spectroscopy (EDS).  Figure 2a shows an exponential increase of gallium concentration between the samples, where the concentration changed between~4 at.% and~16 at.%. The Ga concentration in the sample was doubled as compared to the previous sample. A typical distribution of Cu and Ga elements is presented in Figure 2b based on the sample S3. In Figure 2b, blue indicates Cu, and magenta indicates Ga. The map confirms the homogeneous distribution of both elements and a higher concentration of copper as compared to gallium. Because there were no visible differences in elements distribution between samples containing different amount of gallium, only maps for the sample with the highest Ga concentration are presented in the paper, and all distributions are included in the Supplementary Materials Section. indicates Ga. The map confirms the homogeneous distribution of both elements and a higher concentration of copper as compared to gallium. Because there were no visible differences in elements distribution between samples containing different amount of gallium, only maps for the sample with the highest Ga concentration are presented in the paper, and all distributions are included in the Supplementary Material Section.

XRD/XRR
The X-ray reflectometry analysis showed that the samples had a thickness in the range of 60-70 nm, except the sample with the largest concentration of Ga, where the layer thickness was 40 nm. The various thicknesses were the consequence of the deposition time, which was kept constant (20 min) to kept the same deposition parameters for various gallium metal dopants placed on the copper magnetron target. The obtained results are presented in Figure 3 together with the spectra gathered for CuO:Ga 2 O 3 (sample S4, shown as an insert). The density in all studied cases was found to be around 5.6(1) g/cm 3 regardless of the gallium concentration. This value was significantly lower than that found for bulk copper(II) oxide, which was 6.31 g/cm 3 . Similarly, the lack of dependence of roughness with film composition was observed. The mean value was found to be relatively large, at around 3 nm.

XRD/XRR
The X-ray reflectometry analysis showed that the samples had a thickness in the range of 60-70 nm, except the sample with the largest concentration of Ga, where the layer thickness was 40 nm. The various thicknesses were the consequence of the deposition time, which was kept constant (20 min) to kept the same deposition parameters for various gallium metal dopants placed on the copper magnetron target. The obtained results are presented in Figure 3 together with the spectra gathered for CuO:Ga2O3 (sample S4, shown as an insert). The density in all studied cases was found to be around 5.6(1) g/cm 3 regardless of the gallium concentration. This value was significantly lower than that found for bulk copper(II) oxide, which was 6.31 g/cm 3 . Similarly, the lack of dependence of roughness with film composition was observed. The mean value was found to be relatively large, at around 3 nm. The diffraction patterns for all samples are presented in Figure 4. The continuous lines are the measurements results (thin irregular lines) together with fits (thicker and smooth lines). The longitudinal red and green lines indicate the position of Braggs maxima for the CuO and β-Ga2O3 phases, while the back panel presents the powder X-ray diffraction patterns of those oxides. Both phases had a monoclinic crystal system with the C2/c and C2/m space group, respectively (ICSD card No. 00-045-0937 and ICSD card No. 00-041-1103). During the analysis, for the two samples with the smallest Ga amounts, only one phase of copper oxide was fitted, while for samples with about 8 at.% and 16 at.% of gallium, a small amount of β-Ga2O3 was also found, meaning a sufficient signal from crystalline gallium(III) oxide was detected with XRD measurements. For those diffractograms, The diffraction patterns for all samples are presented in Figure 4. The continuous lines are the measurements results (thin irregular lines) together with fits (thicker and smooth lines). The longitudinal red and green lines indicate the position of Braggs maxima for the CuO and β-Ga 2 O 3 phases, while the back panel presents the powder X-ray diffraction patterns of those oxides. Both phases had a monoclinic crystal system with the C2/c and C2/m space group, respectively (ICSD card No. 00-045-0937 and ICSD card No. 00-041-1103). During the analysis, for the two samples with the smallest Ga amounts, only one phase of copper oxide was fitted, while for samples with about 8 at.% and 16 at.% of gallium, a small amount of β-Ga 2 O 3 was also found, meaning a sufficient signal from crystalline gallium(III) oxide was detected with XRD measurements. For those diffractograms, three fitting lines were included on the graph, indicating patterns for the CuO and β-Ga 2 O 3 phases and a total phase. Additionally, for samples dopped with the largest amount of gallium, a shift in preferential grain orientation was clearly evident. The wide and intensive maxima observed for an angle around 35 deg., in case of sample S1 (CuO:Ga 2 O 3 with 4% at. Ga), became weak and small in samples strongly dopped with gallium. Furthermore, a change in the pattern leading to a shift of the peak was found around 38 deg. due to the larger values and its separation from at least two individual Bragg maxima.
Sensors 2020, 20, 3142 8 of 17 ngle around 35 deg., in case of sample S1 (CuO:Ga2O3 with 4% at. Ga), became weak and small amples strongly dopped with gallium. Furthermore, a change in the pattern leading to a shift of eak was found around 38 deg. due to the larger values and its separation from at least t ndividual Bragg maxima. Because of small amount of gallium oxide in the pattern, the crystallographic parameters us or this phase during the fitting procedure were from the crystallographic database and were k onstant during analysis. On the other hand, the parameters for copper oxide were varied and slo nd a systematic change with the increase of gallium concentration was observed. Figure 5 show hange of cell volume with Ga concentration where a decrease from ~82 A3 to ~80 A3 is clearly se he insert of the graph shows dependence of cell parameters of a, b, c and β constants with galliu oncentration. The dashed lines are values found in the database for the bulk sample. The m ignificant change was with the β angle, where a reduction of around 1 deg. was found for tudied samples, while the difference between the specific Ga concentration was uncertain. entioned previously, the reduction of cell volume is a consequence of the changes of the oncentration, which can be seen especially well for the sample S4 with the 16 at.% of G urthermore, with a shift of the a parameter, a strong (100) crystallographic texture was found uO:Ga2O3 films with 8 at.% (sample S2) and 16 at.% Ga (sample S3). Also, for those samples, ncrease of peaks widths was found.
An estimation of the size of crystallites was performed by applying a Scherrer formula. T cherrer formula, linking coherence length (Lcoh) together with the width of Braggs maxima ound at 2θ angle for used wavelength λ, allowed us to estimate the size of crystallites: Lcoh = kλ/ω , where k is a Scherrer constant typically equal to 0.95 for thin films [102]. The samples S0 and ithout and with small gallium concentration (i.e., pure CuO and CuO:Ga2O3 with 4% at. Ga) ha ean grain size of 12-14 nm, while the S3 and S4 samples with 8 at.% and 16 at.%, respective howed a decrease of mean grain size to ~7 nm. Furthermore, the mean size of gallium(III) ox rystallites for those two samples was found to be even lower, at around 5 nm. The XRD analy llowed us to estimate the amount of crystalline β-Ga2O3 phase in the spectra to be of seve ercents, i.e., 3(1)% and 5(1)% for S2 and S3 samples with Ga concentration of 8 at.% and 16 at Because of small amount of gallium oxide in the pattern, the crystallographic parameters used for this phase during the fitting procedure were from the crystallographic database and were kept constant during analysis. On the other hand, the parameters for copper oxide were varied and slow, and a systematic change with the increase of gallium concentration was observed. Figure 5 shows a change of cell volume with Ga concentration where a decrease from~82 A3 to~80 A3 is clearly seen.
The insert of the graph shows dependence of cell parameters of a, b, c and β constants with gallium concentration. The dashed lines are values found in the database for the bulk sample. The most significant change was with the β angle, where a reduction of around 1 deg. was found for all studied samples, while the difference between the specific Ga concentration was uncertain. As mentioned previously, the reduction of cell volume is a consequence of the changes of the Ga concentration, which can be seen especially well for the sample S4 with the 16 at.% of Ga. Furthermore, with a shift of the a parameter, a strong (100) crystallographic texture was found in CuO:Ga 2 O 3 films with 8 at.% (sample S2) and 16 at.% Ga (sample S3). Also, for those samples, an increase of peaks widths was found. Sensors 2020, 20, x FOR PEER REVIEW 10 of 18

Gas-Sensing Characteristics
The gas-sensing characteristics of the CuO-Ga2O3-based gas sensors under exposure to various An estimation of the size of crystallites was performed by applying a Scherrer formula. The Scherrer formula, linking coherence length (L coh ) together with the width of Braggs maxima (ω) found at 2θ angle for used wavelength λ, allowed us to estimate the size of crystallites: L coh = kλ/ωcos θ, where k is a Scherrer constant typically equal to 0.95 for thin films [102]. The samples S0 and S1 without and with small gallium concentration (i.e., pure CuO and CuO:Ga 2 O 3 with 4% at. Ga) had a mean grain size of 12-14 nm, while the S3 and S4 samples with 8 at.% and 16 at.%, respectively, showed a decrease of mean grain size to~7 nm. Furthermore, the mean size of gallium(III) oxide crystallites for those two samples was found to be even lower, at around 5 nm. The XRD analysis allowed us to estimate the amount of crystalline β-Ga 2 O 3 phase in the spectra to be of several percents, i.e., 3(1)% and 5(1)% for S2 and S3 samples with Ga concentration of 8 at.% and 16 at.%, respectively.

Gas-Sensing Characteristics
The gas-sensing characteristics of the CuO-Ga 2 O 3 -based gas sensors under exposure to various acetone concentrations in time function, with gas-in/gas-out phases, are presented in Figure 6a-d. As can be observed, the p-n heterostructure made of p-type CuO and n-type Ga 2 O 3 reacted in increasing the resistance when acetone was introduced to the measurement cell and decreased the resistance in synthetic air. Therefore, the sensor response was defined as the resistance ratio R gas /R air , where R gas and R air are electrical resistance under exposure to acetone and synthetic air, respectively. The sensor response as a function of acetone concentration at 300 • C and 50% RH is presented in Figure 7. As can be noticed, the doping effect of Ga 2 O 3 to pure CuO slightly increased the sensor response in the 0.1-1.25 ppm of acetone. However, the highest responses were obtained for samples with lower gallium oxide content, i.e., around~4% at. Therefore, further investigations will be focused on the experiments with 2-6% at. doping. Moreover, sample S1 exhibited 40% faster response time in comparison with pure CuO samples ( Figure 8). All response/recovery times are shown in Table 4. The resistance changes measured at various relative humidity concentrations are presented in Figure 9. The fact that the gas-sensing performance of thin film-based sensors is correlated with humidity is widely known. However, as can be noticed, the fabricated sensors remained stable in the full range of various relative humidity due to the p-n structure. The relative resistance changes in the 10-90% relative humidity range were around~30% for all samples. However, in the range, 50-90% of changes were below 5% which makes the developed sensors very attractive for utilization in the highly humidified samples, such as exhaled human breath analysis.  Sensors 2020, 20, x FOR PEER REVIEW 12 of 18 Figure 6. Sensor response as a function of acetone concentration at (a) 300 °C and 50% RH for CuO (sample S0), (b) CuO:Ga films deposited with 4% at. (sample S1); (c) 8% at. (sample S2); (d) 16% at.

Response and Recovery Times
The response and recovery times are presented in the Table 4 as well as an example of the calculation is given in Figure 8. Briefly, in the gas-sensing applications, the response and recovery time (s) is defined as the time to reach a 90% variation of the sensor signal. In this study, resistance under exposure to target gas and air, respectively. It has to be underlined, that the obtained times are quite high and further works should be focused on the reduction of response and recovery time (s), for example by using noble metal dopings.  Figure 8. The response and recovery times at 300 °C for CuO:Ga2O3 films (sample S1) with ~4% at. Ga at the exposition to 0.25 ppm acetone. Figure 9. The resistance changes of the fabricated sensors S0-S4 measured at 300 °C at various relative humidity levels.

Conclusions
The GLAD magnetron sputtering was used for CuO-Ga2O3 film deposition from the Cu:Ga mosaic target for gas sensor purposes. The sputtering conditions were presented for the thin CuO-Ga2O3 deposition with an average deposition rate of about 2.5 nm/min. The sample structure was investigated by XRD, XRR, and EDS measurements. The content of gallium was determined to be 4 at.% to 16 at.%, and film density was found to be 5.6 g/cm 3 regardless of the gallium Figure 8. The response and recovery times at 300 • C for CuO:Ga 2 O 3 films (sample S1) with~4% at. Ga at the exposition to 0.25 ppm acetone.
Sensors 2020, 20, x FOR PEER REVIEW 13 of 18 Figure 8. The response and recovery times at 300 °C for CuO:Ga2O3 films (sample S1) with ~4% at. Ga at the exposition to 0.25 ppm acetone. Figure 9. The resistance changes of the fabricated sensors S0-S4 measured at 300 °C at various relative humidity levels.

Conclusions
The GLAD magnetron sputtering was used for CuO-Ga2O3 film deposition from the Cu:Ga mosaic target for gas sensor purposes. The sputtering conditions were presented for the thin CuO-Ga2O3 deposition with an average deposition rate of about 2.5 nm/min. The sample structure was investigated by XRD, XRR, and EDS measurements. The content of gallium was determined to be 4 at.% to 16 at.%, and film density was found to be 5.6 g/cm 3 regardless of the gallium Figure 9. The resistance changes of the fabricated sensors S0-S4 measured at 300 • C at various relative humidity levels.

Response and Recovery Times
The response and recovery times are presented in the Table 4 as well as an example of the calculation is given in Figure 8. Briefly, in the gas-sensing applications, the response and recovery time (s) is defined as the time to reach a 90% variation of the sensor signal. In this study, resistance under exposure to target gas and air, respectively. It has to be underlined, that the obtained times are quite high and further works should be focused on the reduction of response and recovery time (s), for example by using noble metal dopings.

Conclusions
The GLAD magnetron sputtering was used for CuO-Ga 2 O 3 film deposition from the Cu:Ga mosaic target for gas sensor purposes. The sputtering conditions were presented for the thin CuO-Ga 2 O 3 deposition with an average deposition rate of about 2.5 nm/min. The sample structure was investigated by XRD, XRR, and EDS measurements. The content of gallium was determined to be 4 at.% to 16 at.%, and film density was found to be 5.6 g/cm 3 regardless of the gallium concentration. Diffraction patterns indicated Braggs peaks for the monoclinic phase of CuO for samples with a low Ga content, and CuO and β-Ga 2 O 3 monoclinic phases for samples with a higher Ga content. Also, the grain size changed from around 7 nm for a low content of gallium to 12-14 nm for samples with a higher content of gallium. For the measurements of the sensing properties of the CuO-Ga 2 O3 system, all investigated samples were able to detect acetone on levels of 1.25 ppm, 0.25 ppm, and 0.1 ppm. The signal values are presented in Figure 8. The response and recovery times were high, however, it has to be underlined that the results were related to the measurement setup, in which a high volume quartz-glass tube was used. The advantage of this type of sensor is its low humidity influence on the signal level. In conclusion, according to the CuO-Ga 2 O 3 system review and our results, the design was properly designed to detect very low level acetone concentration sensors on ppm and even below ppm levels.
Supplementary Materials: The following are available online at http://www.mdpi.com/1424-8220/20/11/3142/s1, Figure S1: Gallium/copper concentrations: (a) distribution map of sample S1 for Cu and Ga elements, (b) distribution map of sample S2 for Cu and Ga elements, (c) distribution map of sample S3 for Cu and Ga elements.

Conflicts of Interest:
The authors declare no conflict of interest.