GLAD Magnetron Sputtered Ultra-Thin Copper Oxide Films for Gas-Sensing Application

Copper oxide (CuO) ultra-thin films were obtained using magnetron sputtering technology with glancing angle deposition technique (GLAD) in a reactive mode by sputtering copper target in pure argon. The substrate tilt angle varied from 45 to 85° and 0°, and the sample rotation at a speed of 20 rpm was stabilized by the GLAD manipulator. After deposition, the films were annealed at 400 °C/4 h in air. The CuO ultra-thin film structure, morphology, and optical properties were assessed by X-ray diffraction (XRD), energy-dispersive X-ray spectroscopy (EDX), X-ray reflectivity (XRR), and optical spectroscopy. The thickness of the films was measured post-process using a profilometer. The obtained copper oxide structures were also investigated as gas-sensitive materials after exposure to acetone in the sub-ppm range. After deposition, gas-sensing measurements were performed at 300, 350, and 400 °C and 50% relative humidity (RH) level. We found that the sensitivity of the device is related to the thickness of CuO thin films, whereas the best results are obtained with an 8 nm thick sample.


Introduction
Gas sensors have been investigated over the last few decades, and the first gas sensor was introduced to the market in 1968 by Figaro Company. It is still available. Since then, the gas sensor market has increased year by year, and it is expected to be worth USD 1.4 billion in 2024 [1]. Generally, a gas sensor consists of gas-sensitive films and gas sensor substrate with electrodes, a package, and front-end electronic circuits. The gas-sensitive layer can be realized based on various materials, including organic compounds (e.g., phthalocyanines [2,3]) and metal oxides [4] (e.g., WO 3 [5,6], TiO 2 [7,8], SnO 2 [9,10], In 2 O 3 [11,12], Fe 2 O 3 [13,14], MoO 3 [13,15], ZnO [16][17][18], CuO [19][20][21][22][23][24][25][26][27][28]). The most common methods for metal oxide depositions are magnetron sputtering [29,30], sol-gel [31,32], thermal oxidation [33,34], hydrothermal techniques [35,36], the spray pyrolysis technique [37,38], and the microwave-assisted method [39,40]. Among them, magnetron sputtering is widely accepted for industrial purposes, because it can be easily adapted to Complementary Metal Oxide Semiconductor (CMOS) technology; therefore, front-end electronics compounds can be realized by using the same technology. Gas sensors are characterized by the "3-S" parameters: sensitivity, selectivity, and stability. Gas-sensitive layers play an important role in all of these parameters. For instance, sensitivity is related to the surface-volume ratio of the gas-sensitive layer; therefore, the nanocolumnar structure During these experiments, a copper target (Kurt J. Lesker Company) of 50 mm diameter (5N purity) was employed for reactive sputtering at various argon/oxygen mixtures, such as 66%O2/34%Ar, 80%O2/20%Ar, 100%O2/0%Ar. The flows of argon and oxygen were controlled by mass flow controllers (MFC) 1179B (MKS Instruments (Andover, MA, USA). The substrate tilt angle varied from 45° to 85° and 0° (Table 1), and the sample rotation at a speed of 20 rpm was stabilized by the ECR manipulator. The CuO thin films were deposited onto silicon substrates (Si-Mat, Kaufering, Germany), quartz JGS-2 (Continental Trade, Warsaw, Poland), and BVT (Praha, Czech Republic) for XRD, optical, SEM, and gas-sensing measurements, respectively.  Figure 1b shows a schematic drawing of the GLAD deposition system. The system has been pre-pumped to a vacuum level of 5 × 10 −6 mbar and then to a deposition vacuum level of 3 × 10 −2 mbar. The copper target was first presputtered in pure argon for 10 min at 50 W to remove any contamination, and then, oxygen was introduced into the chamber. After the presputtering processes, the power and temperature were fixed at 50 W and 400 °C, respectively. The sputtering time was adjusted to deposit films with various thicknesses. The copper oxide thin films were fabricated with DC-MF sputtering mode by a power supply from DORA Power System. After deposition, the samples were annealed at 400 °C/4 h in air. During these experiments, a copper target (Kurt J. Lesker Company) of 50 mm diameter (5N purity) was employed for reactive sputtering at various argon/oxygen mixtures, such as 66%O 2 /34%Ar, 80%O 2 /20%Ar, 100%O 2 /0%Ar. The flows of argon and oxygen were controlled by mass flow controllers (MFC) 1179B (MKS Instruments (Andover, MA, USA). The substrate tilt angle varied from 45 • to 85 • and 0 • (Table 1), and the sample rotation at a speed of 20 rpm was stabilized by the ECR manipulator. The CuO thin films were deposited onto silicon substrates (Si-Mat, Kaufering, Germany), quartz JGS-2 (Continental Trade, Warsaw, Poland), and BVT (Praha, Czech Republic) for XRD, optical, SEM, and gas-sensing measurements, respectively.  Figure 1b shows a schematic drawing of the GLAD deposition system. The system has been pre-pumped to a vacuum level of 5 × 10 −6 mbar and then to a deposition vacuum level of 3 × 10 −2 mbar. The copper target was first presputtered in pure argon for 10 min at 50 W to remove any contamination, and then, oxygen was introduced into the chamber. After the presputtering processes, the power and temperature were fixed at 50 W and 400 • C, respectively. The sputtering time was adjusted to deposit films with various thicknesses. The copper oxide thin films were fabricated with DC-MF sputtering mode by a power supply from DORA Power System. After deposition, the samples were annealed at 400 • C/4 h in air.

XRD/XRR Measurements
X-ray diffractometry and X-ray reflectometry were obtained by utilization of X'Pert PRO MPD PANalytical system with a grazing angle of 3 • using monochromatic Cu Kα1 radiation. The thickness and density calculations were performed by fitting the XRR data using dedicated software. The film thickness and density were obtained by the period of oscillations in the XRR curve and the position of the total reflection edge, respectively.

SEM/EDX Measurements
The structure of the CuO films was observed in a scanning electron microscope equipped with an electron backscatter diffraction mode (Nova NANOSEM 200, FEI, Hillsboro, OR, USA) with an attachment for the detection of characteristic X-ray radiation (EDS, EDAX Genesis).

Optical Measurements
Optical measurements were carried out with the Perkin-Elmer double-beam spectrophotometer Lambda 19. The spectra of the transmission coefficient were recorded within the wavelength range from 300 to 2200 nm with a step of 1 nm.

Thickness Measurements
The CuO film thicknesses determination was carried out using a Taly-step profilometer (Taylor Hobson, Leicester, UK.)

Gas-Sensing Setup
The gas-sensing setup was previously described in [51,52,56]. Briefly, the developed sensors ( Figure 2a,b) with various copper oxides serve as the gas-sensitive layers, which were placed in the quartz-tube oven, and the target gas at various concentrations was introduced.
X-ray diffractometry and X-ray reflectometry were obtained by utilization of X'Pert PRO MPD PANalytical system with a grazing angle of 3° using monochromatic Cu Kα1 radiation. The thickness and density ρ calculations were performed by fitting the XRR data using dedicated software. The film thickness and density were obtained by the period of oscillations in the XRR curve and the position of the total reflection edge, respectively.

SEM/EDX Measurements
The structure of the CuO films was observed in a scanning electron microscope equipped with an electron backscatter diffraction mode (Nova NANOSEM 200, FEI, Hillsboro, OR, USA) with an attachment for the detection of characteristic X-ray radiation (EDS, EDAX Genesis).

Optical Measurements
Optical measurements were carried out with the Perkin-Elmer double-beam spectrophotometer Lambda 19. The spectra of the transmission coefficient were recorded within the wavelength range from 300 to 2200 nm with a step of 1 nm.

Thickness Measurements
The CuO film thicknesses determination was carried out using a Taly-step profilometer (Taylor Hobson, Leicester, UK.)

Gas-sensing Setup
The gas-sensing setup was previously described in [51,52,56]. Briefly, the developed sensors ( Figure 2a,b) with various copper oxides serve as the gas-sensitive layers, which were placed in the quartz-tube oven, and the target gas at various concentrations was introduced. Because the measurements were based on the measurements of the electrical resistance changes, an electrometer (34401A HP) was used to record the electrical resistance of the sensor under exposure to air and target gas. Therefore, the gas sensor response (S) was defined as the resistance ratio S = (Rgas-Rair)/Rair, where Rgas and Rair are electrical resistances in gas and air, respectively. By using specialized mass flow controllers 1179B (MKS Instruments, Andover, MA, USA), different concentration levels were obtained. A gas mixture (Air Products, Hersham, UK) containing 5 ppm of acetone, 20.944% mol of oxygen, and 79.054% mol of nitrogen was used for the measurements. Measurements to detect gas were carried out at various temperatures and 50% relative humidity (RH) level. All the methods used to characterize the copper oxide layers obtained by means of the GLAD technique are shown in Table 2.  Because the measurements were based on the measurements of the electrical resistance changes, an electrometer (34401A HP) was used to record the electrical resistance of the sensor under exposure to air and target gas. Therefore, the gas sensor response (S) was defined as the resistance ratio S = (R gas − R air )/R air , where R gas and R air are electrical resistances in gas and air, respectively. By using specialized mass flow controllers 1179B (MKS Instruments, Andover, MA, USA), different concentration levels were obtained. A gas mixture (Air Products, Hersham, UK) containing 5 ppm of acetone, 20.944% mol of oxygen, and 79.054% mol of nitrogen was used for the measurements. Measurements to detect gas were carried out at various temperatures and 50% relative humidity (RH) level. All the methods used to characterize the copper oxide layers obtained by means of the GLAD technique are shown in Table 2.

Material Characterization Results
The crystallographic structure of the deposited gas-sensing layers was determined with X-ray diffraction and reflectivity. However, it is important to note that ultra-thin films were deposited; therefore, the special adapter was used by the XRD system to measure the diffraction of deposited films at the grazing angle of 3 • using monochromatic Cu Kα 1 radiation. Figure Figure 4 shows the XRR measurement results for copper oxide film deposited at a 70 • angle and 80% content of O 2 in the reactive atmosphere of argon/oxygen. The thickness calculated from the XRR results is in agreement with the thickness measured by the mechanical profilometer, i.e., 11.05 nm and 12.00 nm, respectively. The difference can be observed between the density value estimated from measurement and from the theoretical, i.e., 5.66 g/cm 3 and 6.31 g/cm 3 . However, it has to be taken into account that the theoretical density of CuO was obtained for CuO powders [57,58]. Shehayeb et al. [59] found the variation of density values for CuO films deposited by an electrophoretic deposition (EPD) based on dispersed CuO nanoparticles, where density was in the range of~3.3-4.5 g/cm 3 . Chuagan et al. [60] presented the investigation results on nanostructured CuO prepared via sol-gel method, where density ranged from 2.16 to 2.56 g/cm 3 .     Apart from gas-sensing applications, CuO is also an attractive material for optical applications; therefore, the basic optical properties, such as transmission and reflectivity, were measured. The results are presented in Figure 5. As can be observed, the transmission of the films increased from 20% to 90%. The highest transmission was obtained from the thinnest films deposited with the 85°, 80°, and 75° substrate tilt angles. The reflectivity of all the obtained CuO films was less than 20% and higher than 7% through most of the spectral region. The highest reflectivity was assigned to the thin films deposited with 0°, 45°, and 70° substrate tilt angles. Apart from gas-sensing applications, CuO is also an attractive material for optical applications; therefore, the basic optical properties, such as transmission and reflectivity, were measured. The results are presented in Figure 5. As can be observed, the transmission of the films increased from 20% to 90%. The highest transmission was obtained from the thinnest films deposited with the 85 • , 80 • , and 75 • substrate tilt angles. The reflectivity of all the obtained CuO films was less than 20% and higher than 7% through most of the spectral region. The highest reflectivity was assigned to the thin films deposited with 0 • , 45 • , and 70 • substrate tilt angles.  The morphology of all the copper oxide films was investigated using scanning electron microscopy. Each film showed the same morphology. As a result, in Figure 6, only one example was shown. The films were deposited using the GLAD technique; therefore, the shadowing effect can be observed. The main aim of the paper was to verify the acetone detection by ultra-thin copper oxides, and therefore, 8-15 nm thin films were deposited (refer to Figure 8). Increasing the deposition time would lead to increasing the thickness, as shown schematically in Figure 6. The GLAD technique allows for the possibility of controlling the nanocolumnar growth, including the spacing between the nucleation as presented in the SEM photo. However, further optimization is needed, where the substrate tilt angle, substrate rotation, substrate temperature, argon/oxygen mixture, and other deposition parameters are selected.  The morphology of all the copper oxide films was investigated using scanning electron microscopy. Each film showed the same morphology. As a result, in Figure 6, only one example was shown. The films were deposited using the GLAD technique; therefore, the shadowing effect can be observed. The main aim of the paper was to verify the acetone detection by ultra-thin copper oxides, and therefore, 8-15 nm thin films were deposited (refer to Figure 8). Increasing the

Gas-Sensing Characteristics
Working temperature is one of the main factors that affect the response of gas sensors based on metal oxides. Figure 7 shows the sensor response (R gas /R air ) measured at 0.25, 1.25, and 2.5 ppm of acetone vs. the operating temperature for copper oxides deposited at 66%O 2 /34%Ar and 80%O 2 /20%Ar, respectively (Figure 7a), and 100%O 2 content (Figure 7b) in the reactive atmosphere. This shows that the highest responses were obtained at 350 • C for layers deposited at 100%O 2 (fully reactive mode). Generally, 350 • C was the operating temperature with the highest responses; however, in some cases, the results obtained indicated that an operating temperature above 400 • C could give better results. Due to the limitation of the gas-sensing system, such measurements could not be carried out. Moreover, from a practical point of view, the operating temperature should be as low as possible to reduce the power consumption needed to heat up the sensors. A solution to reduce the power consumption could be found in the literature [62,63]. The reaction of oxygen with an adsorption site results in Oand the generation of a hole, which leads to an accumulation layer at the surface of the p-type conducting metal oxides, such as CuO [64]. The common model for the detection of reducing gases is based on the decrease of oxygen species at the surface leading to an increase of resistance, which was observed when the copper oxide-based sensor was exposed to acetone [65,66].
As was already mentioned, the main goal of the experiment was to define the sensor response based on the ultra-thin copper oxide layer under exposure to sub-ppm acetone concentration. Therefore, the gas-sensitive layers were deposited in the 4-15 nm range, and the responses are presented in Figure 8. It is worth noting that the most sensitive layers to acetone were those with a thickness of Coatings 2020, 10, 378 8 of 13 8 nm, which worked at a temperature of 350 • C. Therefore, this thickness was selected for further experiments. The explanation for decreasing the sensor response when the thickness was increasing could lie in the film's deposition behavior. For lower thicknesses, slow nanocolumnar growth was observed with strictly separated nucleation islands. When the deposition took longer, the thicknesses increased, but at the same time, new nucleation appeared close to the previous one, which resulted in the reduced surface to volume ratio of the gas-sensitive layer and, in fact, reduced sensor response. The effect will reverse above a certain thickness, known as a threshold thickness, and the effect is related to the material that is used, gas sensor substrate, and deposition conditions. Therefore, the threshold thickness can only be experimentally determined.
Working temperature is one of the main factors that affect the response of gas sensors based on metal oxides. Figure 7 shows the sensor response (Rgas/Rair) measured at 0.25, 1.25, and 2.5 ppm of acetone vs. the operating temperature for copper oxides deposited at 66%O2/34%Ar and 80%O2/20%Ar, respectively (Figure 7a), and 100%O2 content (Figure 7b) in the reactive atmosphere. This shows that the highest responses were obtained at 350 °C for layers deposited at 100%O2 (fully reactive mode). Generally, 350 °C was the operating temperature with the highest responses; however, in some cases, the results obtained indicated that an operating temperature above 400 °C could give better results. Due to the limitation of the gas-sensing system, such measurements could not be carried out. Moreover, from a practical point of view, the operating temperature should be as low as possible to reduce the power consumption needed to heat up the sensors. A solution to reduce the power consumption could be found in the literature [62,63]. The reaction of oxygen with an adsorption site results in Oand the generation of a hole, which leads to an accumulation layer at the surface of the p-type conducting metal oxides, such as CuO [64]. The common model for the detection of reducing gases is based on the decrease of oxygen species at the surface leading to an increase of resistance, which was observed when the copper oxide-based sensor was exposed to acetone [65,66]. As was already mentioned, the main goal of the experiment was to define the sensor response based on the ultra-thin copper oxide layer under exposure to sub-ppm acetone concentration. Therefore, the gas-sensitive layers were deposited in the 4-15 nm range, and the responses are presented in Figure 8. It is worth noting that the most sensitive layers to acetone were those with a thickness of 8 nm, which worked at a temperature of 350 °C. Therefore, this thickness was selected for further experiments. The explanation for decreasing the sensor response when the thickness was increasing could lie in the film's deposition behavior. For lower thicknesses, slow nanocolumnar growth was observed with strictly separated nucleation islands. When the deposition took longer, the thicknesses increased, but at the same time, new nucleation appeared close to the previous one, which resulted in the reduced surface to volume ratio of the gas-sensitive layer and, in fact, reduced Coatings 2020, 10, x FOR PEER REVIEW 9 of 13 sensor response. The effect will reverse above a certain thickness, known as a threshold thickness, and the effect is related to the material that is used, gas sensor substrate, and deposition conditions. Therefore, the threshold thickness can only be experimentally determined. As mentioned in the Introduction, higher acetone levels are characteristic of diabetics. For healthy people, the level of acetone in exhaled air is below 1 ppm. In contrast, diabetic acetone levels range from 1.5 to 2.5 ppm. Based on the literature review, both proposed regions are not strictly defined; therefore, the acetone detection has to be carried out in a wide range of concentrations. It should be pointed out that commercially available gas sensors have a limit of detection in tens of ppm for acetone (i.e., Figaro TGS 822: 50-5000 ppm). Figure 9 presents the gas sensor calibration curves. The various gas-sensitive layers (66%O2/34%Ar, 80%O2/20%Ar, and 100%O2 refer to the content of oxygen in the argon/oxygen mixture during the layers deposition) were investigated. As can be observed, the limit of detection was 0.25 ppm, which covers the acetone concentration in the exhaled human breath for healthy people, which is generally in the 0.3-0.9 ppm range. As mentioned in the Introduction, higher acetone levels are characteristic of diabetics. For healthy people, the level of acetone in exhaled air is below 1 ppm. In contrast, diabetic acetone levels range from 1.5 to 2.5 ppm. Based on the literature review, both proposed regions are not strictly defined; therefore, the acetone detection has to be carried out in a wide range of concentrations. It should be pointed out that commercially available gas sensors have a limit of detection in tens of ppm for acetone (i.e., Figaro TGS 822: 50-5000 ppm). Figure 9 presents the gas sensor calibration curves. The various gas-sensitive layers (66%O 2 /34%Ar, 80%O 2 /20%Ar, and 100%O 2 refer to the content of oxygen in the argon/oxygen mixture during the layers deposition) were investigated. As can be observed, the limit Coatings 2020, 10, 378 9 of 13 of detection was 0.25 ppm, which covers the acetone concentration in the exhaled human breath for healthy people, which is generally in the 0.3-0.9 ppm range. healthy people, the level of acetone in exhaled air is below 1 ppm. In contrast, diabetic acetone levels range from 1.5 to 2.5 ppm. Based on the literature review, both proposed regions are not strictly defined; therefore, the acetone detection has to be carried out in a wide range of concentrations. It should be pointed out that commercially available gas sensors have a limit of detection in tens of ppm for acetone (i.e., Figaro TGS 822: 50-5000 ppm). Figure 9 presents the gas sensor calibration curves. The various gas-sensitive layers (66%O2/34%Ar, 80%O2/20%Ar, and 100%O2 refer to the content of oxygen in the argon/oxygen mixture during the layers deposition) were investigated. As can be observed, the limit of detection was 0.25 ppm, which covers the acetone concentration in the exhaled human breath for healthy people, which is generally in the 0.3-0.9 ppm range. The gas-sensing characteristics of CuO-based gas sensors under exposure to various acetone concentrations in time function, with gas-in/gas-out phases, are presented in Figure 10a-c. Figure  10d presents the influence of the resistance changes to various relative humidity concentrations, which influences the gas-sensing behavior of the metal oxide-based gas sensors. The fact that the The gas-sensing characteristics of CuO-based gas sensors under exposure to various acetone concentrations in time function, with gas-in/gas-out phases, are presented in Figure 10a-c. Figure 10d presents the influence of the resistance changes to various relative humidity concentrations, which influences the gas-sensing behavior of the metal oxide-based gas sensors. The fact that the gas-sensing performance of thin film-based sensors is correlated with humidity is widely known. The effect of RH will decrease the total gas sensor response. Moreover, the relative humidity is above 80% in the exhaled breath, and therefore, it always has to be taken into account.
Coatings 2020, 10, x FOR PEER REVIEW 10 of 13 gas-sensing performance of thin film-based sensors is correlated with humidity is widely known. The effect of RH will decrease the total gas sensor response. Moreover, the relative humidity is above 80% in the exhaled breath, and therefore, it always has to be taken into account.

Conclusions
In this paper, the investigation results on copper oxide ultra-thin films deposited by a commercially available glancing angle deposition system (magnetron sputtering technology at various conditions) are presented and discussed. The deposited thin films were used as gas-

Conclusions
In this paper, the investigation results on copper oxide ultra-thin films deposited by a commercially available glancing angle deposition system (magnetron sputtering technology at various conditions) are presented and discussed. The deposited thin films were used as gas-sensitive layers for sub-ppm acetone detection, which is considered to be a biomarker of diabetes present in exhaled breath. The CuO ultra-thin film structure, morphology, and optical properties were assessed by X-ray diffraction, energy-dispersive X-ray spectroscopy, X-ray reflectivity, and optical spectroscopy. The optimal thickness and optimal operating temperatures were 8 nm and 350 • C, respectively. The highest responses were obtained for samples deposited in the fully reactive mode: 100% oxygen during the magnetron sputtering deposition. It was confirmed that sensors can detect acetone as low as 0.25 ppm and that they work well in the 0.25-2.5 ppm range, which covers the exhaled acetone range. However, exhaled breath is fully humidified, and relative humidity above 80% decreases the total gas sensor response. Therefore, relative humidity needs to be reduced by, for instance, applying a humidity trap.