Optical diagnostics of the magnetron sputtering process of copper in an argon–oxygen atmosphere

X-ray Diffraction (XRD) in Bragg–Brentano (θ–2θ) geometry was applied to study the crystal structure and phase composition of the samples. A panalytical Empyrean X-ray diffractometer with a Cu lamp was used to perform the measurements. The primary beam optics consisted of a programmable divergence slit (0.5°) and solar slits (2.29°). The secondary beam optics was composed of a soller, parallel plate collimator (0.18°) and a PIXcel3D® detector in the scanning line mode. A Ni B-filter was employed during measurements. The samples were scanned with a 0.026° step in the range of 20°–120° at room temperature.

Optical measurements were performed with the Perkin-Elmer double-beam spectrophotometer Lambda 19. The spectra of transmission coefficient and reflection coefficient were recorded within the wavelength range from 250 nm to 2000 nm with a step of 1 nm. The optical band gap of the deposited oxides was determined by the model given by formula (1): where, m takes a value of 2 for direct and ½ for indirect band gap, A is a proportional constant, hv is the incident photon energy, α is the absorption coefficient and E_g is the optical band gap. Here, the absorption coefficient α is directly related to the transmittance, reflectance and the thickness of the deposited thin films. The absorption coefficient (αhν) can be evaluated from the following relationship: where, T is the measured transmittance, R is the measured reflectance and d is the thickness of the deposited films.

The optical emission spectroscopy was realised by utilisation of the SPM2 diffraction grating monochromator with a resolution of 0.1 nm and the R928 Hamamatsu photomultiplier connected to the deposition chamber via a glass optical fibre. The OES measurement was controlled by dedicated software run on a PC. Based on previous experiments [24], the spectrum of plasma radiation was recorded over a range of 450–550 nm containing the most significant spectral lines of Cu [25].

The vacuum technological chamber was pumped out using a 450 l/s turbomolecular pump to a pressure of 5 × 10^–6 hPa; the pumping speed was then limited to 12 l/s and the working gases—argon and oxygen—were dosed using the MKS-1179B (MKS Instruments, USA) controllers. The total gas flow was constant and kept at 20 ml/min; the composition of the argon/oxygen mixture was changed in the range of 0–100% O₂, where 0% refers to 100%Ar and 0%O₂ and 100% refers to 0%Ar and 100%O_2, respectively. The pressure in the chamber during the sputtering process varied in the range of 1.8–3.0 × 10^–2 hPa, depending upon the argon/oxygen ratio. The copper target with a diameter of 80 mm was used as the source material and a DC-MF power supply operating in the current source regime was used. The material was deposited onto quartz glass substrates, and the deposition time was 20 min. The target–substrate distance was 45 mm, and the substrate temperature was set to 100 °C. The deposition of each sample was preceded by stabilising the process conditions for 20 min on the basis of previously obtained results [7, 15]. For the RBS experiments, the thickness was reduced to 80–160 nm and the thin films were deposited onto C foils (flexible graphite from GoodFellow) keeping the same deposition procedures as for films deposited on quartz glass substrates (Table 1).

The thickness of the obtained layers was determined using a Taylorstep profilometer (Rank-Taylor Hobson, UK). The measured thicknesses are summarised in Table 2. As can be observed, above 40% O₂ the thickness was constant and equal to approximately 240 nm ± 5%. It has to be pointed out that during the deposition a constant deposition time was used. As a result of the constant deposition time controlled by the power supply, samples of various thicknesses were obtained and the deposition rate as a function of oxygen concentration could, therefore, be calculated.

The chemical composition and layer thickness of the films were investigated by Rutherford Backscattering spectrometry. RBS measurements were performed using the 2 MeV He + ions at a backscattering angle of θ = 171° at the Institute of Nuclear Physics at Goethe University of Frankfurt am Main. The beam current was around 20 nA [26, 27]. The beam spot on the target had a circular shape with a diameter of 1–1.5 mm. Prior to the measurements, the beam energy calibration was performed on C and 5000 Å-thick Au calibration samples. The incident ion beam was directed along the normal to the sample surface. For data evaluation, the computer code SIMNRA was used [28]. For our RBS analysis [26, 27], we took into account the electronic stopping power data by Ziegler and Biersack, Chu + Yang’s theory for electronic energy loss straggling and Andersen’s screening function for the Rutherford cross section. The contribution from double and/or multiple scattering into the RBS spectra was taken into account using the calculating facilities of SIMNRA [29].

The RBS film thickness is determined from the energy loss of the projectile He + ions. It can also be estimated from the total number of counts of the signal (the peak area). The film thickness t is related to the energy difference ΔE (i.e. the energy width of the RBS signal) between the ions scattered from the front surface (KE₀; where E₀ is the primary energy and K is the kinetic factor) and those from the rear surface (E_1,t) of the film (i.e. from the film surface and those from the film-substrate or film-film interface) by the following formula [30]: where N is the film density and \(\overline{\varepsilon }\) is the stopping cross section (-factor).

The relationship between the film thickness and the total number of counts of the signal A₀ is [33] given as follows: where σ(E₀) is the Rutherford cross section, Ω is the detector solid angle, Q is the total number of incident particles hitting the target (charge), θ is the scattering angle and the (Nt) product is the areal density or the number of target atoms per unit area.

The RBS simulated spectrum from the SIMNRA program is for the (Nt) product (d_SIMNRA), which is in the unit of 10¹⁵ atoms/cm². The simulated areal density can then be converted into layer thickness value d (in nm) by using the mass density value of the bulk: where a is the conversion coefficient. The conversion coefficient a for a selected single element can also be found from SIMNRA. For instance, for Cu, the value for the conversion coefficient is a [1⋅10¹⁵ atoms/cm²] = 0.1179 [nm] estimated by using the bulk mass density ρ(Cu) = 8.951 g/cm³ (the estimated atomic density is δ(Cu) = 8.483⋅10²² atoms/cm³).

For the layer consisting of only pure metal (M) or its stoichiometric oxide (e.g. MO), the layer thickness d[nm] can be easily estimated according to Eq. (5). For a mixed layer consisting of, for example, metal and oxides, the type of oxides consisting in the layer can be defined from the estimated value of metal (M) and oxygen (O) content by SIMNRA on the basis of its percentage. For instance, the M and O content in M₂O are , respectively, 66.7% and 33%, while they amount to 50% and 50% for MO. In the case of a mixed layer consisting of m% M and n% MO, for example, the layer thickness was estimated as follows: where a₁ and a₂ are the conversion coefficients for M and MO, respectively. In most cases, the thickness could be estimated with a high degree of accuracy, if the layer either consists of only pure metal stoichiometric oxides or a mixture. Some difficulty arises in the thickness conversion for a layer consisting of a mixture of different oxides, but with a similar mass density because it is more difficult to estimate the exact ratio of different oxides based only on the ratio of the M content. It is important to remember, here, that, besides the uncertainty in thickness conversion, the ambiguity in determining the layer thickness is also related to the layer quality itself—the sputter-deposited films may have some columnar growth or defects, and thus, the mass density of the film is certainly different from that of the bulk.

One of the possible methods of verifying the deposited structures is X-ray diffraction, which allows determining the crystallographic structure and phase content. An XRD analysis was conducted on all the samples deposited at various argon/oxygen ratios (Table 2); however, for better data visualisation, only the spectra showing the structural change in samples are presented in Fig. 3. For thin film deposited in a 100% Ar gas atmosphere, four well-crystallised peaks—(111), (200), (220), (311)—related to Cu metallic phase can be seen in Fig. 3a. The crystal structure of copper is cubic close packed (ccp) and the space group is Fm3m space group [33, 34]. In addition, the less intense peak (111) which is assigned to the Cu₂O phase can be seen in the diffraction pattern for a thin film deposited in a 100% Ar gas atmosphere. The oxidation process of copper in air at room temperature after sample deposition might be responsible for oxygen contamination and the formation of the Cu₂O phase (cubic crystal structure, Pn3m space group) [34].

Choundhary et al. [35] reported their investigation results on the oxidation effects of copper metal that can occur at room temperature when the film is exposed to oxygen or ambient air. As a result of this exposure, a thin layer of native oxide is formed on the thin copper surface. The authors claimed that afterwards, the oxide growth slows down and effectively stops within a few nanometers because the oxygen atoms have too little energy to diffuse through the native oxide layer already formed at room temperature [35]. As the oxygen concentration increases, the crystal structure of the deposited thin films changes and only peaks related to the Cu₂O phase can be seen in Fig. 3b. Further increases of the oxygen concentration of up to 20% cause the formation of the CuO phase (Fig. 3c.) [36]. A thin film deposited in a gas atmosphere of 80% Ar + 20% O₂ is well crystallised and shows a strong (002) peak of monoclinic CuO phase (C2/c space group) [36]. For the sputtering process in pure oxygen, only two peaks assigned for the monoclinic CuO phase can be observed (Fig. 3d). The lower intensity of those two peaks might be due to the smaller thickness of that sample or might suggest a lower crystallinity of that thin film in comparison with a sample deposited in a gas atmosphere of 80% Ar + 20% O₂. The XRD analysis proves that by changing the composition of the argon/reactive gas mixture, it is possible to control the thin film phase composition from metallic Cu to CuO.

The magnetron sputtering deposition process is conducted in the vacuum chambers filled with reactive gases; therefore, as presented in Fig. 1, an instability effect can occur in situations where the deposition parameters such as working pressure, gas flow rate and cathode current are not well chosen or stabilised. During the experiments in this research, the abovementioned deposition parameters were set to keep the deposition in the stable mode as is presented in Fig. 4a (no hysteresis effect as shown in Fig. 1). As already mentioned, the reactive magnetron sputtering deposition process can be divided into three modes (metallic, transition and reactive) depending on the amount of reactive gas used. Figure 4a shows the experiment results, where the abovementioned modes are represented with green, blue and red lines, which refer to below 15%, in the 15–20% range, and above 20% of oxygen in the reactive gas mixture, respectively. In the transition mode, cuprous oxide (Cu₂O) structures can be obtained which exhibit the same gas-sensing behaviour as CuO [36]. Figure 4b shows the dependence of the deposition rate as a function of the O₂ concentration in the Ar/O₂ mixture during the magnetron sputtering deposition process. As can be observed, the deposition rate is constant (~ 12 nm/min) for 40–100% oxygen concentration in the argon/oxygen mixture, which is related to the target surface oxidation.

Su et al. [38] presented investigation results on the self-oxidation effect of Cu metallic targets during magnetron sputtering deposition including an in-depth discussion of the possible mechanism, avoidance and utilisation. The authors sputtered the Cu films in high vacuum systems and deposition was performed in an atmosphere of only argon. The results have shown that the main O₂ source for the oxidation of Cu was the residual O₂ in the deposition chamber as a result of the limited pumping process and the impurity of working gas [38]. Therefore, pure Cu films, Cu/Cu₂O composite films, pure Cu₂O films, Cu₂O/CuO composite films and pure CuO films could be obtained by simply controlling the deposition parameters. However, the target surface oxidation effect should be controlled to avoid any uncontrolled depositions and non-reproducibility of results.

Optical emission spectroscopy is based on the measurement of the intensity of spectral lines emitted as radiation in the plasma. Figure 5 shows the selected emission spectra for structures deposited at various oxygen content levels in the argon/oxygen mixture, including excited copper atoms, marked with an asterisk (Cu^*), and excited copper ions, marked with a plus (Cu⁺). The specified copper atoms and ions were observed at 510.5 nm, 515.3 nm, 512.8 nm, 495.0 nm and 502.0 nm. The peak revealing the presence of ionised O₂⁺ 525.7 nm oxygen molecules was observed for structures deposited in the oxide mode (for oxygen content above 20% in the Ar/O₂ mixture). It can be seen that the intensity of peaks is affected by O₂ concentration, as shown in Fig. 6.

Figure 6 shows the intensity of recorded spectral lines for various gas mixture compositions dispensed to the deposition vacuum chamber. There is a clear division of the deposition process into the three modes—metallic, transition and oxidized—associated with the surface conditions of the sputtered target. An increase in the intensity of the spectral line of ionised oxygen molecules (O₂⁺) between the depositions in the 80%/20% (Fig. 5c) and the 0%/100% (Fig. 5d) argon/oxygen mixtures was observed. In general, the magnetron sputtering deposition process under a fully oxidised atmosphere (100% O₂) is not common due to the lower deposition rate in comparison with Ar/O₂ mixtures. However, as presented in Fig. 4b, the deposition rate of the developed system for copper oxide depositions in the 40–100% O₂ is approximately constant. This can be explained by considering similar values of the atomic mass of argon (40 u) and the molecular mass of oxygen (32 u), which plays an important role in physical vapour deposition, such as in reactive magnetron sputtering technology.

The deposition rates of thin copper oxide films deposited by magnetron sputtering technology at various argon/oxygen mixture ratios were also investigated by Aghilizadeh et al. [39]. The obtained results confirmed that increasing the amount of oxygen in the reactive gas composition led to a decrease in the deposition rate. The Cu₂O phases were obtained for 10% and 20% oxygen in the argon/oxygen mixture, which is similar to the results presented in this paper. However, it has to be underlined that in [39] the DC mode was used, and in this paper, the DC-MF mode was used as well, as there were different deposition systems and different substrate materials. Therefore, the deposition parameters should be optimised for a specific application. Dolai et al. [40] reported the results of copper oxide thin films prepared by DC magnetron sputtering for the photovoltaic application, where the argon/oxygen ratio and the substrate temperature during depositions were key parameters. By contrast, Nayan et al. [41] presented the results of cuprous and cupric oxide thin films prepared by RF magnetron sputtering for gas-sensing applications, where apart from the argon/oxygen ratios and substrate temperature, substrate bias voltage was the key parameter. As a result, the electrical, optical and structural properties of the thin CuO films can be optimised for a specific application by various sets of deposition parameters, including target–substrate distance and sputtering geometry, such as glancing angle deposition [41].

A linear relationship between the material deposition rate and the copper spectral line intensity was found as presented in Fig. 7. Therefore, the OES can be used as a deposition rate monitor for the various cooper oxide thin film depositions. This can replace the conventional quartz monitors used in the sputtering technology to control the thicknesses of the deposited films. However, the observation of the intensity of Cu* 515 nm works well for monitoring the deposition rate of copper oxides, but not for pure metallic films, where other spectral lines should be chosen.

Spectrophotometry measurement results (transmittance and reflectance spectra) within the range of the fundamental absorption edge used to determine the band gap energy E_g (see Sects. 2 , 2.2) are presented in Fig. 8. The results show significant differences between the characteristics of films obtained at the lowest oxygen content in the sputtering atmosphere (10% O₂) and those obtained at higher oxygen contents. One can especially notice differences in the position of the transition from high to low transmittance. This can primarily be attributed to differences in the materials energy gap. The films obtained at a low oxygen content were found to be formed of Cu₂O, while films formed at a higher content of oxygen were of CuO. It is usually understood that the energy gap of Cu₂O is lower than the energy gap of CuO [42]. Looking at the transmittance presented in Fig. 8a, the opposite dependence is suggested. However, it has to be stressed that the presented results were measured on films of significantly different thicknesses as the deposition time was the same, but the deposition rate was very different (Sect. 2.5). The film thickness, as evaluated from the interference region of the transmission curves, are in agreement with the thicknesses measured by the mechanical profilometer; these values are summarised in Table 2. Based on the literature review, it can be stated that CuO mostly possesses a direct band gap, although some studies have reported that CuO has an indirect band gap. Table 3 summarises the energy band gap (Eg) values that can be found for CuO and Cu₂O which are the most common copper oxides.

As can be observed, the direct band gap varies in the range of 1.3–2.45 eV and 1.71–2.66 eV for CuO and Cu₂O, respectively. The calculated band gap values for copper oxide thin films deposited in this research at various oxygen content levels in the argon/oxygen mixture are in accordance with the results given in the literature, and some differences are related to various deposition parameters used during the deposition. Sai Guru Srinivasan et al. [57] presented the nanostructured copper oxide thin films deposited by RF magnetron sputtering at 5–50% oxygen content levels in the reactive gas. The films were further investigated for solar light-driven photocatalysis applications; the extensive optical properties were evaluated in that paper. The films deposited at 5% of O₂ partial pressure (Cu₂O) have a band gap of 2.12 eV and at 20–50% (CuO) the gap varies within the range of 1.79–1.82 eV [57]. Alajani et al. [58] reported their investigation results on Cu_xO_y thin films for solar cell applications, where the copper oxides were deposited at various argon/oxygen ratios; argon was set to 18 sccm and oxygen flow rates were 8, 9, 10, 11, 13, 15, and 17 sccm, which correspond to approximately 30.7%, 33.3%, 35.7%, 37.9%, 41.9%, 45.4% and 48.5%, respectively. The various mixtures allow the obtaining of various Cu_xO_y films such as Cu₄O₃, Cu₂O and CuO, which is confirmed by XRD and Raman. The energy band gap changes from 2.32 eV to 1.81 eV, and 2.32 eV was measured for samples deposited at 35.7% and 1.81 eV at 45.4% [58]. Similar results were reported by Hari Prasad Reddy et al. [59] where the copper oxides deposited in the RF magnetron sputtering technology at various O₂ partial pressures have various energy band gaps, i.e. as the oxygen partial pressure increased from 5·10^–3 to 2 × 10^–2 Pa, the optical band gap of the films shifted from 2.10 eV to 2.35 eV, and when increased to 8 × 10^–2 Pa, the E_g decreased to 1.96 eV, which is explained by the growth of CuO phase [59]. The effect of the energy band gap changes of the copper oxides deposited at various sputtering powers was investigated by Sibasankar Reddya et al. [60] The results have shown that the optical band gap of the films increased from 1.92 to 2.04 eV with an increase of sputtering power from 0.38 to 1.08 Wcm⁻². At a higher sputtering power of 1.50 Wcm⁻², the E_g decreased to 1.78 eV. The films exhibited lower band gap at a sputtering power of 0.38 Wcm⁻² and 1.50 Wcm⁻², which was due to the presence of CuO and metallic copper phases along with Cu₂O, respectively [60]. The recently presented results have shown that these deposition parameters have an impact on the optical properties; therefore, a direct comparison is not suitable.

The measured and simulated RBS spectra for various copper oxide thin films deposited at various argon/oxygen ratios are shown in Fig. 9. The films for RBS measurements were deposited on C foil, as mentioned in Sect. 2.5. Firstly, we discuss the results obtained for the thin films deposited at two different limits of the argon/oxygen mixture: in 100% argon gas and 100% oxygen.

The RBS spectrum of Cu film deposited in 100% of Ar gas [with the film notation (Cu/C, 100%Ar)] is characterised by a large Cu peak at an energy level of around 1320 keV from the film and a steep C edge from the substrate at 500 keV. We notice, here, that even if this film was deposited in the fully argon gas atmosphere without any supply of oxygen, traces of oxygen were detected as revealed by a small, but visible peak around 720 keV in the RBS spectra. The RBS results are in a strong agreement with XRD data obtained on the Cu films deposited on the quartz glass substrate; even for samples deposited in the pure argon atmosphere (0% oxygen content in the Ar/O₂ mixture), traces of oxygen contamination from the surface oxidation of films appeared. It is important to remember, here, that the RBS analysis generally provides the composition of different elements in the layer. The large Cu peak at around 1320 keV and especially the large O peak at around 720 keV from the film and the steep C edge from the substrate at 500 keV were revealed in the RBS spectrum of the film obtained from Cu deposition in pure oxygen gas [100% of oxygen, the film notation (CuO/C, 100% O₂)]. The RBS analysis indicated that the film is a homogeneous CuO layer, i.e. with the Cu–O ratio of 1:1 (or 50% Cu and 50% O concentration).

We observe, here, that for those two films, the peak width of the Cu peak is similar because the layer thickness is similar (87 nm and 92 nm, respectively). However, the peak intensity of the Cu peak for the Cu/C, 100% Ar film is visibly larger than that for the CuO/C, 100% O₂ film since the atomic amount of copper in the first film is twice as big as that in the second film (i.e. the Cu concentration is 100% in the Cu film and 50% in the CuO film). Similar to the (CuO/C, 100%O₂) film, the RBS spectra of the films deposited in the various argon/oxygen mixtures (of 95% Ar + 5% O₂, 90% Ar + 10% O₂ and 80% Ar + 20% O₂) exhibit the Cu peak and O peak from the film and the C edge from the substrate; the film notations are, respectively, (Cu₂O/C, 95%Ar + 5% O₂), (CuO/C, 90% Ar + 10% O₂) and (CuO/C, 80% Ar + 20% O₂). These films have an estimated layer thickness in the range of 140–170 nm (please refer to Sect. 2.4), and thus, the width of the Cu peak as well the O peak is visibly larger than that in the (Cu/C, 100% Ar) and (CuO/C, 100% O₂) film. Increasing the oxygen concentration in the gas mixture implies a systematic decrease of the Cu-peak intensity and an increase of the O-peak intensity revealing the decrease and increase, respectively, of the Cu and O content in the film. The RBS analysis of the film deposited in the 5% oxygen gas concentration (Cu₂O/C, 95% Ar + 5% O₂) indicates that the entire layer consisted of the Cu₂O phase. Increasing the oxygen concentration to 10%, our analysis indicates that the film (Cu₂O/C, 90% Ar + 10% O₂) consists of two layers: the surface layer of the Cu₂O phase (~ 22 nm) and the layer of the CuO phase (~ 146 nm). In this case, since the oxygen content is small (~ 10%), it is difficult to conclude from only RBS data whether it is related to the Cu₂O or the CuO phase. Taking into account the possibility of the existence of the Cu₂O phase determined from the XRD results, we can, then, estimate that the film consisted of 85% Cu and 15% Cu₂O phase. Increasing the oxygen concentration to 20%, the (CuO/C, 80% Ar + 20% O₂) film revealed the CuO composition in the entire film, which was in strong agreement with observation from XRD.

For all investigated films, a substantial carbon diffusion into the film was observed as revealed by an upturn on the right side of the C edge; this is similar to the situation which occurs with other films deposited by the magnetron sputtering technology on the C foils. The average carbon concentration in the film was estimated to be around 10%. It is originated from the out–diffusion of carbon from the substrate which is independent from the substrate temperature during deposition [27, 28]. However, the non-zero background signal between the Cu and O peak and between the O peak and the C edge indicate the presence of about 0.3% Cu and 3% O diffusion deep into the substrate. For all the copper oxides thin films, the estimated thickness from the RBS data analysis is about 7–15% larger than the nominal thickness measured by the mechanical profilometer (Sect. 2.4). As we mentioned earlier, the ambiguity in the determination of the layer thickness is related to the layer quality itself. For the films deposited by sputtering, the existence of possible porosities or defects would lead to a lower value of the mass density of the film in comparison with that of the bulk. Consequently, the conversion coefficient would be lower, and thus, the estimated thickness value would be lower than that obtained using the conversion coefficient based on the mass density of bulk.