Semiconducting p-Type Copper Iron Oxide Thin Films Deposited by Hybrid Reactive-HiPIMS + ECWR and Reactive-HiPIMS Magnetron Plasma System

: A reactive high-power impulse magnetron sputtering (r-HiPIMS) and a reactive high-power impulse magnetron sputtering combined with electron cyclotron wave resonance plasma source (r-HiPIMS + ECWR) were used for the deposition of p-type CuFe x O y thin ﬁlms on glass with SnO 2 F conductive layer (FTO). The aim of this work was to deposit CuFe x O y ﬁlms with di ﬀ erent atomic ratio of Cu and Fe atoms contained in the ﬁlms by these two reactive sputtering methods and ﬁnd deposition conditions that lead to growth of ﬁlms with maximum amount of delafossite phase CuFeO 2 . Deposited copper iron oxide ﬁlms


Introduction
Generally, highly efficient materials for photocathode thin film applied in solar water splitting cells are intensively searched.Key factors are low cost, high photoactivity and long-term stability.These materials should have suitable value of bang gap E g compatible with the solar spectrum, high absorption coefficient, p-type conductivity and good transport properties for photogenerated holes and electrons.One suitable candidate for efficient photocathode is Cu 2 O with direct band gap E g = 2 eV [1] and p-type conductivity.These Cu 2 O films were used as effective photocathode in photoelectrochemical (PEC) cells but the stability problem was solved with additional layers ZnO:Al and TiO 2 made by atomic layer deposition (ALD) in order to improve stability during PEC measurement [2].Because this approach is technologically complicated further approaches was tested.Some progress was achieved for films CuO.Cupric oxide CuO is p-type semiconductor material with band gap E g = 1.2-2.5 eV and high absorption coefficient.CuO with increased stability was applied as photocathode and presented in reference [3].In this work the improved stability was achieved with oxygen excess contained in films and with deposition of Au-Pd nanostructures on the surface of CuO films.Further approach for improvement PEC performance and stability of these photocathodes is to deposit CuFeO 2 with delafossite structure.This material has suitable band gap E g = 1.47 eV and good stability in alkaline electrolytes which is used in PEC cells [4,5].One of the problems with delafossite is the poor photoinduced electron-hole separation and charge collection [6].Further improvement of PEC performance of this material is to deposit a CuFeO 2 /CuO composite structure with heterojunction between those two phases helping to improve PEC performance as photocathodes [7].
Recently, reactive high-power impulse magnetron sputtering (r-HiPIMS) has become promising technology for various material preparation [18][19][20][21][22][23][24][25].The magnetron discharge is generated in short pulses (10-100 µs) with low frequency of pulsing (100 Hz-1 kHz).The duty factor is very low in r-HiPIMS, and although the high power is applied in the discharge plasma in the active part of the pulse, the average applied power can be low with similar values like in classical DC magnetron sputtering [24].This method relies on the intensive ionization of sputtered and reactively sputtered particles in high current discharge pulse [26][27][28], and the growing films is built from ions rather than from neutrals.
In order to extend possible conditions in r-HiPIMS plasma, a hybrid plasma source based on the combination of r-HiPIMS with RF inductively coupled plasma working in electron cyclotron wave resonance (ECWR) was developed [29][30][31] and tested for the deposition of TiO 2 textured rutile [32] and Fe 2 O 3 with a hematite phase [33].The possibility to operate the r-HiPIMS magnetron plasma at very low-pressure p < 0.05 Pa, further increase of ionization of sputtered particles, higher electron temperature T e during the active part of discharge pulse are great benefits of this hybrid r-HiPIMS + ECWR deposition system [31].
In this paper, the study of hybrid r-HiPIMS + ECWR plasma system and pure reactive r-HiPIMS with two magnetrons cathodes is presented together with RF probe plasma diagnostics during the deposition of copper iron oxide thin films.These films were investigated as photocathodes in a PEC solar water splitting system.The aim of this work was to deposit CuFe x O y films with different atomic ratio of Cu and Fe atoms contained in the films by these two reactive sputtering methods.The aim Coatings 2020, 10, 232 3 of 14 was to study this ternary oxide for different molar ratio Cu and Fe contained in the film and to find deposition conditions which lead to growth of films with maximum amount of delafossite phase CuFeO 2 .

Experimental
Copper iron oxide thin films were deposited on fluorine doped tin oxide (FTO) conductive glass substrates.This is the glass substrate with 500 nm thick SnO 2 :F conductive layer with the sheet resistance 15 Ω/ .The hybrid r-HiPIMS + ECWR deposition system can be seen in Figure 1.The reactor stainless steel chamber was continuously pumped by the turbomolecular (500 L s −1 ) as a secondary pump and by the rotary vane pump (25 m 3 •h −1 ) as primary pump on the base pressure p b = 1 × 10 −4 Pa.Argon and oxygen gas flow Q Ar and Q o2 were fed into the reactor through the gas flow controller and the pressure p in the chamber was controlled during the deposition by the control gate valve and was held for all the deposition conditions on the value p = 1 Pa.Argon gas flow was held constant on the value Q Ar = 26 sccm for all the deposition conditions.The gas flow of oxygen for particular conditions is shown in Table 1.The system was equipped with two magnetrons with circular planar cathodes with diameter 50 mm.The first cathode was made of pure iron (99.9%) with thickness 1.5 mm and the second one was made of pure copper (99.9%) with thickness 6 mm.The substrate was electrically floating and was placed 150 mm from the surface of both targets.The substrate was heated during the deposition with controlled temperature up to 500 • C. The temperature of the substrate at particular deposition conditions are in Table 1.The substrate was rotating during the process with rotation speed 1 rpm.The water-cooled RF ICP coil was placed between magnetrons and the substrate (Figure 1).The diameter of the ICP coil was 200 mm, and the length was 120 mm.The power fed to this ICP coil was P ECWR , and the values for particular conditions are in Table 1.The DC magnetic field was set to take the system as close as to ECWR resonance and the magnetic field was around B ≈ 1.7 mT.Deposition time was always 60 min for all the deposited samples.
Coatings 2020, 10, 232 3 of 14 deposition conditions which lead to growth of films with maximum amount of delafossite phase CuFeO2.

Experimental
Copper iron oxide thin films were deposited on fluorine doped tin oxide (FTO) conductive glass substrates.This is the glass substrate with 500 nm thick SnO2:F conductive layer with the sheet resistance 15 Ω/□.The hybrid r-HiPIMS + ECWR deposition system can be seen in Figure 1.The reactor stainless steel chamber was continuously pumped by the turbomolecular (500 L s −1 ) as a secondary pump and by the rotary vane pump (25 m 3 •h −1 ) as primary pump on the base pressure pb = 1 × 10 −4 Pa.Argon and oxygen gas flow QAr and Qo₂ were fed into the reactor through the gas flow controller and the pressure p in the chamber was controlled during the deposition by the control gate valve and was held for all the deposition conditions on the value p = 1 Pa.Argon gas flow was held constant on the value QAr = 26 sccm for all the deposition conditions.The gas flow of oxygen for particular conditions is shown in Table 1.The system was equipped with two magnetrons with circular planar cathodes with diameter 50 mm.The first cathode was made of pure iron (99.9%) with thickness 1.5 mm and the second one was made of pure copper (99.9%) with thickness 6 mm.The substrate was electrically floating and was placed 150 mm from the surface of both targets.The substrate was heated during the deposition with controlled temperature up to 500 °C.The temperature of the substrate at particular deposition conditions are in Table 1.The substrate was rotating during the process with rotation speed 1 rpm.The water-cooled RF ICP coil was placed between magnetrons and the substrate (Figure 1).The diameter of the ICP coil was 200 mm, and the length was 120 mm.The power fed to this ICP coil was PECWR, and the values for particular conditions are in Table 1.The DC magnetic field was set to take the system as close as to ECWR resonance and the magnetic field was around B ≈ 1.7 mT.Deposition time was always 60 min for all the deposited samples.Typical r-HiPIMS and r-HiPIMS + ECWR discharge waveforms are synchronized and can be seen in Figures 2 and 3, respectively.The length of pulses, their relative positions, and their pulsing frequencies were held constant for all investigated conditions.The "ON" time on the iron cathode was TONFe = 100 μs, and the "ON" time on the copper cathode was TONCu = 20 μs for samples 5x-10x (Table 1).TONCu = 30 μs for the sample 4x, 3x, TONCu = 100 μs for the sample 2x and TONCu = 50 μs for the sample 1x.The pulsing frequency on both cathodes was always Tp = 100 Hz.Typical r-HiPIMS and r-HiPIMS + ECWR discharge waveforms are synchronized and can be seen in Figures 2 and 3, respectively.The length of pulses, their relative positions, and their pulsing frequencies were held constant for all investigated conditions.The "ON" time on the iron cathode was T ONFe = 100 µs, and the "ON" time on the copper cathode was T ONCu = 20 µs for samples 5x-10x (Table 1).T ONCu = 30 µs for the sample 4x, 3x, T ONCu = 100 µs for the sample 2x and T ONCu = 50 µs for the sample 1x.The pulsing frequency on both cathodes was always T p = 100 Hz.  1.
Transmission spectra of selected deposited films on FTO glass were measured by spectrometer (Carry 100, Varian, Palo Alto, CA, USA) with integrating sphere in the spectral range 200-900 nm.The thickness of the films was determined by profilometry measurements (Alpha-Step 500 KLA Tencor MDL.no.ASIQ, San Jose, CA, USA).The deposited films were analyzed by X-ray diffraction (XRD).XRD data were measured at room temperature on a Bruker AXS D8 powder diffractometer (Billerica, MA, USA) in a Bragg-Brentano geometry using a CoKα radiation wavelength (λ = 1.7903Å, U = 34 kV, I = 30 mA).Chemical composition regarding molar ratio of Cu/Fe was measured by XRF.ED-XRF spectrometer Niton XL5 (Thermo Scientific, Waltham, MA, USA) equipped with Ag anode X-Ray Tube (6-50 kV, 0-500 μA, max 5 W) was used.Measure area diameter was 3 mm.Niton Connect PC software (Version 1.0) was used for evaluation of the measured data.
In order to get a good and high-quality crystalline structure, the films were annealed after the deposition in air, argon or in high vacuum.A long annealing for 6 h in argon atmosphere on temperature 600 °C was already used for these CuFeO2 delafossite films prepared by the RF reactive co-sputtering, as reported in reference [17].In our experiments, the annealing was done for selected samples in the air atmosphere.Temperature increase rate 10 °C/min to the temperature 650 °C with 3 h retention, and slow cooling (1 °C/min) was used.These conditions were found to be optimal by several preliminary experiments.Also, some samples received annealing in the high vacuum or in the argon directly in the plasma reactor after the deposition.A temperature of 500 or 650 °C and prolonged time of 6 h were used to provide enough time for crystallization process and formation of new crystalline phases in the material under high vacuum conditions.Other discharge conditions like average iron cathode discharge current I Feav , and average copper cathode current I Cuav , copper or iron cathode voltage U Fe , and copper cathode voltage U cu are mentioned in Table 1.
Transmission spectra of selected deposited films on FTO glass were measured by spectrometer (Carry 100, Varian, Palo Alto, CA, USA) with integrating sphere in the spectral range 200-900 nm.
The thickness of the films was determined by profilometry measurements (Alpha-Step 500 KLA Tencor MDL.no.ASIQ, San Jose, CA, USA).The deposited films were analyzed by X-ray diffraction (XRD).XRD data were measured at room temperature on a Bruker AXS D8 powder diffractometer (Billerica, MA, USA) in a Bragg-Brentano geometry using a CoKα radiation wavelength (λ = 1.7903Å, U = 34 kV, I = 30 mA).Chemical composition regarding molar ratio of Cu/Fe was measured by XRF.ED-XRF spectrometer Niton XL5 (Thermo Scientific, Waltham, MA, USA) equipped with Ag anode X-Ray Tube (6-50 kV, 0-500 µA, max 5 W) was used.Measure area diameter was 3 mm.Niton Connect PC software (Version 1.0) was used for evaluation of the measured data.
In order to get a good and high-quality crystalline structure, the films were annealed after the deposition in air, argon or in high vacuum.A long annealing for 6 h in argon atmosphere on temperature 600 • C was already used for these CuFeO 2 delafossite films prepared by the RF reactive co-sputtering, as reported in reference [17].In our experiments, the annealing was done for selected samples in the air atmosphere.Temperature increase rate 10 • C/min to the temperature 650 • C with 3 h retention, and slow cooling (1 • C/min) was used.These conditions were found to be optimal by several preliminary experiments.Also, some samples received annealing in the high vacuum or in the argon directly in the plasma reactor after the deposition.A temperature of 500 or 650 • C and prolonged time of 6 h were used to provide enough time for crystallization process and formation of new crystalline phases in the material under high vacuum conditions.Photoelectrochemical (PEC) measurements were done by the use of potentiostat (VoltaLAB PGZ100, RadiometerAnalytical SAS, Villeurbanne, France), connected to the three-electrode optical cell filled with 0.1 M Na 2 SO 4 (pH = 5.8) electrolyte.All experiments were done under an air atmosphere.The sample, with irradiated area set to 1 cm 2 by Teflon tape, was connected as working electrode.Platinum sheet was used as counter electrode and Ag/AgCl in 3 M KCl E Ag/AgCl = 0.207 V) as reference electrode.All potentials reported in our text are related to the mentioned reference electrode.Linear sweep voltammetry (+200 to −600 mV, sweep rate 10 mV/s) was done under the pulsed modulated light illumination with 5 s intervals of light and dark.The light was modulated by an electronic controlled mechanical shutter.Chronoamperometry (at −400 mV) was done under continuous illumination after short 30 s dark conditioning.A solar simulator (Newport, Irvine, CA, USA), emitting artificial AM1.5G solar spectrum with the intensity 100 mW•cm −2 , was used as the light source.

Plasma Diagnostics by RF Planar Probe
Plasma diagnostics at conditions of reactive r-HiPIMS and r-HiPIMS + ECWR depositions were done by a so-called RF planar probe, which is described in detail in reference [34].This method has origin in Sobolewski method presented in reference [35][36][37].The detailed experimental description together with Figures S1-S4 are in the Supplementary Materials.The plasma diagnostics with the RF planar probe is presented in this paper just for two selected types of plasma generation.The first measured type is the pure reactive r-HiPIMS without applied RF power to ECWR electrode P RF = 0 W. The particular conditions for this first type were: I Feav = 600 mA, I Cuav =38 mA, Q O2 =2 sccm.The second measured type is the hybrid reactive r-HiPIMS + ECWR with applied RF power to ECWR electrode P RF = 180 W. The conditions for this second type were: I Feav = 550 mA, I Cuav = 38 mA, Q O2 = 2 sccm.For both these measured types (r-HiPIMS, r-HiPIMS + ECWR) the other parameters were the same, as in case of all deposited samples presented in this paper.

Results and Discussion
The results of plasma diagnostics of the reactive r-HiPIMS and hybrid r-HiPIMS + ECWR are in Figures 2 and 3, respectively.The first difference between these two specific cases is the character of discharge current time evolution during the active discharge pulse on the Fe magnetron cathode where the active discharge pulse is quite long (100 µs).In case of pure r-HiPIMS we see typical discharge current decrease after ≈ 30 µs from the beginning of the pulse (Figure 2).This is a typical phenomenon in r-HiPIMS caused by so called gas rarefaction at the surface of the cathode due to the sputtering wind [26].On the other hand, in case of r-HiPIMS + ECWR, a quite flat discharge current time evolution (Figure 3) can be observed during the active discharge pulse.Similar behavior was already discussed in reference [31] and this effect was explained by higher ionization of the plasma during the active pulse and by higher concentration of double ionized Fe ++ in r-HiPIMS + ECWR.We have observed difference between r-HiPIMS and r-HiPIMS + ECWR in value and character of time evolution of the ion flux density i ionflux (Figures 2 and 3).In case of r-HiPIMS + ECWR, i ionflux values during active pulses on both cathodes were apparently higher (around 6 mA/cm 2 ) than in case of pure r-HiPIMS where i ionflux ≈ 3 mA/cm 2 .If we compare the tail electron temperature T e (Figures 2 and 3), it can be seen that T e is much higher in case of r-HiPIMS + ECWR than in pure r-HiPIMS.This is important phenomenon of superimposed ECWR plasma because higher T e means higher probability of ionization of sputtered particles which is the main feature of r-HiPIMS sources.Furthermore, this T e is very high (T e ≈ 4 eV) during all the time of the "OFF" time period of the pulse in case of hybrid r-HiPIMS + ECWR with measured plasma density n i ≈ 7 × 10 16 m −3 (Figure 3).It means the growing films in hybrid source are interacting with quite intensive plasma during the time between active discharge pulses and this is further significant difference from the pure r-HiPIMS which may influence parameters of deposited film.If we compare the time evolution of measured n i during active pulses in both cases, Figures 2  and 3, it can be seen that in the iron cathode active discharge pulse n i decreases from the beginning to the half of the pulse in case of r-HiPIMS + ECWR (Figure 3) and in case of pure r-HiPIMS, n i gradually increases during all the pulse (Figure 2).During the active discharge pulses of both cathodes, the value of n i does not significantly differ in hybrid and pure HiPIMS case.
The results of photoelectrochemical measurement of samples 1x, 2x, 3x, and 4x from the Table 1 can be seen in Figure 4.The molar ratios of Cu and Fe measured by XRF for these samples are also in the Table n Fe is dependent on the ratio of average discharge currents I CuAV and I FeAV but is also strongly dependent on the oxygen flow in plasma Q O2 .It is clear that this oxygen flow strongly influences the sputtering rate of iron target which is partially covered by iron oxide layer which has a smaller sputtering yield.As it can be seen from Table 1, when Q O2 decreased the amount of iron in deposited films strongly increased.In Figure 4a, we can see the linear voltammetry of these samples under light illumination and at dark conditions as the light from the solar simulator is chopped by shutter.All the mentioned samples exhibit the p-type conductivity and have quite large photocurrents.The highest photocurrents can be identified for the sample 2x with the molar ratio n Cu n Fe = 12.3, and the highest thickness was 2200 nm. Figure 4b shows the chronoamperometry under the constant illumination by solar simulator for 20 min.The photocurrent at the beginning quickly drops and then very slowly decreases.It looks like the most stable sample from this set of samples is 3x with n Cu n Fe = 4.3 and a thickness of 580 nm.
Coatings 2020, 10, 232 7 of 14 discharge pulses of both cathodes, the value of ni does not significantly differ in hybrid and pure HiPIMS case.
The results of photoelectrochemical measurement of samples 1x, 2x, 3x, and 4x from the Table 1 can be seen in Figure 4.The molar ratios of Cu and Fe measured by XRF for these samples are also in the Table 1.The measured molar ratio was found in the interval between ≈ 1.8 up to ≈ 12.3.
This ratio is dependent on the ratio of average discharge currents ICuAV and IFeAV but is also strongly dependent on the oxygen flow in plasma QO₂.It is clear that this oxygen flow strongly influences the sputtering rate of iron target which is partially covered by iron oxide layer which has a smaller sputtering yield.As it can be seen from Table 1, when QO₂ decreased the amount of iron in deposited films strongly increased.In Figure 4a, we can see the linear voltammetry of these samples under light illumination and at dark conditions as the light from the solar simulator is chopped by shutter.All the mentioned samples exhibit the p-type conductivity and have quite large photocurrents.The highest photocurrents can be identified for the sample 2x with the molar ratio 12.3, and the highest thickness was 2200 nm. Figure 4b shows the chronoamperometry under the constant illumination by solar simulator for 20 min.The photocurrent at the beginning quickly drops and then very slowly decreases.It looks like the most stable sample from this set of samples is 3x with 4.3 and a thickness of 580 nm.
The influence of the presence of oxygen dissolved in electrolyte was studied for sample 2x (Figure S5 in the Supplementary Information).Photoelectrochemical measurement in the absence of oxygen was performed under an argon atmosphere in an electrolyte bubbled by argon for five hours.Measurement in the presence of oxygen was performed under an air atmosphere.It was found that there was an almost negligible decrease in photocurrent in the absence of oxygen.Because of the negligible effect of dissolved oxygen, all photoelectrochemical experiments were done in the presence of oxygen (at concentrations corresponding to the equilibrium with air).Phase composition of deposited layers 1x-4x changed after post deposition annealing in air or Ar as documented in Figure 5 for samples 1x and 2x.As the deposited sample contains only CuO crystalline phase, after calcination at 650 °C in air, the XRD peak at 45° corresponding to CuO strongly increased as the result of improved crystallinity (see Figure 5a).No other phase than CuO was detected.On the other side, Cu2O phase appears when sample (2x) was annealed in the argon atmosphere as can be seen in Figure 5b.The formation of Cu2O phase can be explained by the reduction of CuO during annealing.The influence of the presence of oxygen dissolved in electrolyte was studied for sample 2x (Figure S5 in the Supplementary Information).Photoelectrochemical measurement in the absence of oxygen was performed under an argon atmosphere in an electrolyte bubbled by argon for five hours.Measurement in the presence of oxygen was performed under an air atmosphere.It was found that there was an almost negligible decrease in photocurrent in the absence of oxygen.Because of the negligible effect of dissolved oxygen, all photoelectrochemical experiments were done in the presence of oxygen (at concentrations corresponding to the equilibrium with air).
Phase composition of deposited layers 1x-4x changed after post deposition annealing in air or Ar as documented in Figure 5 for samples 1x and 2x.As the deposited sample contains only CuO crystalline phase, after calcination at 650 • C in air, the XRD peak at 45 • corresponding to CuO strongly increased as the result of improved crystallinity (see Figure 5a).No other phase than CuO was detected.
On the other side, Cu 2 O phase appears when sample (2x) was annealed in the argon atmosphere as can be seen in Figure 5b.The formation of Cu 2 O phase can be explained by the reduction of CuO during annealing.The XRD spectra of samples 1x after annealing in air at 650 °C before electrochemical measurement "NM" and after electrochemical measurement "ME" can be also seen in Figure 5a.The sample 1x contains after electrochemical measurement (linear voltammetry and 20 min chronoamperometry at −400 mV vs. Ag/AgCl under light) "ME" also phase Cu2O (peaks 42.5° and 49.7°).It means that films were slowly reduced during photoelectrochemical measurement.Photocurrents of samples 1x-4x annealed in argon (not shown in graphs) were much lower than samples annealed in air.As can be seen from Table 1, samples 1x-4x have Cu/Fe molar ratio much higher than in the desired delafossite CuFeO2 (1:1).This is probably the reason why the films 1x-4x contain only CuO and not CuFeO2.Furthermore, although samples 1x-4x contain quite significant fraction of Fe no iron or iron oxide phases were identified in these samples.Conditions of the plasma deposition were then further changed and optimized for samples 5x to 10x to achieve the desired molar ratio of Cu/Fe =1.
The next investigated series consists of samples 6x, 7x, 8x which were heated during the deposition at 500 °C (Table 1).The results of linear voltammetry and chronoamperometry measurement can be seen in Figure 6 for sample 6x as deposited and samples 6x, 7x, 8x annealed in air at 650 °C.The ratio of Cu and Fe measured by XRF in these films were in the interval ≈ 0.8 − 1.The results of XRD measured on the sample 6x without postdeposition annealing, after annealing The XRD spectra of samples 1x after annealing in air at 650 • C before electrochemical measurement "NM" and after electrochemical measurement "ME" can be also seen in Figure 5a.The sample 1x contains after electrochemical measurement (linear voltammetry and 20 min chronoamperometry at −400 mV vs. Ag/AgCl under light) "ME" also phase Cu 2 O (peaks 42.5 • and 49.7 • ).It means that films were slowly reduced during photoelectrochemical measurement.Photocurrents of samples 1x-4x annealed in argon (not shown in graphs) were much lower than samples annealed in air.As can be seen from Table 1, samples 1x-4x have Cu/Fe molar ratio much higher than in the desired delafossite CuFeO 2 (1:1).This is probably the reason why the films 1x-4x contain only CuO and not CuFeO 2 .Furthermore, although samples 1x-4x contain quite significant fraction of Fe no iron or iron oxide phases were identified in these samples.Conditions of the plasma deposition were then further changed and optimized for samples 5x to 10x to achieve the desired molar ratio of Cu/Fe =1.
The next investigated series consists of samples 6x, 7x, 8x which were heated during the deposition at 500 • C (Table 1).The results of linear voltammetry and chronoamperometry measurement can be seen in Figure 6 for sample 6x as deposited and samples 6x, 7x, 8x annealed in air at 650 • C. The ratio of Cu and Fe measured by XRF in these films were in the interval n Cu n Fe ≈ 0.8 − 1.The results of XRD measured on the sample 6x without postdeposition annealing, after annealing in air on 650 • C not electrochemically measured "MN" and after annealing in air on 650 • C after photoelectrochemical (PEC) measurement "ME" can be seen in Figure 7.As deposited sample 6x contains delafossite phase (peaks at 36.5 • and 72.1 • ), Cu 2 O phase and CuO phase but it does not have any photoresponse in PEC measurement (see Figure 6).After annealing in air, sample 6x lost the delafossite phase which was transformed to CuFe 2 O 4 phase, Cu 2 O phase disappeared and amount of CuO increased.But annealing in air results in a significant increase in photocurrent response for all samples 6x, 7x and 8x (see Figure 6a), with the highest photocurrents for sample 6x from this set of samples.The chronoamperometry shown in Figure 6b shows that observed photocurrents significantly decrease with time which suggests that these films are not stable in time during PEC polarization.Instability was confirmed by XRD measurement of sample 6 after electrochemical measurement "ME" (Figure 7) which shows the formation of Cu 2 O due to the photoelectrochemical reduction of CuO.
Coatings 2020, 10, 232 9 of 14 photoelectrochemical (PEC) measurement "ME" can be seen in Figure 7.As deposited sample 6x contains delafossite phase (peaks at 36.5° and 72.1°), Cu2O phase and CuO phase but it does not have any photoresponse in PEC measurement (see Figure 6).After annealing in air, sample 6x lost the delafossite phase which was transformed to CuFe2O4 phase, Cu2O phase disappeared and amount of CuO increased.But annealing in air results in a significant increase in photocurrent response for all samples 6x, 7x and 8x (see Figure 6a), with the highest photocurrents for sample 6x from this set of samples.The chronoamperometry shown in Figure 6b shows that observed photocurrents significantly decrease with time which suggests that these films are not stable in time during PEC polarization.Instability was confirmed by XRD measurement of sample 6 after electrochemical measurement "ME" (Figure 7) which shows the formation of Cu2O due to the photoelectrochemical reduction of CuO.
(a) (b)  In order to get some photocurrents at films with delafossite structure, samples 9x and 10x were annealed after the deposition in vacuum (10 −3 Pa) at 650 and 500 °C.
Figure S6 (in SI) shows XRD patterns of sample 9x after annealing in a vacuum at 650 °C.After this thermal treatment, sample 9x contains large fraction of delafossite (CuFeO2) phase and a small amount of Cu2O.No other crystal phase containing iron was found.Figure 8 shows chopped light linear voltammetry and chronoamperometry.Chronoamperometry (Figure 8b) of these delafossite samples shows a decrease of photocurrents by time.Although photocurrents obtained from linear voltammetry (Figure 8a) were rather small in comparison with previous set of samples (6x-8x, Figure Coatings 2020, 10, 232 9 of 14 photoelectrochemical (PEC) measurement "ME" can be seen in Figure 7.As deposited sample 6x contains delafossite phase (peaks at 36.5° and 72.1°), Cu2O phase and CuO phase but it does not have any photoresponse in PEC measurement (see Figure 6).After annealing in air, sample 6x lost the delafossite phase which was transformed to CuFe2O4 phase, Cu2O phase disappeared and amount of CuO increased.But annealing in air results in a significant increase in photocurrent response for all samples 6x, 7x and 8x (see Figure 6a), with the highest photocurrents for sample 6x from this set of samples.The chronoamperometry shown in Figure 6b shows that observed photocurrents significantly decrease with time which suggests that these films are not stable in time during PEC polarization.Instability was confirmed by XRD measurement of sample 6 after electrochemical measurement "ME" (Figure 7) which shows the formation of Cu2O due to the photoelectrochemical reduction of CuO.
(a) (b)  In order to get some photocurrents at films with delafossite structure, samples 9x and 10x were annealed after the deposition in vacuum (10 −3 Pa) at 650 and 500 °C.
Figure S6 (in SI) shows XRD patterns of sample 9x after annealing in a vacuum at 650 °C.After this thermal treatment, sample 9x contains large fraction of delafossite (CuFeO2) phase and a small amount of Cu2O.No other crystal phase containing iron was found.Figure 8 shows chopped light linear voltammetry and chronoamperometry.Chronoamperometry (Figure 8b) of these delafossite samples shows a decrease of photocurrents by time.Although photocurrents obtained from linear voltammetry (Figure 8a) were rather small in comparison with previous set of samples (6x-8x, Figure In order to get some photocurrents at films with delafossite structure, samples 9x and 10x were annealed after the deposition in vacuum (10 −3 Pa) at 650 and 500 • C.
Figure S6 (in SI) shows XRD patterns of sample 9x after annealing in a vacuum at 650 • C.After this thermal treatment, sample 9x contains large fraction of delafossite (CuFeO 2 ) phase and a small amount Coatings 2020, 10, 232 10 of 14 of Cu 2 O.No other crystal phase containing iron was found.Figure 8 shows chopped light linear voltammetry and chronoamperometry.Chronoamperometry (Figure 8b) of these delafossite samples shows a decrease of photocurrents by time.Although photocurrents obtained from linear voltammetry (Figure 8a) were rather small in comparison with previous set of samples (6x-8x, Figure 6a), it is noticeable that the decrease of photocurrent with polarization time is not as fast as in the case of samples 6x-8x.
Coatings 2020, 10, 232 10 of 14 6a), it is noticeable that the decrease of photocurrent with polarization time is not as fast as in the case of samples 6x-8x.
(a) (b)  Table 2 represents the phase composition of samples annealed in vacuum at 650 °C, before (NM) and after (ME) photoelectrochemical polarization.Data are expressed semiquantitative weight percent phase composition of the measured sample including support FTO glass (SnO2 phase).There still exist some CuFeO2 phase after photoelectrochemical polarization, but phase composition for each sample is different.While sample 9x contains a smaller amount of CuFeO2, an increased amount of Cu2O, and a new phase Fe3O4, sample 10x contains similar amount of CuFeO2, and Cu2O disappears.This means that sample 9x contains CuFeO2, which is stable during photoelectrochemical polarization and at the same time exhibits promising photocurrent values.Further work will be devoted to the deposition of films with higher content of CuFeO2 at deposition conditions corresponding to sample 9x to achieve more efficient and stable p-type photocathode films.
The results of measurement of optical transmission in UV-VIS region can be seen in Figure 9 for particular deposited samples.The optical band gap can only be roughly estimated because we used only transmission spectra.All optical diagnostics was done on the films before PEC measurements.The right graph shows also optical transmission of bare FTO glass after annealing on 650 °C.It can be seen that FTO glass is still very well transparent up to 3.5 eV.The molar ratio of Cu/Fe is mentioned at each optical spectrum.Samples in the left graph 2x-5x were all annealed in air up to 650 °C.As it was shown from the diffraction experiment, sample 2x with the molar ratio ncu/nFe = 12.5 exhibits the crystal structure of CuO (Figure 5).For the sample 2x the band gap was Eg ≈ 1.5 eV, which is close to reported value of CuO crystalline material [38].When the iron amount grows in the films the band gap increases to the values close to Eg ≈ 2 eV.On the right graph we can see transmission spectra of samples 9x, 10x annealed in vacuum on 650 °C, and these samples contain a large fraction of delaffosite phase (Table 2).Samples 9x and 10x with the highest amount of delafossite phase have the band gap Eg ≈ 1.8 eV, this is in the region typically reported for delafossite [4].These two samples have also ncu/nFe ≈ 1.This means that sample 9x contains CuFeO 2 , which is stable during photoelectrochemical polarization and at the same time exhibits promising photocurrent values.Further work will be devoted to the deposition of films with higher content of CuFeO 2 at deposition conditions corresponding to sample 9x to achieve more efficient and stable p-type photocathode films.The results of measurement of optical transmission in UV-VIS region can be seen in Figure 9 for particular deposited samples.The optical band gap can only be roughly estimated because we used only transmission spectra.All optical diagnostics was done on the films before PEC measurements.The right graph shows also optical transmission of bare FTO glass after annealing on 650 • C. It can be seen that FTO glass is still very well transparent up to 3.5 eV.The molar ratio of Cu/Fe is mentioned at each optical spectrum.Samples in the left graph 2x-5x were all annealed in air up to 650 • C. As it was shown from the diffraction experiment, sample 2x with the molar ratio n cu /n Fe = 12.5 exhibits the crystal structure of CuO (Figure 5).For the sample 2x the band gap was E g ≈ 1.5 eV, which is close to reported value of CuO crystalline material [38].When the iron amount grows in the films the band gap increases to the values close to E g ≈ 2 eV.On the right graph we can see transmission spectra of samples 9x, 10x annealed in vacuum on 650 • C, and these samples contain a large fraction of delaffosite phase (Table 2).Samples 9x and 10x with the highest amount of delafossite phase have the band gap E g ≈ 1.8 eV, this is in the region typically reported for delafossite [4].These two samples have also n cu /n Fe ≈ 1.

Conclusions
A reactive high-power impulse magnetron sputtering system (r-HiPIMS) and a hybrid high power impulse magnetron sputtering combined with electron cyclotron wave resonance plasma source (r-HiPIMS + ECWR) were used for the deposition of p-type CuFexOy thin films onto FTO substrates.
These experiments confirm that at certain conditions in this r-HiPIMS and in hybride r-HiPIMS + ECWR reactive sputtering process a delafossite CuFeO2 phase can be formed in the deposited material after the post deposition annealing in vacuum.On the other hand, the material still contains different phases like Cu2O or some phases of iron oxides.For this reason, the stability of prepared cathodes was still limited.For future experiments, more precise stoichiometry control of the sputtering rate of copper and iron should be implemented, together with some emission spectroscopy measurements of Cu and Fe atoms in the magnetron plasma.Furthermore, more stable and thermal resistant substrates with conductive electrode should be used for deposition experiments.Deposition conditions were optimized to achieve the desired Cu/Fe ratio = 1 in deposited films, as deposited films have negligible photocurrent response.Annealing in 650 °C results in photoactive films where the p-type material is CuO.However, the photocurrent decreases with the time of the photoelectrochemical polarization due to the reduction of CuO to Cu2O.High temperature of deposition (500 °C) together with post deposition heat treatment at 650 °C in vacuum is essential for the formation of stable and photoactive CuFeO2 phase.Further effort will be devoted to increase CuFeO2 content and to achieve more efficient p-type photocathode films.

Supplementary Materials:
The following are available online at www.mdpi.com/xxx/s1, Figure S1: Experimental configuration of guarded planar radio frequency (RF) probe for measurement of current and voltage waveforms and plasma diagnostics, Figure S2 Author Contributions: Z.H. was involved in writing of original draft manuscript and its revision, he also performed thin film deposition, M.Z. was involved in film characterization, photoelectrochemical measurement and data analysis, J.O. performed thin film analysis, D.T. and M.Č. were involved in plasma diagnostics and J.K.

Conclusions
A reactive high-power impulse magnetron sputtering system (r-HiPIMS) and a hybrid high power impulse magnetron sputtering combined with electron cyclotron wave resonance plasma source (r-HiPIMS + ECWR) were used for the deposition of p-type CuFe x O y thin films onto FTO substrates.
These experiments confirm that at certain conditions in this r-HiPIMS and in hybride r-HiPIMS + ECWR reactive sputtering process a delafossite CuFeO 2 phase can be formed in the deposited material after the post deposition annealing in vacuum.On the other hand, the material still contains different phases like Cu 2 O or some phases of iron oxides.For this reason, the stability of prepared cathodes was still limited.For future experiments, more precise stoichiometry control of the sputtering rate of copper and iron should be implemented, together with some emission spectroscopy measurements of Cu and Fe atoms in the magnetron plasma.Furthermore, more stable and thermal resistant substrates with conductive electrode should be used for deposition experiments.Deposition conditions were optimized to achieve the desired Cu/Fe ratio = 1 in deposited films, as deposited films have negligible photocurrent response.Annealing in 650 • C results in photoactive films where the p-type material is CuO.However, the photocurrent decreases with the time of the photoelectrochemical polarization due to the reduction of CuO to Cu 2 O. High temperature of deposition (500 • C) together with post deposition heat treatment at 650 • C in vacuum is essential for the formation of stable and photoactive CuFeO 2 phase.Further effort will be devoted to increase CuFeO 2 content and to achieve more efficient p-type photocathode films.

Figure 1 .
Figure 1.Experimental apparatus with two magnetrons and the radio frequency (RF) electron cyclotron wave resonance (ECWR) electrode for the deposition of copper iron oxide thin films.

Figure 1 .
Figure 1.Experimental apparatus with two magnetrons and the radio frequency (RF) electron cyclotron wave resonance (ECWR) electrode for the deposition of copper iron oxide thin films.

Figure 2 .
Figure 2. Time evolution of plasma parameters like, ion plasma density ni (a), tail electron temperature Te (b) and ion flux density on the probe iionflux (c) of reactive r-HiPIMS without applied ECWR RF power.

Figure 2 .Figure 3 .
Figure 2. Time evolution of plasma parameters like, ion plasma density n i (a), tail electron temperature T e (b) and ion flux density on the probe i ionflux (c) of reactive r-HiPIMS without applied ECWR RF power.

Figure 3 .
Figure 3.Time evolution of plasma parameters like ion plasma density n i (a), tail electron temperature T e (b) and ion flux density on the probe i ionflux (c) of reactive r-HiPIMS + ECWR plasma with applied RF power P RF = 180 W.

Figure 8 .
Figure 8. Linear voltammetry (a) and chronoamperometry (b) of samples 9x and 10x exposed to post deposition annealing in vacuum for 6 h at 500 • C (500 • C vac) and 650 • C (650 • C vac).Table 2 represents the phase composition of samples annealed in vacuum at 650 • C, before (NM) and after (ME) photoelectrochemical polarization.Data are expressed as semiquantitative weight percent phase composition of the measured sample including support FTO glass (SnO 2 phase).There still exist some CuFeO 2 phase after photoelectrochemical polarization, but phase composition for each sample is different.While sample 9x contains a smaller amount of CuFeO 2 , an increased amount of Cu 2 O, and a new phase Fe 3 O 4 , sample 10x contains similar amount of CuFeO 2 , and Cu 2 O disappears.This means that sample 9x contains CuFeO 2 , which is stable during photoelectrochemical polarization and at the same time exhibits promising photocurrent values.Further work will be devoted to the deposition of films with higher content of CuFeO 2 at deposition conditions corresponding to sample 9x to achieve more efficient and stable p-type photocathode films.

Figure 9 .
Figure 9. Transmission spectra of selected deposited films 2x, 3x, 4x, 5x (a) and 9x, 10x (b) on FTO glass substrates measured by UV-VIS optical spectrometry.All optical diagnostics was done on the films before PEC measurements.

:
The example of time evolution of RF current IRFprobe RF voltage URFprobe measured on the RF planar probe.Time evolution of cathode voltage on Cu cathode is shown as well in case of reactive r-HiPIMS + ECWR deposition copper iron oxide with PECWR = 180 W, Figure S3: Example of reconstructed current voltage characteristics for the single URFp voltage sweep in the middle of copper discharge pulse in case of reactive r-HiPIMS + ECWR deposition copper iron oxide with PECWR = 180 W, Figure S4: Example of reconstructed current voltage characteristics for the single URFp voltage sweep in the middle of copper pulse in case of reactive r-HiPIMS + ECWR deposition copper iron oxide with PECWR = 180 W, logarithm of second derivative is shown with fitted electron temperature Te from the tail of electron distribution function, Figure S5: Linear voltammetry of the sample 2x annealed in air on 650 °C in electrolyte without degassing by argon and in electrolyte degassed by flow of argon for 5 h before PEC measurement, Figure S6: X-ray diffraction (XRD) of sample 9x with significant delafossite structure after annealing of the sample after deposition on 650 °C in vacuum.

Figure 9 .
Figure 9. Transmission spectra of selected deposited films 2x, 3x, 4x, 5x (a) and 9x, 10x (b) on FTO glass substrates measured by UV-VIS optical spectrometry.All optical diagnostics was done on the films before PEC measurements.

:
The example of time evolution of RF current IRFprobe RF voltage URFprobe measured on the RF planar probe.Time evolution of cathode voltage on Cu cathode is shown as well in case of reactive r-HiPIMS + ECWR deposition copper iron oxide with PECWR = 180 W, Figure S3: Example of reconstructed current voltage characteristics for the single U RFp voltage sweep in the middle of copper discharge pulse in case of reactive r-HiPIMS + ECWR deposition copper iron oxide with PECWR = 180 W, Figure S4: Example of reconstructed current voltage characteristics for the single U RFp voltage sweep in the middle of copper pulse in case of reactive r-HiPIMS + ECWR deposition copper iron oxide with PECWR = 180 W, logarithm of second derivative is shown with fitted electron temperature Te from the tail of electron distribution function, Figure S5: Linear voltammetry of the sample 2x annealed in air on 650 • C in electrolyte without degassing by argon and in electrolyte degassed by flow of argon for 5 h before PEC measurement, Figure S6: X-ray diffraction (XRD) of sample 9x with significant delafossite structure after annealing of the sample after deposition on 650 • C in vacuum.

Table 1 .
List of films with significant deposition parameters.

I Feav [mA] U Fe [V] I Cuav [mA] U Cu [V] Q O2 [sccm] t dep. [ • C] Thickness [nm] mol. Ratio Cu/Fe P ECWR [W]
Other discharge conditions like average iron cathode discharge current IFeav, and average copper cathode current ICuav, copper or iron cathode voltage UFe, and copper cathode voltage Ucu are mentioned in Table

Table 1 .
List of films with significant deposition parameters.