Hollow Cathode Gas Flow Sputtering of Nickel Oxide Thin Films for Hole‐Transport Layer Application in Perovskite Solar Cells

Nickel oxide (NiO1+δ) is a versatile material used in various fields such as optoelectronics, spintronics, electrochemistry, and catalysis which is prepared with a wide range of deposition methods. Herein, for the deposition of NiO1+δ films, the reactive gas flow sputtering (GFS) process using a metallic Ni hollow cathode is developed. This technique is distinct and has numerous advantages compared to conventional sputtering methods. The NiO1+δ films are sputtered at low temperatures (100 ºC) for various oxygen partial pressures during the GFS process. Additionally, Cu‐incorporated NiO1+δ (Cu x Ni1−x O1+δ) films are obtained with 5 and 8 at% Cu. The thin films of NiO1+δ are characterized and evaluated as a hole‐transporting layer (HTL) in perovskite solar cells (PSCs). The NiO1+δ devices are benchmarked against state‐of‐the‐art self‐assembled monolayers (SAM) ([2‐(3,6‐dimethoxy‐9H‐carbazol‐9‐yl)ethyl]phosphonic acid (also known as MeO2PACz)‐based PSCs. The best‐performing NiO1+δ PSC achieves an efficiency (η) of ≈16% without a passivation layer at the HTL interface and demonstrates better operational stability compared to the SAM device. The findings suggest that further optimization of GFS NiO1+δ devices can lead to higher‐performing and more stable PSCs.


Results and Discussion
In experimental settings where the properties of the thin films were examined, the thickness was varied between 100 and 500 nm (for an overview of various NiO 1+δ film thicknesses used in experiments, see Table S1, Supporting Information).For the investigation of the elemental composition of the films, we used the electron probe microanalyzer (EPMA).The results are shown in Figure 1a,b and summarized in Table S2, Supporting Information.Figure 1a,b illustrates the variation of δ in NiO 1 +δ and Cu x Ni 1Àx O 1+δ and the atomic ratio of Cu to the total cation (Ni+Cu).In NiO 1+δ and Cu x Ni 1Àx O 1+δ (1-ring) films, as the p(O 2 ) increased, the cation/anion ratio surprisingly increased.This was contrary in the case of Cu x Ni 1Àx O 1+δ (2-ring) films where the anion content in the film increased with increasing p(O 2 ).In all Cu x Ni 1Àx O 1+δ films, the Cu content decreased with increasing p(O 2 ).The effect of decreasing anion (oxygen) content in NiO x films with p(O 2 ) has also been previously reported by Sun et al. [62] In this work, [62] optical emission spectroscopy disclosed that at higher p(O 2 ), the oxygen concentration in plasma dropped significantly, indicating that more oxygen reacted with the Ni-target surface ("target poisoning").It was supported by the fact that the deposition rate simultaneously decreased. [62]The deposition rate for the GFS NiO 1+δ also decreases with increasing p(O 2 ). [63]However, target poisoning in an intensive argon flow is less probable.We suggest a formation of bound Ni─O species in a gas phase and their elimination by the gas flow at higher p(O 2 ). [44,64]As shown in Figure 1b, with an increasing p(O 2 ), the relative ratio of Cu-to-Ni decreased.This also counters target poisoning since NiO is thermodynamically more stable than Cu 2 O. Copper can readily bind to oxygen at a lower O/metal ratio due to its stable oxidation states (+1 and +2), potentially resulting in the binding of more oxygen atoms compared to nickel, which has oxidation states (+2 and +3).This implies that, theoretically, a single oxygen atom could bind to more copper atoms than to nickel atoms.For simplicity, the NiO 1+δ films will be denoted as NiO 1.23 , NiO 1.15 , NiO 1.14 , and NiO 1.12 (based on at% oxygen/ metal ratio from EPMA).While the two Cu x Ni 1Àx O 1+δ films obtained at 0.24 Pa, i.e., Cu 0 .10 Ni 0.90 O 1.11 and Cu 0.15 Ni 0.85 O 1.0 , will be denoted as NiO x :Cu_1 and NiO x :Cu_2, respectively.
To study the crystallinity of the film both the grazing incidence X-Ray diffraction (GI-XRD) and the Bragg-Brentano XRD (BB-XRD) were performed (see Figure S1 and Table S3, Supporting Information).[67][68] GI-XRD detects diffraction from the crystallographic planes with increasing tilting angle to the surface when 2θ increases.One can therefore estimate if residual stresses exist in the film.The lattice constant determined from each diffraction maximum decreases with 2θ for all NiO 1+δ films considered.This suggests the likelihood of compressive stress in the films.The relative changes of lattice constant are shown in Figure S2, Supporting Information.The BB-XRD indicated a strong preferential orientation along the (111) crystallographic plane which is typical for low-temperature sputtered NiO 1+δ films with a columnar growth. [63,69,70]According to Song et al., [70] the preferential orientation along the (111) crystallographic plane originates from the growth mechanism according to the Van der Drift competitive model. [71]The calculated texture coefficient (TC) along the (111) crystallographic plane and the average crystallite size from BB-XRD are shown in Figure 1c.All films exhibited a texture along the (111) crystallographic plane, with a TC (111) value greater than 1.A maximum TC (111) of 4.14 was obtained at the highest p(O 2 ) of 0.47 Pa.According to Ryu et al., [72] when the active species of Ni and O produced by the sputtering process collide separately with the growing film surface, the arrangement of O 2À ions dictates the crystallographic orientation in resultant films.O 2À ions (radius: 1.40 A), which are larger than Ni 2+ ions (radius: 0.69 A), are most densely packed along the (111)-plane in the NiO crystal structure.This suggests that the (111) orientation minimizes the surface free energy during the film growth. [72,73]Chen et al. demonstrated that when sputtered with <50% O 2 /Ar mixture and substrate temperatures up to 673 K, (111)-oriented NiO 1+δ are obtained. [69]or Cu x Ni 1Àx O 1+δ , the BB-XRD peak intensities were relatively weaker when compared to the NiO 1+δ and the diffraction angle shifted to lower values (Figure S1 and Table S3, Supporting Information).[76][77] Additionally, XRD patterns did not reveal any Cu-related peaks, suggesting that the addition of Cu did not alter the NiO-type phase structure.For reference, the BB-XRD pattern of a Cu x O film obtained by sputtering with only a Cu target, as shown in Figure S2, Supporting Information.[80] The amount of Cu was found to influence the crystallographic orientation and lattice deformation in the NiO 1+δ films as reported in the literature. [81]In contrast to the NiO 1+δ films without Cu, certain crystal orientations were detected at a specific amount of Cu (see BB-XRD, Figure S1, Supporting Information), and the TC (111) progressively decreased with increasing amount of Cu.The secondary electron emission microscopy (SEM) images are shown in Figure S3, Supporting Information.A direct correlation of crystallite sizes from SEM and BB-XRD is not meaningful as the former are obtained from ≈100 nm films while the latter is from ≈500 nm films.Moreover, the SEM images represent the surface morphology of the material while the XRD analysis employs a mathematical model to estimate average sizes.This is achieved in the latter by analyzing the broadening of diffraction peaks resulting from the constructive interference of scattered monochromatic X-rays at specific angles from a set of lattice planes. [82]he atomic force microscopy (AFM) images of the indium tin oxide (ITO) substrate and various NiO 1+δ coated on ITO are shown in Figure S4, Supporting Information.The average roughness (R a ) determined from the AFM images for ITO, NiO 1.23 , NiO 1.15 , NiO 1.14 , and NiO 1.12 were 5.5, 10.1, 5.5, and 9.1 nm, respectively.In the case of NiO x :Cu_1 and NiO x :Cu_2, the R a decreased to 5.4 and 6.6 nm, respectively, in comparison to NiO 1.15 with R a 10.1 nm which were all deposited at a p(O 2 ) of 0.24 Pa.
The optical bandgap determined from the Tauc plot, the lattice constant determined from BB-XRD, and the oxygen/metal ratio (from EPMA) in the films are plotted in Figure 1d,e.Figure S5, Supporting Information, displays the UV-vis spectra and Tauc plots.The determined lattice constant (a) for the NiO 1+δ films was in the range 4.1777-4.1709A when the p(O 2 ) was varied.The decreasing a with increasing p(O 2 ) is attributed to a reduction in the V Ni .When V Ni exists in the lattice, the six nearest oxygen atoms move away from the center of the V Ni . [83]Furthermore, high-resolution XRD in Bragg-Brentano geometry (HR-BB-XRD) was also performed in the 2θ region of the (111)-plane for NiO 1+δ (Figure 1f,h).The XRD peak detected at 37.17º/37.22º(Figure 1g,h The variation of p(O 2 ) did not have a strong influence on the optical bandgap (E g ) of NiO 1+δ films, which was ≈3.68 eV (Figure 1d).In Cu x Ni 1Àx O 1+δ films, the optical bandgap (E g ) was marginally lower compared to NiO 1+δ , with an increase in Cu content (Figure 1e).The narrowing of optical bandgap in Cu x Ni 1Àx O 1+δ films is consistent with previously reported literature. [84,85]The transmittance of NiO 1+δ films was found to be influenced by p(O 2 ), with films deposited at higher p(O 2 ) exhibiting greater transparency compared to those deposited at lower p(O 2 ) levels (Figure S5, Supporting Information).[88] Therefore, an increase in the transmittance of NiO 1+δ with increasing p(O 2 ) suggests a decrease in the Ni 3+ ion content (in agreement with the X-Ray photoelectron spectroscopy (XPS) analysis presented later).This trend is contrary to the one seen in certain literature where a higher oxygen flow during sputtering results in increased Ni 3+ content in the films. [47,87,89,90]In the case of Cu x Ni 1Àx O 1+δ films, the decrease in transmittance correlates with decreasing crystallinity as detected by BB-XRD (weaker diffraction peaks).It is known that good crystallization can improve transparency and vice versa.[93] For NiO x :Cu_1 and NiO x :Cu_2, the value of lattice constant increased compared to the NiO 1.15 (Figure 1e).Therefore, along with the 2θ shift as discussed earlier, this suggests that the larger Cu cations (Cu + and Cu 2+ ) replaced the Ni 2+ cations in the NiO cubic lattice. [94]aman spectroscopy was performed to investigate the microstructure of the NiO 1+δ films.The fitted Raman spectra are shown and tabulated in Figure S6 and Table S4, Supporting Information, respectively.In the bulk NiO with rock salt crystal structure, the one-phonon (1P) transverse optical (TO) and longitudinal optical (LO) modes are forbidden.[97][98] Generally, the presence of LO mode is attributed to imperfections within the crystal lattice or point defects such as nickel vacancies or Ni 3+ ions. [83,99,100]For the NiO 1+δ , the 1P TO and LO modes are resolved into two bands each (TO 1 /TO 2 and LO 1 /LO 2 ) which arise from the anisotropy of optical phonons. [97,98]The second order (2P) modes (TO + LO and 2LO) are detected at 994-1001 cm À1 and 1078-1091 cm À1 , respectively.105] Furthermore, the SO mode is known to be particle sizedependent, it enhances with a decrease in crystallite size. [106,107]rom the Raman spectra, it was detected that with increasing Cu content in the films, the typical NiO 1+δ modes due to defects (LO and TO modes) and surface effects (SO mode) are enhanced without bringing any other fundamental changes to the NiO-type structure in agreement with the GI-and BB-XRD. [108]PS was used to investigate the surface chemistry of the NiO 1+δ films.Before XPS measurements, the surface of the NiO 1+δ was sputtered with Ar + ions (2.7 Â 10 À6 mbar, 20 W, 20 mins) to remove adsorbed contaminants from the ambient (Figure S7, Supporting Information).With the Ar + sputtering, an obvious partial reduction of NiO 1+δ to metallic Ni 0 was not detected. [109]111][112] To simplify this approach, we have estimated the ratio of Ni 3+ /Ni 2+ states from the O1s spectra, as shown in Figure S8, Supporting Information. [113,114]Accurately determining the Ni 3+ /Ni 2+ ratio, particularly in the case of Cu x Ni 1Àx O 1+δ , remains a challenging task with such an approach due to the presence of different valency of Cu bound to oxygen within the material. [115]or the sake of simplicity, this contribution is neglected in the determination of the Ni 3+ /Ni 2+ ratio from the XPS O1s spectra for Cu x Ni 1Àx O 1+δ . [76]The O1s spectra were deconvoluted into NiO (≈529.4eV), NiOOH (≈530.6 eV), Ni 2 O 3 (≈531.3eV), and NiOOH (≈532.1 eV) components (Figure S8, Supporting Information). [112]Despite carrying out an Ar + -etching before the XPS measurements, traces of moisture (feature at ≈532.7 eV) were detected on samples NiO 1.15 and NiO 1.14 .Ratcliff et al. reported such physisorbed H 2 O on their Ar + sputter-cleaned NiO 1+δ film. [109]The Ni 3+ /Ni 2+ ratio (Table S5, Supporting Information) decreased as the p(O 2 ) increased and showed a proportionate relationship with the amount of Cu in NiO 1+δ .This trend aligns with the variation of transparency with p(O 2 ) (Figure S5a, Supporting Information), i.e., an increase in transparency of the NiO 1+δ due to decreasing Ni 3+ .The Cu 2p spectra (Figure S9, Supporting Information) confirm the presence of Cu in the Cu x Ni 1Àx O 1+δ .The feature detected at 932.6 eV corresponds to the Cu 0 (metal) or Cu +1 (Cu 2 O).Cu 0 and Cu +1 are indistinguishable by XPS due to the superimposition of their binding energies. [116,117][120] The Cu atoms were incorporated predominantly as Cu + acceptors with 92.7 and 86.6% on the surface of NiO x :Cu_1 and NiO x :Cu_2, respectively (Figure S9, Supporting Information).It is to be noted that the XPS-induced reduction of Cu 2+ to Cu + could influence the determination of Cu + and Cu 2+ from the Cu 2p 3/2 spectra. [121]In Cu x Ni 1Àx O 1+δ , for each Cu + substituted position, the adjacent Ni 2+ will be converted to Ni 3+ to maintain overall charge neutrality in the crystal. [122]Also, the excessive O i can trap two electrons to become O i 0 0 , and to balance the O 2À formed, two Ni 2+ must be converted to Ni 3+ . [122]The Cu + substitution-induced Ni 3+ /Ni 2+ increment is thus noticed in the GFS Cu x Ni 1Àx O 1+δ (Table S5, Supporting Information), i.e., the Ni 3+ /Ni 2+ ratio increases from 0.27 in NiO x :Cu_1 to 0.44 in NiO x :Cu_2.However, the Cu x Ni 1Àx O 1+δ in general had a lower Ni 3+ /Ni 2+ ratio compared to the undoped NiO 1+δ which could be explained by the filling of Cu ions in V Ni .
][125][126] Small polarons are quasiparticles formed by the strong coupling between charge carriers and phonons in a crystal lattice where the distortion of the lattice due to the charge carrier extends over distances smaller than the lattice constant. [124]According to Zhang et al., [125] in NiO 1+δ , which exhibits lower mobilities (<0.1 cm 2 V À1 s) due to strong electron correlation and localized nature of the valence band (VB), the conduction mechanism is well described by SPH.All undoped NiO 1+δ films in this study except NiO 1.15 had a mobility (μ h ) value of ≈0.1 cm 2 V À1 s À1 , while the latter had a value of 3.1 cm 2 V À1 s À1 (Figure 2a).The antiferromagnetic property of the material, like in the case of NiO 1+δ , could add uncertainty to Hall measurements. [126,127]For example, the mobility and carrier density values determined by Hall measurements and those derived from electrochemical impedance spectroscopy can vary by a few orders of magnitude. [127]The inclusion of Cu in NiO x :Cu_1 resulted in a reduction of μ h , which decreased from 3.1 cm 2 V À1 s À1 measured in NiO 1.15 to 0.59 cm 2 V À1 s À1 .Ideally, with Cu + substitution (at Ni 2+ sites), the hole mobility of the films should increase according to the chemical modulation of valence band theory. [128,129]However, the μ h enhancement could be concealed by the increased scattering due to grain boundaries or crystallinity degradation. [129,130]With the further addition of Cu, in NiO x : Cu_2, the μ h increased to 2.48 cm 2 V À1 s À1 .
The kinetic effects in NiO 1+δ films were investigated by Hall measurements and the concentration of holes (p Hall ), and their mobilities were determined.The concentration of charge carriers was in addition calculated with the data from combined Kelvin probe (KP) and photoelectron yield spectroscopy (PYS) measurements. [131]Vacancy defects are prevalent in the densely packed NiO structures, whereas the interstitial defects having high formation energies are difficult to form spontaneously. [132] V Ni is known to be the common point defect in NiO (promoting p-type conductivity), however, oxygen-deficient growth conditions are known to promote oxygen vacancies (V O ) ("hole-killer"), which are stable. [132,133][134] Such V O has been identified with photoluminescence investigations in doped NiO nanostructures. [135]Thus, the number of vacancy defects (V Ni and V O ) would influence the hole concentration in NiO 1+δ films.
Charge-carrier density found for as-deposited NiO 1+δ films from Hall measurements was in the range of p Hall 10 13 -10 15 cm À3 (Figure 2a).The concentration of holes from KP-PYS investigations (p KP-PYS ) was calculated by considering for statistical distribution of holes: 1) the valence band maximum energy (E VBM ) relative to the Fermi level (E F ) as calculated with the WF and ionization energy (E i ) data (Figure 2b,c); and 2) the effective density of states in the valence band for an effective hole mass (m h *) of 0.86 m o , [130] where m o is the free electron mass.The hole density p KP-PYS of as-deposited NiO 1.15 films was found as similar to p Hall .For other film compositions, p KP-PYS was up to 2 orders of magnitude lower than p Hall .This is explained by differences between bulk properties as studied by Hall measurements and near-surface properties as investigated by KP-PYS.Note that this concentration of holes, as determined by both Hall and KP-PYS methods, is still too low for the application of the as-deposited films as HTLs in solar cell devices.
To achieve a higher hole density, we have further conditioned the as-deposited thin films by a postdeposition treatment (PDT) in O 3 ambient for 5 min.It is worth mentioning that such a PDT procedure is usually applied during the PSC fabrication before the spin coating with the perovskite film. [47]The measured bulk resistivity (ρ) for the 100 nm thick NiO 1.15 was 1.34 Â 10 5 Ωcm and 3.82 Â 10 2 Ωcm before and after O 3 PDT treatment, respectively.The electronic properties of the NiO 1.15 and NiO x :Cu_1 samples were then studied by KP-PYS.The determined p KP-PYS of the O 3 -PDT conditioned films was found of ≈6 orders of magnitude higher than that of the as-deposited films.The WF determined by KP and the E i measured by PYS for the NiO 1.15 and Cu 0 .10 Ni 0.90 O 1.11 PDT samples almost coincided, thus denoting a near degeneration state (Figure 2b).For as-deposited NiO 1+δ and Cu x Ni 1Àx O 1+δ layers, variations of the WF and E i with the film composition were found of ≈180 and ≈100 meV, respectively, in agreement with Ref. [72] For NiO 1+δ films, an increase in WF and E i , and respectively of E F and E VBM (Figure 2b,c) correlates with the estimated increase of the Ni 3+ concentration with O 3 treatment.138] The device structure of the inverted PSCs based on NiO 1+δ is shown in Figure 3a.The control samples were based on SAM, ([2-(3,6-dimethoxy-9H-carbazol-9-yl)ethyl]phosphonic acid) (also known as MeO-2PACz), [60] instead of the NiO 1+δ as HTL.Among all the devices, the SAM-based devices had superior photovoltaic parameters obtained from current-voltage ( J-V ) characteristics (Figure 3b,e and Table S6, Supporting Information).The J-V characteristics, series resistance (R s ), and parallel resistance (R p ) of the devices are shown in Figure S10, Supporting Information.Among the NiO 1+δ -based solar cells, the one with NiO 1.15 had superior characteristics in terms of best-performing devices.When considering the best-performing devices, the short-circuit current density ( Jsc) of the NiO 1.15 device was comparable to that of the SAM device.However, the former exhibited decreased open-circuit voltage (V OC ), fill factor (FF), and η by 132 mV, ≈8%, and 4.5%, respectively, in comparison to the control device.With the addition of ≈5 at% of Cu into the NiO 1+δ (NiO x :Cu_1 device), a ≈55 mV improvement in V OC of the device was found when compared to the NiO 1.15 device.However, the aforementioned improvement is compensated by a ≈5% loss in FF leveling the η obtained with NiO 1.15 -device.With a higher amount of Cu (≈8 at%) in the NiO 1+δ , all photovoltaic parameters of the device deteriorated (i.e., for the NiO x : Cu_2 device) due to degrading optical and interfacial charge transfer properties as known from the literature. [139,140]The V OC and FF losses in the NiO 1+δ devices are generally attributed to the increased interfacial nonradiative recombination. [141]Additionally, a higher R s is also known to detrimentally affect the FF. [142]The R p of NiO 1+δ devices exhibited a direct correlation with the crystallinity determined by BB-XRD.In other words, higher crystallinity in the sample corresponded to an increase in R p . [140]o further evaluate the performance of the devices, transient photocurrent (TPC), open-circuit voltage decay (OCVD), transient photovoltage (TPV), and capacitance-frequency (C-f ) measurements were performed (Figure 3f,h).It is important to note that a white LED was used for illumination in the aforementioned measurements, unlike the sun-simulator with AM 1.5G spectrum used for J-V characteristics (see Supporting Information for more details).The TPC rise curve pertains to the devices' charge extraction ability, which is accumulated at its transport layers (in short-circuit conditions). [143]The control device, the NiO 1+δ device, and the Cu 0 .10 Ni 0.90 O 1.11 device portray similar features up to 10 2 μs after which the extracted current density of the control device continues to increase.The maximum extracted current density values for all devices are in correlation to the J sc values measured.[145][146] A relatively faster photovoltage decay is associated with the reduced number of surface defect states and deep-level defect states, thus with the better performance of the device. [145]From the OCVD decays, it was found that the control device had a faster decay corresponding to its superior V OC and FF.It was then followed by the Cu 0 .10 Ni 0.90 O 1.11 device which had the best V OC among all the NiO 1+δ -based devices.The TPV decays for the devices are shown in Figure S11, Supporting Information.At a light pulse duration of ≈30 μs, an overshoot of the TPV was found for all devices in the increasing order for control, NiO 1.15 , NiO x :Cu_1, and NiO x : Cu_2 devices which correspond to the charge trapping at the interfaces with the influence of ion migration. [147,148]This observation correlated to the FF, as seen in the devices.
Figure 3h presents the C-f data for the PSCs measured under illumination.The capacitance response under 10 kHz in the PSCs arises from the electrode polarization due to electronic and ionic accumulation. [149]From the C-f data, it is inferred that the ion migration is higher in the Cu-incorporated NiO 1+δ devices and the same was least in SAM-passivated control devices.Light-induced degradation and biasing in PSCs are known to increase the I À ion concentration at the perovskite and chargeÀ transport layer interface. [150,151][155] To investigate the stability of the solar cells, the devices were tracked at their maximum power point (MPP) under a 1-sun condition in a nitrogen ambient at an elevated temperature of 85 °C, and the data are presented in Figure 3i.The spectrum of the lamp used in the stability test and the absolute η of the PSCs from the tracking are shown in Figure S12 and S13, Supporting Information, respectively.The NiO 1.15 device retained more of its initial efficiency (η) compared to the control device, while the Cu x Ni 1Àx O 1+δ devices outperformed others in terms of η retention.One of the reasons why the Cu x Ni 1Àx O 1+δ devices outperform NiO 1+δ devices in terms of stability is the lower Ni 3+ /Ni 2+ ratio in Cu x Ni 1Àx O 1+δ , as determined from the XPS analysis. [156]The surface Ni 3+ of NiO 1+δ is known to react with perovskite resulting in the formation of a PbI 2Àx Br x Àrich, organic cation-deficient perovskite, and, thus, creating energy barriers and recombination centers at the interface. [50]The second reason for improved operational stability is the reduced surface roughness in Cu x Ni 1Àx O 1+δ films compared to NiO 1.15 which is more favorable for coverage of perovskite film on top. [156]59]

Conclusion
In this research, we employed the hollow cathode GFS technique to fabricate NiO 1+δ and Cu x Ni 1Àx O 1+δ thin films.The NiO 1+δ films displayed a preferred orientation along the (111) crystallographic plane, and an increased p(O 2 ) during the deposition led to a higher degree of this orientation.However, the introduction of Cu for Cu x Ni 1Àx O 1+δ films disrupted this orientation, giving rise to the observation of multiple planes in the BB-XRD pattern.The inclusion of Cu, as indicated by XRD and Raman analyses, demonstrated the preservation of the inherent NiO-like structure.XPS unveiled a decrease in the Ni 3+ /Ni 2+ ratio in NiO 1+δ films with an increase in p(O 2 ) during deposition.Additionally, the predominant Cu valency on the surface of Cu x Ni 1Àx O 1+δ was found to be +1 (Cu + ).
The NiO 1+δ and Cu x Ni 1Àx O 1+δ films were successfully demonstrated as HTL in PSCs.Ion-migration effects were found to be higher in the NiO 1+δ device, and it was further pronounced in the case of Cu x Ni 1Àx O 1+δ devices when compared to the SAM device.In the operational stability test at 85 ºC, the NiO 1+δ PSC retained its original efficiency better than the SAM device.Our key finding is that the Cu x Ni 1Àx O 1+δ devices, although exhibited an inferior η compared to the NiO 1+δ Àdevice, outperformed other devices in the stability test (≈90% η retained after 9 d).Additional optimization and exploring interface engineering options such as the use of a passivation layer, bilayer HTL, etc. can further enhance the device performance and operation stability of the PSCs reported here.The GFS NiO 1+δ is a promising substitute for organic HTL and solution-processed NiO 1+δ for mass production of large-area PSCs.

Figure 1 .
Figure 1.a) δ in NiO 1+δ and Cu x Ni 1Àx O 1+δ and b) the at.ratio of Cu/Cu+Ni obtained from EPMA data for NiO 1+δ and Cu x Ni 1Àx O 1+δ films.The Cu x Ni 1Àx O 1+δ films were deposited with either one or two 5 mm Cu-ring targets.c) The calculated texture coefficient (TC (111) ) from BB-XRD and the estimated crystallite size (D) with the Scherrer equation along the (111) crystallographic plane for various 500 nm thick NiO 1+δ films (see Supporting Information for more details).Optical bandgap, at.oxygen/metal ratio from EPMA, and lattice constant determined from BB-XRD for d) NiO 1+δ and e) NiO 1+δ and Cu x Ni 1Àx O 1+δ films deposited at 0.24 Pa p(O 2 ).The gray, blue, and red lines correspond to the optical bandgap, lattice constant, and δ (at.O/metal ratio) in NiO 1+δ , respectively.The lines connecting the data points are a guide to the eye.f-h) High-resolution BB-XRD of the various films deposited at 0.24 Pa p(O 2 ).

Figure 2 .
Figure 2. Thin film electronic properties as a function of composition: a) Hall mobility, charge carrier concentration from Hall measurements, and charge carrier concentration estimated from KP-PYS measurements.The lines connecting the data points are a guide to the eye.b) WF and ionization energy (E i ) were determined from KP and PYS measurements, respectively.c) Valence band maximum position (E VBM ) with respect to Fermi level (E F ).The ozone PDT data are only shown for NiO 1.15 and NiO x :Cu_1 samples.

Figure 3 .
Figure 3. a) Schematic of the PSC architecture.Photovoltaic parameters of the GFS NiO 1+δ -based PSCs: b) short-circuit current ( J SC ), c) open-circuit voltage (V OC ), and d) FF, and, I efficiency (η) from current-voltage ( J-V ) measurement under AM 1.5G illumination.f ) TPC rise, g) normalized OCVD, and h) C-f measurements of the various PSCs.i) Normalized averages of maximum power point tracking-η (average of 7 devices of control, 4 for NiO 1.15 , 5 for NiO x :Cu_1, and 6 for NiO x :Cu_2) of the PSCs under the operational stability test.The spikes and step-like η loss noticed are due to the periodical J-V measurement every 24 h.