Deposition of spinel IGZO thin films with increased indium contents on textured GZO templates

 View Online  Export Citation RESEARCH ARTICLE | NOVEMBER 10 2025 Deposition of spinel IGZO thin films with increased indium contents on textured GZO templates Evangelos Agiannis  ; Hendrik F. W. Dekkers  ; Marta Agati ; Annelies Delabie J. Appl. Phys. 138, 185301 (2025) https://doi.org/10.1063/5.0276210 Articles You May Be Interested In Study of the interfacial reaction of high-κ Al 2O3 gate oxides with IGZO J. Appl. Phys. (August 2025) Design of oxide-based tunnel FETs using amorphous IGZO and p-type oxide semiconductors J. Appl. Phys. (September 2025) Low-temperature deposition of crystalline IGZO films using high-power pulsed magnetron sputtering J. Appl. Phys. (August 2025) 10 November 2025 16:23:47 Deposition of spinel IGZO thin films with increased indium contents on textured GZO templates Cite as: J. Appl. Phys. 138, 185301 (2025); doi: 10.1063/5.0276210 View Online Export Citation CrossMar k Submitted: 16 April 2025 · Accepted: 23 October 2025 · Published Online: 10 November 2025 Evangelos Agiannis, 1,2,a) Hendrik F. W. Dekkers, 2,a) Marta Agati, 2 and Annelies Delabie 1,2 AFFILIATIONS 1 Department of Chemistry, KU Leuven, Celestijnenlaan 200F, Leuven 3001, Belgium 2 IMEC, Kapeldreef 75, Leuven 3001, Belgium a) Authors to whom correspondence should be addressed: [email protected] and Dekk[email protected] ABSTRACT Spinel indium gallium zinc oxide (IGZO) has been identified as an interesting alternative to amorphous IGZO (a-IGZO) due to its higher resilience to the formation of oxygen deficiencies. Currently, it is not understood whether the indium content in spinel IGZO can increase beyond the atomic concentration ratio (mole ratio) of In/(In + Ga + Zn) = 33%. In this study, we explored the deposition and structural stability of spinel IGZO films with increased indium contents. Polycrystalline spinel GZO thin films were used as a template for the growth of polycrystalline spinel IGZO by physical vapor deposition, without active heating of the substrate and without the addition of oxygen in the sputtering gas mixture. Using co-sputtering, we were able to deposit spinel IGZO thin films with different compositions. The use of sputtering with pulsed DC enhanced crystallization into the spinel phase, allowing the deposition of films with In/(In + Ga + Zn) concentration ratios up to 44%. Our results suggest that the indium rich spinel IGZO phase is structurally unstable to decomposition to In 2 O 3 and metallic indium crystallites, during forming gas annealing at 350 °C in a H 2 /N 2 ambient for 1 h. The insights of this work contribute to the development of spinel IGZO thin films with improved properties for various thin-film transistor applications. © 2025 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). https://doi.org/10.1063/5.0276210 INTRODUCTION Indium gallium zinc oxide (IGZO) has been identified as a promising wide bandgap semiconducting oxide for thin-film transistors (TFTs), due to the absence of mobile holes and the relatively high electron hall mobility of its amorphous phase (typically between 10 and 20 cm 2 /V s). 1,2 These characteristics, in combination with the low deposition temperatures (T d ) of the amorphous phase, make IGZO a promising candidate for applications as a semiconducting channel of thin-film transistors (TFTs) in optical displays, 3 flexible electronics, 2 and dynamic random-access memory (DRAM) cells. 4 Despite these advantages, amorphous IGZO (a-IGZO) is sensitive to the formation of oxygen deficiencies, which are mainly associated with indium. Indium atoms contribute substantially to the formation of the conduction band and increase carrier mobility due to the overlap of their extended 5s 0 orbitals. 1 For example, Ref. 5describes the effects of the composition of a-IGZO, on the performance of TFTs with a-IGZO channels. Channels with increased In concentrations were characterized by increased mobilities, more negative threshold voltages, but lower I on /I off ratios. The positive effects of indium on the mobility of TFTs with different IGZO channels were also mentioned in Refs. 6 and 7. On the other hand, indium atoms bind weaker with oxygen, compared to Zn and Ga, enhancing the creation of oxygen deficiencies 8–10 during different processing steps (e.g., annealing). These oxygen-deficient local configurations are stable when double positively charged. 8 This charge is compensated by the addition of delocalized electrons inside the conduction band in order to maintain the overall charge neutrality, leading to an increase in the free electron concentration. Since oxygen atoms are more strongly embedded in crystalline IGZO, n-type doping is expected to be better controlled in the crystalline phase, making the material electrically more stable. Hexagonal IGZO (h-IGZO) is the most stable crystal structure of IGZO. This phase consists of two polytypes represented by space groups R 3m (160) and P63/mmc (194). 11 Their structure comprises InO 2 sheets that laminate Ga x Zn y O (x + y) layers. C-axis aligned crystalline (CAAC) IGZO is a semi-crystalline polytype of IGZO, derived from the hexagonal phase. Here, the structure has no Journal of Applied Physics ARTICLE pubs.aip.org/aip/jap J. Appl. Phys. 138, 185301 (2025); doi: 10.1063/5.0276210 138, 185301-1 ©Author(s)2025 10 November 2025 16:23:47 periodicity along the a and b axes but maintains a layered arrangement along the c axis. 11 Spinel IGZO, the other crystalline polytype, is represented by space group Fd 3m (227) and can be formed using PVD 12 [Fig. 1(a)]. The sixfold oxygen-coordinated (octahedral) positions are occupied by Ga 3+ and In 3+ cations, while Zn 2+ occupies the fourfold oxygen-coordinated (tetrahedral) sites. This crystal structure results in a unique configuration for oxygen anions, which are always neighbored by only one Zn 2+ cation. Therefore, the oxygen atoms are more strongly embedded in the spinel IGZO lattice compared to h-IGZO. 9 Gallium zinc oxide (GZO) is stable in the spinel phase, but gradual replacement of gallium by indium atoms makes it unstable compared to h-IGZO. Despite this instability, high indium concentrations can be achieved with physical vapor deposition (PVD) using pulsed DC (p-DC) power. 12 The different crystalline phases of IGZO can be obtained with PVD by modifying deposition parameters such as the substrate temperature (T sub ) and the O 2 flow ratio (RO2) in the Ar/O 2 gas mixture, and by employing different post-deposition techniques like annealing. 13 Spinel IGZO can be deposited under intermediate process conditions compared to the amorphous and CAAC phases. 13 Polycrystalline spinel IGZO, without significant quantities of additional phases, is formed when a polycrystalline spinel GZO template is used. The templating effect of GZO films with thicknesses between 1.8 and 3.4 nm was previously reported in Ref. 12. The structure and orientation of the template propagate into the IGZO layer that is grown on top. One of the challenges related to the use of IGZO for electronic applications is maintaining its structural and electrical stability in the presence of hydrogen. Forming gas annealing (FGA) in a H 2 /N 2 ambient is a process often used in the semiconductor industry for passivating dangling bonds at semiconductor/dielectric interfaces. However, positively charged interstitial hydrogen is stabilized near oxygen anions in IGZO, making hydrogen an n-type dopant. 8,14,15 Moreover, during FGA, IGZO may decompose, resulting in metallic cations at the surface of the film, especially metallic indium as the indium content in the film increases. Precipitation of any metallic cations will likely result in further degradation of the electronic properties. So far, the PVD of spinel IGZO could be obtained in a single target p-DC PVD tool (Applied Materials Impulse) using a polycrystalline IGZO target with In:Ga:Zn atomic concentration ratios equal to 1:1:1, and with the best crystallinity being observed at T sub = 200 °C and RO2¼90%. 12 The structural stability of spinel IGZO films with different compositions has not yet been studied. More specifically, exploring the possibility of increasing the In content and, thus, improving the carrier mobility of the films could pave the way for high mobility and electrically stable spinel IGZO TFT channels. In this study, the deposition of textured spinel IGZO thin films with different indium concentrations was investigated and the structural stability of spinel IGZO was determined. Our work focused on the deposition of IGZO in the spinel phase on top of the GZO template by PVD without the intentional heating of the substrate to avoid the formation of CAAC and h-IGZO, as is typically the case at elevated temperatures. 12 Since the addition of oxygen to the PVD process enhances the spinel formation, the effects of RO2on the deposition of spinel IGZO were first examined. Next, the possibility of obtaining the spinel phase with an as high as possible indium concentration, by using only radio frequency (RF) PVD, was investigated. This was followed by adding p-DC PVD from the IGZO target. The presence of the amorphous phase in the spinel IGZO films was always monitored by annealing the films at 700 °C in O 2 for 1 h. This process crystallizes any amorphous parts to h-IGZO. Finally, we tested the structural stability of films with different indium concentrations by subjecting them to FGA at 350 °C for 1 h. EXPERIMENTAL SECTION In this work, the atomic composition of the films was modified by co-sputtering from multiple targets positioned around the rotating substrate [Fig. 1(b)]. The deposition of both GZO and IGZO films was performed using the “multi-cathode”chamber of an Applied Materials Endura platform. During the deposition of the films, the substrates were effectively cooled by electrostatic clamping on a watercooled rotating chuck. However, it must be noted that during sputtering,thetemperatureatthefilmsurfacemightbeofanincreased value, due to the impact of the sputtered species. 16 The cathodes were equipped with a 6 in. polycrystalline IGZO target (In:Ga:Zn = 1:1:1) that was connected to a p-DC generator and three ZnO, Ga 2 O 3 ,and In 2 O 3 targets, connected to RF generators [Fig. 1(b)]. With this FIG. 1. (a) The crystal structure of spinel IGZO represented by the cubic space group Fd  3m (227) with interchanging indium and gallium atoms. (b) A multi-cathode chamber with five targets. The substrate, a 300 mm wafer, is positioned at the center. Journal of Applied Physics ARTICLE pubs.aip.org/aip/jap J. Appl. Phys. 138, 185301 (2025); doi: 10.1063/5.0276210 138, 185301-2 ©Author(s)2025 10 November 2025 16:23:47 configuration, it is possible to deposit films with different compositions by modifying the sputtering rate of each target. All targets had fixed distances from the substrate at angles of 40° to the normal of the substrate. The pulses of the p-DC generator had a frequency of 100 kHz, and the duty cycle was set to 10%. Before each deposition, the wafers were placed in a degassing chamber at 350 °C in an Ar ambient at 3–10 Torr for 60 s. Before entering the PVD chamber, each wafer was cooled for 60 s at a cooling station. Sputtering was interrupted every approximately 60 s to avoid overheating of the targets. During this time, the substrates’surface could cool down. GZO and IGZO deposition For obtaining the spinel GZO template, we used thermally oxidized 12 in. (100) Si wafers and co-sputtering from the RF-powered Ga 2 O 3 and ZnO targets in the multi-cathode chamber. As mentioned before, the crystallization of the initially amorphous GZO layer into the spinel phase happens during annealing at 700°C for 1 h in an atmospheric furnace (ASM A412) under an O 2 atmosphere. An optimally textured GZO template was achieved at a thickness of 3nm. For the deposition of IGZO, the chamber pressure was set to 1.8mTorr by regulating the total gas flow at 50 SCCM. Films with varying indium concentrations were deposited from the atomic concentration ratio (mole ratio) of In/(In + Ga+ Zn)= 6% (Ga rich) to In/(In + Ga + Zn) = 72% (In rich) with the Zn/(In + Ga + Zn) ratio maintained close to 33%. Three sets of IGZO films were deposited for analysis. The first set was deposited by only p-DC sputtering from the IGZO 1:1:1 target. The second set was deposited by only RF co-sputtering from the ZnO, Ga 2 O 3 , and In 2 O 3 targets, while the third set was deposited by co-sputtering with RF and p-DC power. All IGZO films had thicknesses of approximately 22 to 25 nm. Thickness measurements were performed using the F5-SCD (KLA Tencor) metrology platform. Annealing at 700 °C in 1 atm of O 2 for 1 h was used to identify the presence of possible amorphous fractions in the as-deposited IGZO films. During this process, a-IGZO crystallizes into h-IGZO, causing an additional peak in the out-of-plane x-ray diffraction (XRD) pattern at approximately 2Θ= 31.4°. 17 Annealing at temperatures up to 700 °C does not affect the spinel crystallites. 12 Finally, for determining the stability of the spinel phase in a reducing environment, different films were annealed in forming gas (10% H 2 in N 2 ) at 350 °C for 1 h in the same ASM A412 furnace. Material characterization The atomic composition of the films was measured by wavelength-dispersive x-ray fluorescence spectroscopy (XRF) using a Malvern PANalytical 2830ZT Wafer Analyzer equipped with an Rh x-ray source, set at 32 kV and 125 mA. Zn K-α,GaK-α,InL-α, and O K-αwere calibrated using an x-ray photoemission spectroscope (UlvacPhi, Quantes, Al K-αand Cr k-αsources). The measurements were performed using a 40 mm diameter x-ray spot at the center of the wafer. The absolute accuracy of the composition values was estimated within ±2%, as calculated from the systematic errors of the XRF measurements. For obtaining the metal concentrations of the IGZO films that were deposited on top of the template, the Ga and Zn signals that originated from GZO were subtracted from the overall measurement. The XRD and x-ray reflectivity (XRR) measurements were performed using a Jordan Valley JVX tool with a Cu Kαsource. For the out-of-plane Θ–2ΘXRD, a 1° sample tilt was employed. Since the higher intensity spinel IGZO diffraction peak is the one that originates from the (222) planes, 12 the 2Θrange was scanned between 25° and 38°. The grazing incidence (GI) XRD measurements were performed at an incidence angle of ω= 1°. For characterizing the texture of the films, Θ–ωscans were used with the detector and incident beam fixed at a characteristic diffraction angle, while ωwas scanned by tilting the substrate. To suppress the forbidden Si peaks during XRD, the wafers were rotated by w= 22°. For obtaining the films’mass density values, diffraction patterns from XRR measurements were used. The theoretical mass densities that correspond to defect-free crystals were extracted using the ρ¼ZM NA Vc formula with Z being the number of formula units per unit cell, M being the molar mass of the formula unit, NA being the Avogadro number, and V c being the volume of the unit cell. The calculation of the molar mass incorporated the atomic concentration ratios that were extracted by XRF measurements, while the calculation of V c was performed using the lattice constant values that were extracted by in-plane and out-of-plane XRD measurements. Transmission electron microscopy (TEM) characterization was employed to perform further structural analyses. TEM characterization was performed using a double corrected Titan3 G2 60–300 microscope by Thermo Fisher, with 200 kV of acceleration voltage. To accomplish the energy dispersive x-ray (EDX) measurements, the microscope was equipped with a Super-X system, where four silicon drift detectors, each with an area of 30 mm 2 , formed a solid angle of approximately 0.7 sr. The detectors were equally separated at 45° from the holder tilt axis at an elevation angle of approximately 18° from the horizontal axis. TEM sample preparation was performed, using a Helios 460 focused ion beam (FIB) system by Thermosteric, employing spin on carbon (SoC) and Pt capping layers. RESULTS AND DISCUSSION Template assisted PVD of spinel IGZO First, we describe the XRD pattern of a 24 nm thick spinel IGZO film that was sputtered using the IGZO 1:1:1 target with 500 W of p-DC power. In Ref. 12, the templating effect of GZO with different thicknesses was examined, with the results suggesting that the layers with thicknesses below 1.8 nm fail to act as a reliable template. For this reason, deposition was performed on a 3 nm GZO template without any intentional heating of the substrate and without adding any oxygen in the sputtering gas mixture. The film was then subjected to a 700 °C crystallization anneal for 1 h. The cation ratios In:Ga:Zn were calculated from XRF measurements to be 33%:43%:24%. The difference in the cation ratios, when compared to the target (In:Ga:Zn = 1:1:1), agrees with previous observations in the literature where IGZO was also deposited by PVD. 18 One possible explanation for this difference is the resputtering of Zn atoms from the film, and their substitution by Ga, allowing them to occupy tetrahedral sites in the spinel lattice, as is the case with γ-Ga 2 O 3 . 19,20 Journal of Applied Physics ARTICLE pubs.aip.org/aip/jap J. Appl. Phys. 138, 185301 (2025); doi: 10.1063/5.0276210 138, 185301-3 ©Author(s)2025 10 November 2025 16:23:47 The resputtering of Zn atoms from ZnO and Zn films was previously studied in Ref. 21. There, the important role of O − ions in the resputtering process of Zn was pointed out. In the out-of-plane XRD diagram [Fig. 2(a)], the peaks observed at 18.7°, 36.4°, and 57.2° are assigned to the (111), (222), and (333) spinel IGZO planes, respectively, indicating that the IGZO film was textured. Additional diffractions from planes that are not aligned with the surface of the sample can be observed in Fig. 2(b), which shows the grazing incidence (GI) XRD pattern. The smaller diffraction angle of the (111) planes allows their detection by GI XRD. Although additional phases were not detected by the TEM cross section analysis and annealing at 700 °C followed by XRD measurements, small quantities might still be present between the spinel crystallites of every IGZO film deposited for this work. Their limited presence inside the film could be the reason why they were not detected by out-of-plane XRD. In Fig. 2(b), the broadening of the peak at 2Θ= 33.4° may be due to the convolution of the diffraction from the (311) spinel planes and a weak diffraction signal that might originate from small quantities of hexagonal 17 or CAAC 11 IGZO. The in-plane misalignment of the (311) spinel planes is probably an additional reason for the broadening of this peak. Figure 3(a) shows the comparison of the out-of-plane XRD pattern of the spinel IGZO film of Fig. 2,beforeandafterthe 700 °C annealing step. XRF measurements did not reveal any differences in composition. The shift of the (222) peak to larger angles after annealing is due to the reduction in the crystal lattice constant from 8.7 to 8.6 Å. This is caused by the elimination of defects inside the crystal. Metal atoms move to energetically more favorable positions, resulting in fewer structural defects that cause the expansion of the lattice. Examples of such defects are theexchangebetweentheZnandGaatomsintheoccupationof tetrahedral and octahedral sites, 22 and the occupation of nonspinel lattice sites. 19 Moreover, annealing at higher temperatures, andinanoxygenenvironment,bringsthecrystaltoastoichiometric balance. Figure 3(b) shows the deposition of 25 nm of IGZO on top of the textured 3 nm GZO template, using different RO2and the IGZO 1:1:1 target with 500 W of p-DC power. During depositions, there was no intentional heating of the substrate. As previously described, annealing at 700 °C in an O 2 ambient was used to identify the presence of amorphous parts in the as-deposited films. 17 It can be seen that spinel IGZO films were grown on top of the template for every RO2. The template’s texture was transferred into the IGZO layer that was grown on top, while a slightly different (222) peak was observed for RO2¼0%, with the rest of the films resulting in similar peak intensities and widths. It must be noted that as RO2increases, the deposition rate drops significantly, leading to much longer deposition times that help the material crystallize. 23 Moreover, increasing the RO2increases the number of oxygen interstitials (as it is described in the section titled Spinel IGZO with increased indium concentration), which act as electron traps in IGZO channels. 8,13 For these reasons, our work focused on the study of PVD of spinel IGZO at RO2¼0%. In summary, textured spinel IGZO was deposited on crystallized spinel GZO without the addition of oxygen in the sputtering gas mixture and without heating of the substrate. The deposition of spinel IGZO was also observed for different RO2(4%, 10%, and 86%) when the GZO template was employed. The 3 nm thick polycrystalline GZO layer was identified as a reliable template for the deposition of textured spinel IGZO films. Spinel IGZO with increased indium concentration Next, we investigated the spinel formation for IGZO films with increased indium concentrations. First, we examined the spinel formation by co-sputtering from the ZnO, Ga 2 O 3 , and In 2 O 3 targets employing only RF power. Figure 4(a) shows the out-of-plane XRD patterns of films that had the Zn/(In + Ga + Zn) atomic concentration ratio always close to 33% and varied the In to Ga ratio, i.e., (In x Ga 1−x ) y ZnO 4 , where y is approximately equal to 2 and x is the In/(In + Ga) atomic concentration ratio, quantifying the percentage of octahedral lattice sites that are occupied by indium atoms. All films were later annealed at 700 °C in 1 atm of O 2 for 1 h for examining the presence of a-IGZO. For the case of In/(In + Ga) = 29%, a wide peak at approximately 31.2°, which corresponds to diffraction mainly from the (009) planes of h-IGZO, 17 appears after annealing. At the same time, two lower intensity peaks at 2Θ= 34.6 o and 2Θ=36.1 o that correspond to the (104) and (105) planes, respectively, will appear in the GI XRD pattern 17 (not shown). This indicates the presence of significant parts of a-IGZO in the as-deposited films with In/(In + Ga) ≥29%. For the films with In/(In + Ga) up to 21% (i.e., In:Ga:Zn at 14%:53%:33%), no additional diffractions can be seen. Hence, the use of RF limits the formation of the spinel FIG. 2. Out-of-plane (a) and grazing incidence (after subtracting the background) (b) XRD patterns of spinel IGZO on 3 nm of spinel GZO after a 700 °C anneal in an O 2 ambient for 1 h. The vertical lines represent the simulated diffractions of spinel [(a) blue dots] and hexagonal [(b) black dots] IGZO. Journal of Applied Physics ARTICLE pubs.aip.org/aip/jap J. Appl. Phys. 138, 185301 (2025); doi: 10.1063/5.0276210 138, 185301-4 ©Author(s)2025 10 November 2025 16:23:47 phase, while with p-DC only, In/(In + Ga) = 45% can be reached. Therefore, higher indium concentrations can possibly be achieved with the aid of p-DC. For this reason, co-sputtering with RF power from the ZnO, Ga 2 O 3 ,andIn 2 O 3 targets and p-DC power from the IGZO 1:1:1 target was tested. In this case, we deposited spinel films with In/(In + Ga) up to 63% [Fig. 4(b)]. The shift of the spinel (222) peak to lower angles indicates an increase in the crystal lattice constant as the indium concentration increases. The relatively larger peak FIG. 3 (a) Out-of-plane XRD of 24 nm of IGZO deposited on 3 nm of GZO without actively heating the substrate and with RO2¼0%, before and after annealing at 700 °C in 1 atm of O 2 for 1 h. (b) out-of-plane XRD of the (222) diffraction peaks of 25 nm of spinel IGZO as-deposited on 3 nm of GZO, for different RO2and after a 700 °C crystallization annealing step. FIG. 4. (a) Θ–2ΘXRD of 24 nm spinel IGZO films with approximately 33% Zn/(In + Ga + Zn) that were deposited using the ZnO, Ga 2 O 3 , and In 2 O 3 targets and RF power. Without p-DC, the maximum In/(In + Ga) ratio for the deposition of films without significant quantities of additional phases is 21%. (b) Films that were deposited using the IGZO 1:1:1 target with p-DC in combination with the Ga 2 O 3 ,In 2 O 3 , and ZnO targets powered by RF. As the indium concentration decreases, the spinel (222) peak shifts to higher values, indicating a decrease in the crystal lattice constant. The formation of spinel was limited to In/(Ga + In) up to 63%. Journal of Applied Physics ARTICLE pubs.aip.org/aip/jap J. Appl. Phys. 138, 185301 (2025); doi: 10.1063/5.0276210 138, 185301-5 ©Author(s)2025 10 November 2025 16:23:47 widths observed in the XRD patterns of the films with high indium contents suggest smaller grain sizes. We attribute this increase in the structural stability range when the IGZO 1:1:1 target with p-DC power was used to the bombardment of the substrate with O − ions that originate from the target and are accelerated to higher speeds due to the use of p-DC. During PVD, this could cause an increase in adatom mobilities, making crystallization into the spinel structure easier. The bombardment with oxygen ions is known to have a significant influence on the crystallization of metal oxides during low-temperature PVD. 24–26 Mráz and Schneider 24 studied how the bombardment with O − affects the growth of transition metal oxide thin films during PVD. They suggested that the bombardment with high and medium energy O − enables the crystallization during lowtemperature PVD. Bikowski et al. 25 investigated the effect of O − ion bombardment on the electronic and structural properties of ZnO:Al films deposited by magnetron sputtering. They concluded (among others) that the size of the crystallites and the number of defects are affected by the type of power. In Ref. 26, an overview of the production mechanism and the effects of sputtered ions on the properties of the thin films, during reactive sputtering, is given. There, it is pointed out that these effects can be controlled by modifying technical parameters like the substrate-to-target distance and deposition pressure. The results of Ref. 12 show that the formation of spinel IGZO crystallites is induced by the presence of O 2 in the sputtering gas mixture, indicating that the bombardment with O − ions enhances crystallization into the spinel structure. We note that due to the low deposition rate of the In 2 O 3 target, the deposition of films with increased indium concentrations required the decrease of the contribution from the IGZO 1:1:1 target, leading to lower p-DC powers and, thus, a less pronounced O − bombardment. This could be the reason why the formation of spinel IGZO was limited to In/(In + Ga + Zn) ratios up to 44%. Higher indium concentrations could potentially be achieved by employing sputtering with p-DC power from targets with higher indium contents. For further examining the effects of the O − bombardment on the deposition of spinel IGZO, we compare the mass density of films that were deposited with and without the contribution of p-DC power. Figure 5(a) shows the mass density values of different spinel IGZO films deposited on top of 3 nm of spinel GZO, as a function of their In/(In + Ga) atomic concentration ratio. Mass densities were extracted from XRR measurements and were grouped into two categories: The ones that were deposited using co-sputtering with p-DC and RF power (solid blue points), and those that were sputtered using only RF power (empty blue points). When sputtering with p-DC power was employed, the mass densities were found to have relatively higher values when compared to films deposited using only RF. We suggest that this is due to the incorporation of more oxygen interstitials into the films. This difference seems to become more pronounced as the p-DC power increases. For this reason, the solid blue point at In/(In + Ga) = 8% (deposited using only 50 W of p-DC power) has a mass density similar to the empty blue point at approximately 6% In/(In + Ga). The linear trend until the In/(In + Ga) = 41% value is probably the result of both the increasing In concentration and the higher p-DC powers employed for the deposition of these films, which could have resulted in more oxygen interstitials. The observed saturation of the mass densities after this point is probably caused by the gradual decrease in the p-DC power from 400 W, for the case when In/(In + Ga) = 42% to 160 W when In/(In + Ga) = 64%. FIG. 5. (a) The mass density values of films with different In/(In + Ga) concentration ratios. The solid blue cycles originate from films that were deposited using co-sputtering from multiple targets, including sputtering with p-DC power (IGZO 1:1:1 target). The empty blue cycles originate from films that were deposited using only RF power (ZnO, Ga 2 O 3 , and In 2 O 3 targets). (b) The difference between the theoretical expected and the experimentally extracted mass density values of films that were deposited using the IGZO 1:1:1 target with p-DC power and different RO2. The calculation of the theoretical values incorporated the measured (XRF) film compositions and the lattice constant values that were extracted from in-plane and out-of-plane XRD measurements. Journal of Applied Physics ARTICLE pubs.aip.org/aip/jap J. Appl. Phys. 138, 185301 (2025); doi: 10.1063/5.0276210 138, 185301-6 ©Author(s)2025 10 November 2025 16:23:47 For validating the previous hypothesis regarding the incorporation of more oxygen interstitial atoms when p-DC power is being used, we used XRR measurements to extract the mass density of films that were deposited using the IGZO 1:1:1 target with p-DC power and different RO2.InFig. 5(b), the difference between the theoretically expected (that correspond to defect-free crystals) and the experimentally extracted mass density values of these films is given. All films had similar stoichiometries, with In/(In + Ga) atomic concentration ratios that ranged from approximately 45% (for the case of RO2¼0%) up to 52% (for the case of RO2¼86%). Slightly enhanced indium contents could be explained by a more pronounced resputtering of Zn and Ga atoms. 18 The increasing trend is probably the result of the incorporation of more oxygen interstitials into the crystal structure, as the oxygen flow during sputtering increases. For the films deposited using RO220%, this difference was positive. When considering the fact that a certain number of defects (e.g., grain boundaries), which are expected to decrease the films density, cannot be avoided, the previous observation indicates the existence of oxygen interstitial atoms. After the RO2¼20% value, the mass density saturates. Overall, the previous observations suggest that in addition to the increase in adatom mobilities, the bombardment with O − causes the incorporation of more oxygen interstitials into the films. Figure 6(a) is the ternary diagram that shows the different compositions of the films of Fig. 4. The line represents the optimal spinel IGZO stoichiometry where all tetrahedral positions are occupied by Zn atoms. The existence of spinel IGZO films with Zn concentrations that are slightly higher than 33% indicates that Zn atoms can occupy octahedral sites, as is also the case with spinel GZO. 20 This happens until a concentration limit is reached, after which additional phases appear. Figure 6(b) shows the crystal lattice constant, calculated from the out-of-plane XRD measurements, as a function of In/(In + Ga) cation concentration ratio for the spinel IGZO films of Fig. 4, before and after annealing at 700 °C. The solid points correspond to films sputtered using a combination of RF and p-DC power [Fig. 4(b)], while the empty points correspond to films sputtered using only the unary oxide targets with RF power [Fig. 4(a)]. The solid black point is the GZO lattice constant value found in the literature. 27 The linear trend follows Vegard’s law, meaning that the lattice constant is the weighted mean of the lattice parameter of the two individual constituents (in this case, In 2 ZnO 4 and Ga 2 ZnO 4 ). The slightly different trend for the case when only RF power was used could be the result of fewer defects being created during PVD due to the lower kinetic energies of sputtered ions and fewer interstitial oxygen atoms. It can be seen that the lattice constant of every film decreases after annealing as a result of the elimination of defects and the reduction of interstitial oxygen atoms (as described in the section titled Templated assisted PVD of spinel IGZO). This trend is less pronounced for the films that were deposited using only RF power, probably due to the smaller number of defects created during PVD. For further examining the templating effect of the GZO layer and for verifying the uniform distribution of metal atoms in the film, TEM measurements were performed on the cross section of an as-deposited film with increased indium content (In:Ga: Zn = 42%:26%:32%). TEM confirmed the polycrystalline spinel structure of the IGZO layer. Figure 7(a) shows the Fast Fourier Transform (FFT) images of areas below and above the interface of FIG. 6. (a) Ternary composition diagram of spinel IGZO films that were deposited on top of 3 nm of GZO using the co-sputtering deposition approach. The line represents the optimal spinel IGZO stoichiometry, where all tetrahedral positions are occupied by Zn atoms. (b) The crystal lattice constant, before and after annealing, calculated from the out-of-plane XRD measurements, as a function of the In/(In + Ga) concentration ratio. Journal of Applied Physics ARTICLE pubs.aip.org/aip/jap J. Appl. Phys. 138, 185301 (2025); doi: 10.1063/5.0276210 138, 185301-7 ©Author(s)2025 10 November 2025 16:23:47 IGZO with the GZO template. The templating effect is evident from the continuous propagation of the crystal orientations from the spinel GZO template to the spinel IGZO that is grown on top, indicating that the orientation of the spinel GZO grains is transferred to the spinel IGZO grains. The columnar grain growth of IGZO can be observed in the annular dark field scanning TEM (ADF-STEM) image that is given in Fig. 7(b). No regions containing a-IGZO could be detected. The EDX elemental maps [Fig. 7(d)] reveal a FIG. 7. (a) TEM cross section of 24 nm of as-deposited spinel IGZO (In: Ga:Zn = 42%:26%:32%) on top of 3 nm of GZO, and the fast Fourier transform images from the highlighted areas, indicating that the orientation of the GZO grains propagates into the IGZO layer. (b) ADF-STEM image of the cross section, showing the columnar growth of the grains. (c) EDX line scans showing the atomic concentrations of the GZO/IGZO stack. (d) EDX elemental maps showing the homogeneous distribution of the O, Zn, Ga, and In atoms in the IGZO layer. FIG. 8. Θ–2Θ(a) and GI (after subtracting the background) (b) XRD of spinel IGZO films with different indium concentrations, after FGA at 350 °C. The peaks seen at 30.6 o and 33.0 o correspond to In 2 O 3 and metallic indium, respectively. The vertical lines represent the simulated diffractions of spinel IGZO. The decrease in the indium content suppresses the formation of indium and In 2 O 3 crystallites. Journal of Applied Physics ARTICLE pubs.aip.org/aip/jap J. Appl. Phys. 138, 185301 (2025); doi: 10.1063/5.0276210 138, 185301-8 ©Author(s)2025 10 November 2025 16:23:47