Enabling high-quality transparent conductive oxide on 3D printed ZrO 2 architectures through atomic layer deposition

The conformal atomic layer deposition of a transparent conductive oxide composed of Al-doped ZnO (AZO) over three-dimensional (3D) shaped ZrO 2 microarchitectures produced using two-photon lithography (TPL) is re-ported here for the first time. The effect of ZrO 2 morphology, surface roughness


Introduction
The combination of optical transparency while maintaining electrical conductivity is vital for fabricating advanced electrodes, functioning as building blocks across optoelectronic devices, such as displays, [1] photovoltaic cells, [2] thin-film transistors, [3] or lightemitting diodes.[4] Transparent conductive oxides (TCOs) are the most investigated and applied among the various materials offering these characteristics.Fundamentally, TCOs must be highly transparent within the visible range (a band gap > 3 eV) and exhibit low electrical resistivity (ρ < 10 -4 Ω cm).[5,6] The most prevalent examples of TCOs include n-type derivatives of In 2 O 3 , SnO 2 , and ZnO.[7] Al-doped ZnO (AZO) is an attractive alternative that does not compromise optical and electrical properties, which are essential in electronics used in information systems.The precise control of Al doping in AZO is crucial to tailor its electronic and optical properties.For instance, Deng et al. studied the influence of the Al content (0-8.06 at%) within ZnO wurtzite films.[8] The authors demonstrated that with increased dopant concentration, the optical bandgap abruptly increases (3.26 eV for ZnO and 3.50 eV for AZO with 1.80 at.%Al) and gradually shifts towards higher values, up to 3.66 eV for 4.50 at.%Al. [8] The optical bandgap increase is associated with the Burstein-Moss effect, i.e., the absorption edge is blue-shifted due to the increased population of states close to the conduction band.[8,9] Alongside the optical bandgap shift, the average transmittance (in the 400 -1500 nm range) is also higher for the AZO (1.80-5.24at.%Al) than for pristine ZnO.[8] Such transparency and reduced resistivity have allowed AZO to be used in solar cells as conductive electrodes.In another work, Kan et al. reported that indium tin oxide (ITO) replacement by AZO in a polymer solar cell is possible.[10] The results reveal similar power conversion efficiencies, if not better, than ITO.[10] The power conversion efficiencies are attributed to reduced resistivity and increased transparency compared to undoped ZnO.Low resistivity is associated with the Al substitution effect in ZnO, resulting in oxygen vacancies (V O ) or interstitial Zn (Zn i ).[8] In most cases, AZO is utilized in a film form to minimize the light absorption of conductive layers by reducing the optical path length.Various thin film fabrication approaches have been developed for AZO, i.e., sputtering, [11] electrodeposition, [12] chemical vapor deposition, [13] and atomic layer deposition (ALD).[14] ALD offers several advantages, including exceptional conformality, uniformity, and precise thickness control (Åµm range).[15,16] The films are usually free of inhomogeneities and discontinuities.[15] Furthermore, the precise control of the chemical composition is advantageous, especially the possibility of introducing dopants in a well-controlled fashion.[15] Recently, a novel ALD technique called atomic partial layer deposition (APLD) was formulated for the conformal growth of complex oxides, such as HfO 2 -TiO 2. [17,18] Later, Ramírez-Esquivel et al. used the APLD method for growing AZO films with Al doping cycle contribution up to 4 at%.[19] APLD relies on the variation of the metal precursor flow rates and the exposure time to deposit two or more species on the same surface (depicted in Fig. 1), which is not feasible with conventional ALD.[17] Shortcomings have been addressed, such as the formation of undesired aluminum oxide or metallic aluminum domains at higher doping percentages (Al concentration > 4 at.%). [19][20][21][22] Despite the possibility of the APLD for the fabrication of well-controlled alloys on a single atomic layer scale, the studies on AZO grown using ALD/APLD are oriented towards the film deposition over planar surfaces, i.e., glass substrates, [19] probably because of industrial demands on planarized systems.However, there are examples of AZO deposition over complex structures, i.e., nanopillars, [23] nanowires, [24] or nanostructured glass.[25] Even ALD deposition over genuinely 3D microarchitectures has been the subject of only a limited number of studies, including Al 2 O 3, [26,27] SiO 2 , [28] and TiN, [29] with none exhibiting characteristics similar to AZO.
The common feature among the mentioned 3D microarchitectures and the referred works is the usage of two-photon lithography (TPL) to fabricate the 3D polymeric scaffolds as templates for the deposition of the inorganic materials.[26][27][28][29] However, such acrylate-based polymers are usually not directly applicable in optoelectronic technologies unless unreacted functional groups (e.g., vinyl) on the surface of 3D-printed polymeric features are employed to promote targeted functionalization.[30] From a surface chemistry perspective, using ceramic 3D microarchitectures as templates is an attractive option to maximize the geometrical surface area of a TCO, which functionality can be applied over relatively high temperatures without the risk of decomposition.For example, ZrO 2 (zirconia) owns outstanding mechanical feasibility, nontoxicity, chemical stableness, and high affinity to oxygen-including groups.[31] A key property to account for ZrO 2 /AZO architectures is understanding the effect of the underlying substrate crystallinity and shape that might affect the optoelectronic characteristics of deposited AZO thin films.[32] A way to generate such understanding is using the APLD method optimized for AZO deposition over glass and silicon.Important to mention is that no studies have been conducted on complex 3D ceramic structures varying the Al doping levels (>4%).[19] In this regard, merging APLD and TPL holds the potential to pave the path toward a new generation of 3D devices in which the AZO (nano)layer can serve as an enabling role, such as a transparent electrode for light emission.
In this work, a fundamental investigation to expand the applicability of AZO as an optoelectronic component is driven by two strategies: (i) nucleation study of AZO over the non-studied 3D solid-beam microstructured ZrO 2 and (ii) tunning the AZO electrical and optical properties by varying the Al dopant concentrations to demonstrate that AZO can be conformally deposited over complex ZrO 2 architectures.The results can grant the possibility of exploring the fabrication of 3D optical microstructured materials to direct light-matter interaction in optoelectronics.

APLD of AZO
Ceramic 3D micro-optics is an emerging field where additive manufacturing methods can contribute.[33] However, 3D micro-optics might require electrical conductive interfaces to demonstrate functionality.Therefore, in this section, the deposition of AZO nanolayer is studied over complex high refractive index 3D ZrO 2 microstructures fabricated via TPL.The 3D ZrO 2 microarchitectures are manufactured on diced pieces of Si, following the methodology of Winczewski et al. [34] Three different ZrO 2 architectures (octet-truss lattice, gyroid, and C 60 buckyball-inspired structure, denoted as a buckyball) are selected as a platform for testing the geometrical and conformality limits of APLD.
APLD relies on mixing two metalorganic precursors in a specific number of cycles to facilitate doping.In this research, diethylzinc (DEZ) is used as a precursor forming the ZnO matrix of the thin film, while trimethylaluminum (TMA) is the Al-rich dopant.The schematic representation of the APLD process is presented in Fig. 1.The typical cycles (tc) and doping cycles (dc) are utilized.In the dc, DEZ is introduced in a sufficiently short pulse to produce an unsaturated surface with few remaining reactive -OH groups on the surface (Fig. 1 (a-b)).A nonsaturated layer of monoethyl zinc is adsorbed and controlled with the exposure time, in which ethane is released as a side-product.[35] It has previously been reported that DEZ exposures of 0.1---0.24s do not result in total surface saturation, as the growth per cycle (GPC) variation is below 0.05 Å/cycle.[36] A DEZ pulse is followed by a short TMA pulse, which is the base for the APLD method (Fig. 1 (b-c)).As a result, dimethyl aluminum and monomethyl aluminum sites are introduced on the remaining sites on the same atomic layer, and methane is released as a by-product.[37] In the case of AZO deposition, a 0.24 s dosing exposure leads to a nearly saturated surface that ensures efficient doping using the TMA precursor.[36] Consequently, TMA is dosed to react with the remaining groups or adsorption sites to incorporate a small quantity of aluminum-rich reagent (Fig. 1 (b-c)).For obtaining AZO 4%, AZO 4-5 %, and AZO 5 %, the pulse duration and number of TMA pulses during the second subcycle are modified, as indicated in Table S1.Finally, an oxidation pulse to hydroxylate and thus reactivate the surface is conducted (Fig. 1 (c-d)).At this point, a single dc is completed, and zinc and aluminum species co-exist on the same layers separated by complete oxygen layers (Fig. 1 (d)).As by-products of reactions, ethane, and methane are released.[35,37] In the case of tc, DEZ saturating pulse is followed by the purge, H 2 O oxidation pulse, and a purge, which results in the formation of a ZnO layer.Coverage of the entire surface is achieved by repeating tc and dc at the accurate precursor dosage and exposure times to saturate the surface.A detailed description of the methodology can be found in the experimental section.Furthermore, it is important to mention that the APLD method was previously optimized for AZO deposition (≤4% dc) over planar substrates, i.e., glass and silicon.[19] However, no studies have been conducted for higher Al doping levels (>4% dc contribution) and using non-conventional, complex 3D ceramic microstructures.In this study, three different recipes are used to obtain films with higher aluminum concentrations (>4% dc contribution), accomplished by varying the TMA pulses as indicated in the scheme (Figure S2 and Table S1).The AZO deposition methodology is also applied over 3D ZrO 2 microstructures.The structures have been annealed at 600 and 1200 • C to obtain zirconia in two crystallographic phases, i.e., tetragonal (t-ZrO 2 ) and monoclinic (m-ZrO 2 ).[34]

Microstructural and elemental characterization
For ceramic 3D micro-optics, the film morphology, surface roughness, and conformality are important, as such characteristics can affect either optical or electrical outcomes.The study starts with the AZO coating of ZrO 2 microstructures.For reference, top-view images of the 3D t-and m-ZrO 2 gyroid microstructures before and after the AZO 5% deposition are presented in Fig. 2. The high-magnification micrographs (Fig. 2a' and c') reveal the granular, irregular surface of the pristine ZrO 2 structure.In the vertical direction, the individual, stepped layers stacked over each other can be differentiated, typical for the additivelymanufactured structures (Fig. 2a" and c").Fig. 2b and d show corresponding images of the same microstructures after the AZO 5% deposition.The visual qualitative analysis of the images confirms high surface coverage and conformal deposition of the thin films.The high magnifications of the labeled regions (Fig. 2b', b", d' and d") show that the films grow following the original topography, where the lateral dimensions of the 3D microarchitectures are increased (Figure S3).As a result of film growth, distances between microarchitecture beams are reduced (Figure S3).The analysis implies that chemisorption reactions and nucleation processes proceed on surfaces of 3D-printed ZrO 2 microstructures.[38] The charging of the microstructures by the electron beam of the SEM instrument is visibly reduced upon the deposition of AZO 5%, indirectly proving the electrical conductivity of the film.The AZO 5% crystals are observed to grow as needle-like grains, which agrees with previous reports.[19] SEM image of AZO 5% crystals deposited over silicon substrate is shown.The silicon substrate has been chosen for SEM contrast purposes.Similar growth behavior is expected over t-ZrO 2 and m-ZrO 2 microstructures but with a lower crystallization which is challenging to revolve with SEM (Fig. 2).Regardless of the t-ZrO 2 and m-ZrO 2 microstructure geometry (octet-truss lattices, gyroid, or buckyballs) with AZO 4%, AZO 4-5%, and AZO 5%, lower crystallization is observed.Octet-truss lattices coated with ZnO, AZO 4%, AZO 4-5% and AZO 5% are shown in Figures S4-S7.Gyroid coated with ZnO, AZO 4%, AZO 4-5% and AZO 5% are shown in Figures S8-S11.Buckyball coated with ZnO, AZO 4%, AZO 4-5% and AZO 5% are shown in Figures S12-S15.
A planar surface of the Si/SiO 2 substrate is presented next to the area where 3D ZrO 2 microstructures are located (Figures S16 and S17).The surface is covered with an oxide layer grown during the thermal annealing of the pre-ceramic structures.[34,39] It should be noted that needle-like crystals have been challenging to resolve due to SEM contrast.The SEM images collected for the AZO 4%, AZO 4-5%, and AZO 5% demonstrated similar crystallite morphologies to a previous report.[19] The sizes of the grains are estimated by analyzing SEM images of crystallites deposited on the Si/SiO 2 substrates in the areas neighboring the 3D ZrO 2 microstructures.Forty crystallite length (l) and width (w) measurements are conducted randomly for all four thin film compositions.The statistical analysis is described in Section S5.The respective SEM images are shown in Figure S18.The results are given in Table 1.The aspect ratio (AR, i.e., l to w) of crystallites for all the AZO films is comparable but more consistent over the Si/SiO 2 substrate for t-ZrO 2 .More consistent crystallite sizes are observed over the substrate on which t-ZrO 2 microstructures are located, which could be related to lower t-ZrO 2 surface roughness than the m-ZrO 2 .The observation is likely related to the higher annealing temperature of the latter one (1200 • C versus 600 • C).For ZnO, more round crystallites are observed with a lower significance of the substrate role.
The ZnO films grown with different Al doping cycle contributions (AZO 4%, AZO 4-5%, and AZO 5%) and pure ZnO deposited on the t-ZrO 2 octet-truss lattices are visually inspected at three magnifications (Fig. 3) to evaluate any potential variation of the printed structure configuration during AZO deposition.The different microstructure types are shown in Figures S4-S15.For the Al-doped ZnO in Fig. 3, high aspect ratio needle-like grains are noticed at the highest magnifications.In contrast, the ZnO film appears to form more uniform, possibly hexagonal habit grains, with no additional level of morphology observable.To compare the size of crystallites grown over the planar Si/SiO 2 substrates and 3D ZrO 2 structures, the width, length, and aspect ratio are determined for images presenting the flat round t-ZrO 2 base for all AZO film compositions.The length, width, and aspect ratio for AZO 4% are 96 ± 12 nm, 22 ± 3 nm, and 4 ± 1 nm, respectively.For AZO 4-5%, 93 ± nm, 21 ± 3 nm, and 5 ± 1 values are determined.The parameters are ± 12 nm, 20 ± 2 nm, and 5 ± 1 in the case of AZO 5%.The ZnO grains are not distinguishable from the t-ZrO 2 crystallites.For m-ZrO 2, only AZO 5% crystallite dimensions can be accurately measured with 81 ± nm length, 20 ± 3 nm width, and 4 ± 1 ratio.
Although AZO has been successfully deposited over complex 3D microstructures, an important point should be confirmed, which is conformality.For such an assessment, we use a representative microstructure.In this case, the m-ZrO 2 octet-truss lattice with AZO 4-5% on the surface of the 3D ZrO 2 microstructure is milled by a focused ion beam (FIB) to reveal the structure's core.SEM-EDX maps showing the elemental distribution (Zr, Zn, O, Al, C, Si) are collected (Fig. 4 (ag)).A thin Zn, Al, and O film is observed over solid beams containing Zr and O (Fig. 4 (be)).The contrast in the image corresponding to Al (Fig. 4 (e)) is adjusted to increase the initially low readability due to the low Al content.C signals are assigned to both the printed microstructure and the grown thin film (Fig. 4 (f)), which origin can be the incomplete decomposition of precursors used for the 3D printing and ALD or the contamination from the chamber and FIB processing.Carbon is often an impurity in ceramic structures printed using TPL and tailor-made resins.[34,40] The Si substrate is revealed under the base octet-truss lattice (Fig. 4 (g)), which surface is oxidized due to the annealing step used in the additive manufacturing of the ZrO 2 architectures.[34] Finally, the Zr and Zn distribution maps with 75% opacity overlapped in postprocessing (Fig. 4 (h)).The elemental distribution matches the expected allocation of elements for the AZO grown over ZrO 2 .No specific hot spots, e.g., Al or Zn domains, undesired aluminum oxide islands, or areas differing from the desired film morphology grown over a 3D structure, are found within the spatial resolution of the map.It is concluded that the analyzed AZO film is continuous and conformal, also within the central areas of the specimen, which may be difficult to reach for other methods of surface deposition or materials composed of porous nanoparticles.[15,16] An important parameter not yet mentioned is the chemical content of the multiple specimens encountered over the microstructures after AZO deposition.Therefore, the EDX spectrum is collected from the outer beam of a buckyball-inspired architecture coated with AZO 4% to assess the chemical composition of the thin films.EDX spectrum (Figure S19) reveals the chemical composition 62 at.%O, 13 at.%Zn, 1 at.%Al, 10 at.% Zr, and 14 at.%Si.After excluding Zr and Si, both oxidized and thus compromising the oxygen percentage, the calculated aluminum content in AZO film is approximately 4 ± 0.1 at.%.This approximation accuracy is limited due to high uncertainties in measurements of light elements, i. e., oxygen using EDX, though it confirms the Al doping at the expected level.[40] In addition, confocal Raman spectroscopy indirectly confirmed the deposition of AZO 4% over 3D t-ZrO 2 and m-ZrO 2 microstructures (Figure S1).
To estimate the GPC values for the films deposited over 3D ZrO 2 structures, the feature size increment is determined by analyzing the SEM images.An example of such measurement is given (Figure S3).The images are collected before and after deposition for each film composition for t-ZrO 2 and m-ZrO 2 microstructures (Figures S4 -S15).Then, dividing by the known number of cycles, GPC values are computed (Fig. 5).The standard deviation for the thickness increase is the square root of the squared standard deviations of populations.Due to different measurement methods, this contributes to a higher error than for planar substrates.The planar surfaces of glass and silicon are typically chemically pre-cleaned to remove organic contamination and hydroxylate species from the outer layer.Even though the 3D ZrO 2 structures are not cleaned before the deposition and are used directly after storage, all architectures have reactive surfaces with similar GPC values to planar substrates.The GPC values are comparable between the t-ZrO 2 and m-ZrO 2 phases.In most cases, the GPC value is slightly higher for the m-ZrO 2 structures, which may be correlated with higher porosity and granularity of t-ZrO 2 , and in turn, higher active surface area.
Concerning film conformality, the buckyball can be considered an open architecture.In contrast, the octet-truss lattice consists of an intricate network of beams that may interact differently with the precursor vapor during the deposition.It can be observed that despite the architecture, all of the microstructures show a similar trend in GPC values when the Al content is varied.The holes or cavities within the presented architectures (≈ 1 µm 2.5 µm long) and their polycrystalline porous nature do not impede the precursor flow or cause uneven nucleation.[38,41] The qualitative analysis demonstrates a conformal deposition at the top-view surface growth.Neither the irregular surface of the structures nor the external network of beams seemed to divert AZO nucleation in a different ratio.The reaction seems mainly influenced by the high DEZ, TMA, and H 2 O reactivity.

Chemical state characterization
To provide insights into the chemical environment, potential defects, and impurities of the deposited films, X-ray photoelectron spectroscopy (XPS) analysis is conducted.Since the spot sizes of most of the conventional XPS instruments are relatively larger than the 3D structures presented in this study, planar areas of the samples are analyzed instead.[42] The core spectra of the Al 2p, Zn 2p, and O 1s core levels are recorded for ZnO, AZO 4%, AZO 4-5%, and AZO 5% samples deposited over Si substrates (Fig. 6).Within the Al 2p region, a narrow single doublet with 0.55 eV spin-orbit splitting is observed for all AZO samples (Fig. 6 (a)).For simplicity, the 2p 3/2 and 2p 1/2 levels are not individually plotted.The peak area increases with the Al content and is 1.23, 1.36, and 1.52 square units for AZO 4%, AZO 4-5%, and AZO 5%, respectively.This trend is expected according to the increasing Al saturation in the monolayer within the recipes.In the Zn 2p 3/2 range, two main contributions are found for all the analyzed samples (Fig. 6 (c)).The lower binding energy (BE) peak (Zn I ) corresponds with the Zn lattice within the ZnO wurtzite phase (Fig. 6 (c)).[43] The peak at the higher BE (Zn II ) has been related to zinc point defects such as Zn i or zinc hydroxide (Fig. 6 (c)).[44,45] Two contributions are found in the O 1s spectra (Fig. 6).The more intense peak (O I ) is associated with the ZnO wurtzite crystalline structure lattice oxygen (Fig. 6 (b)).[43][44][45] The second peak (O II ) may be related to low-coordinated ZnO, [46] which forms hydroxylated species upon exposure to the ambient.V O species might also be present but cannot be resolved due to their high reactivity to moisture.[47] Nevertheless, the presence of V O is confirmed optically in the next section.
The Al 2p 3/2 signal is observed at BE around 74.1 eV in AZO 4%, AZO 4-5%, and AZO 5% (Fig. 6 (a)).In all cases, and since only one contribution is found, the peak locations can signify that Al constitutes a substitutional dopant in the wurtzite, likely as Al + Zn point defects.[43,48] Al 2p position shifts for varying concentrations are generally insignificant because no local chemical changes are introduced for the Al specie.The Zn 2p 3/2 photoemission line highly depends on the local chemical environment, i.e., local structure changes, morphology variation, and oxygen-deficient lattices.[49,50] A significant difference in Zn I and Zn II peak positions is observed for all AZO compositions compared to ZnO (Fig. 6 (c)).The BE of AZO 4-5% is also lower than the AZO 4% and 5%.

The decrease in BE upon doping ZnO with Al may indicate the inclusion of Al +
Zn defects in the crystalline structure as the dopant increases the free electron density.[19] Similarly, the O I and O II peak positions are consistently shifted to lower BE for all the Al-doped films.In addition to the free electron density augmentation, the oxygen components are susceptible to Al dopant, which suggests the observations are related to the different local structures in AZO films.

Optical properties
The role of defects present in AZO is an important factor.Defects can modify the optical properties of the aimed optical ceramic material.Previously, we have characterized the optical properties of t-ZrO 2 and m-ZrO 2 microstructures using cathodoluminescence (CL) [34] and similar Fig. 5. Growth per cycle (GPC) analysis for deposition of ZnO, AZO 4%, AZO 4-5%, and AZO 5% over (a) t-ZrO 2 and (b) m-ZrO 2 buckyballs, gyroids, and octettruss lattices.results have been found in Figure S20.In this study, we further investigate the influence of the composition of the AZO thin films deposited over the Si substrates and 3D m-ZrO 2 microstructures using the same technique.The selection of 3D m-ZrO 2 microstructures with deposited thin films is made since, after the deconvolution of CL spectra, considerably fewer components are found for this phase when compared with t-ZrO 2 .[34] This way, fewer defect-related ZnO, AZO, and m-ZrO 2 emissions might overlap.The CL spectra are recorded for spots on the m-ZrO 2 buckyballs and directly next to them on the substrate to facilitate the interpretation.The surface of the Si substrate is covered with a native oxide grown during the thermal annealing in the air, reaching several hundred nm, thus exceeding the thickness of the thin films grown over it in this study.[34,39] According to the literature, for ZnO nanorods, the penetration depth depends on incident electron accelerating voltage (e.g., 0.4 µm for 10 kV and 1.14 µm for 20 kV).[51] Therefore, for the 15 kV beam used in this study, the substrate or 3D ZrO 2 microstructures over which films are grown are also sources of the analytical signals detected.
The CL spectrum collected for the undoped ZnO film deposited over substrate features several contributions (Fig. 7 (a)), with the most intense signals centered around 3.60 eV (344 nm), 3.23 eV (383 nm), 2.96 eV (419 nm), 2.62 eV (473 nm), a weaker red band at 1.95 eV (636 nm) and a peak at 1.62 eV (765 nm).The dominant component at around 3.23 eV (≈ 383 nm) is assigned to the ZnO near band emission (NBE), stemming from the free exciton recombination occurring after exciton collisions.[52,53] Previously, peaks at 3.27 eV and 3.65 eV have been distinguished for thick and thin ZnO rods.[54] The latter signals are due to the increased free-bound exciton energy combined with the confinement effect in the thin structure.[54] In SEM images of the films at high magnifications, needle-like features could be resolved (Figure S16 -S17), which may imply the existence of two NBE components.Due to oxygen and zinc vacancies (V Zn ), blue (2.96 -2.62 eV shoulder) excitonic emission can be observed.[55,56] This signal is assigned to the electron transitions from the V O to the valence band and from the conduction band to the V Zn .[55] The near-infrared (near-IR) peak at 1.62 eV indicates the presence of structural imperfections, V Zn, and interstitial oxygen (O i ).[57,58] The peak at 1.65 eV was previously attributed to V O below the conduction band in ZnO nanotubes.[59] The origin of the peak at approximately 1.9 eV is debatable and may relate to the V Zn point defects or a non-bridging oxygen hole center in the underlying SiO 2 formed during the annealing of the Si substrate.[34,57,60] Since the luminescence spectra are typically not presented in the UV but mostly visible range, the accurate attribution and referencing of the 3.94 eV (315 nm) peak are nontrivial.The signal might be related to Fermi-level electrons recombining with valence band holes.[61] A shoulder ranging to higher energy (≈ 290 nm) was also observed for ZnO nanorods.[62] For AZO 4% (Fig. 7 (b)), the main peak maximum is shifted to 3.35 eV (367 nm).The optical bandgap depends on the electron-hole mobility within the semiconductor and is reported to widen from 3.28 eV for undoped ZnO to 3.35 eV for AZO 4%, consistent with our observations.[63] No new distinct signals are found.In the case of the AZO 4-5% on Si/SiO 2 (Fig. 7 (c)), the local maxima are found at 3.94 eV (315 nm), 3.6 eV (344 nm), 3.35 eV (372 nm), 2.96 eV (419 nm), 2.62 eV (473 nm), 1.95 eV (636 nm), and 1.65 eV (752 nm).For AZO 5% on Si/SiO 2 (Fig. 7  (d)), the maxima are found around 3.65 eV (340 nm), 3.35 eV (372 nm), 2.96 eV (419 nm), 2.62 eV (473 nm), 1.82 eV (681 nm) and 1.65 eV (742 nm).An additional contribution at 2.20 eV (564 nm) can be distinguished, likely stemming from the donor-acceptor shallow-level transitions from V O to V Zn .[55] The shoulder at around 3.65 eV (≈ 340 nm) is more intense in all AZO samples when compared with ZnO, correlating with the higher crystallite aspect in Table 1.[54] For AZO 5%, the broadening towards higher energy is observed with a weak peat at 4.22 eV (294 nm).Such broadening of the emission into the UV range was also previously observed with increased Mg concentration in doped ZnO films grown by ALD.[64]  Within the indigo-blue-green region (2.9 -2.3 eV), contributions from underlying 3D m-ZrO 2 structures, ZnO and AZO, can be superpositioned.The typical CL spectrum collected for the additively manufactured m-ZrO 2 microstructures is deconvoluted into the components around 2.3, 2.6, and 2.9 eV.The dominant peak at 2.6 eV can be assigned to V O or the electronic transitions between the F *+ and F + levels.[34,65] The 2.3 and 2.9 eV contributions are associated with interstitial carbon (C i ), likely in ceramics obtained from the metalorganic precursor and carbon-rich polymer.[34] Consequently, due to the overlap of the main ZnO and m-ZrO 2 contributions, in the case of thin films grown over the 3D ZrO 2 structures, the main focus is dedicated to the most intense ZnO signals correlated with the NBE.For ZnO deposited over m-ZrO 2 buckyball, the CL spectrum is deconvoluted into the components at 2.28, 2.56, 2.9, and 3.18 eV (Fig. 8 (a)).The latter, the most intense signal, can be assigned to ZnO NBE.[52] A weak near-IR signal around 1.62 eV is found, previously attributed to structural imperfections, V Zn, and O i .[57,58] Besides the broader band deconvoluted to peaks at 2.3 eV, 2.6 eV, and 2.9 eV, for AZO 4%, AZO 4-5%, and AZO 5% grown over 3D m-ZrO 2 structures, the NBE peak is found at 3.34, 3.30, and 3.34 eV (Fig. 8 (ad)).With the increase of Al content in AZO, a minor widening of the band gap can be expected.[19] The intensity of the NBE peak is significantly lower than for the ZnO deposited over m-ZrO 2 and the ZnO and AZO films grown over the substrate.These differences indicate that the radiative emission recombination paths may be modified.[66] As in the CL, the carriers are excited with high-energy electron beams, and the abundant electrons could be trapped in unfilled defect levels related to V O .[66] In turn, the electrons are unavailable for transitions associated with other CL bands.[66] In addition, complimentary transmittance spectra of the thin films deposited on glass slides are recorded in the ultraviolet-visible-nearinfrared range (UV-Vis-NIR) (Figure S21(a)).Since only minor differences have been observed between ZnO and AZO (with 1 -4% dc contributions), in this study, AZO 4% and AZO 5% are representative samples.[19] There is no significant difference in the UV range concerning the absorption edge between the samples.The increased Al doping results in slightly higher transmittance over the visible range, correlated with the higher charge carrier concentration and Burstein-Moss effect.[8,19] Using the Tauc plot (Figure S21(b)), an optical band gap of 3.5 ± 0.1 and 3.6 ± 0.1 eV for AZO 4% and AZO 5%, [67,68] respectively, are determined, which are higher than the NBE positions determined by CL.
Monochromatic cathodoluminescence images of the m-ZrO 2 gyroid with deposited AZO 4% are collected (Fig. 9) to study the spatial signal distribution at 2.58, 2.88, and 3.30 eV over the conformally deposited film.Interestingly, no increase in CL intensity is noted for the edges of lamellas composing gyroid, which could be expected due to the higher specific free surface area.For 2.58 and 2.88 eV, higher intensity is noted within inner regions of the 3D structures between parallel segments of lamellas, which may occur due to the increased electric field amplitude inside a cavity.[69] The intensity of the signals at 3.30 eV, associated with NBE, is lower but homogenous throughout the architecture, illustrating the uniformity of the AZO layer.It is important to note that the shape of the 3D microstructures does not influence the optical properties of the m-ZrO 2 octet-truss lattice, gyroid, and buckyball coated with AZO 5% (Figure S20).The only observable difference concerns the minor   difference in the relative intensity of the 2.3, 2.6, and 2.9 eV peaks.

Electronic properties 2.6.1. Hall effect measurements
The AZO electrical properties (carrier concentration, mobility, and resistivity) are determined via Hall effect measurements.Upon the placement of an electrical conductor carrying electric current in the transverse magnetic field, a potential difference is produced, termed the Hall effect.[70,71] The principle is often employed to determine the material electronic properties.[71] To establish the Hall effect, measurements are conducted using a dedicated system for depositing thin films on glass substrates by applying a magnetic field of 0.5 T. The derived results are shown in Table 2.
All AZO films exhibit n-type semiconductor behavior, as can be expected for Al-doped ZnO.[19] As a reference, the bare ZnO film showed a mean resistivity (ρ) value of 6.02 × 10 -3 Ω cm, following the lowest values reported.[72] The n-type extrinsic semiconductor resistivity (ρ) is inversely proportional to the elemental electric charge (e), the free electron density (n), and the electron mobility (µ n ), as given in Equation (1).
Thus, the charge carrier concentration and electron mobility play an important role in resistivity evaluation as intrinsic contributions.Regarding the fabricated AZO-ZrO 2 heterostructures, the electrical resistivity shows values reaching a minimum of 1.1 × 10 -3 Ω cm, which is an improvement from previous results using the supercycle (sc) approach by APLD.[19] Compared with ZnO and AZO 4%, the carrier concentration shows an increment for the AZO 4-5% and 5% films with values near 7.7 × 10 20 cm − 3 .[19] The donor doping effect may explain the latter, in which Al atoms substituting Zn in the ZnO crystalline structure (as discussed in XPS) impact the electrical properties.For AZO 4%, 4-5%, and 5%, the mobility is reduced compared with AZO of lower Al content and reaches the minimum value of 5.79 cm 2 V − 1 s − 1 ) for AZO 4-5%.The mobility behavior is mainly influenced by the charge carrier concentration, which can be detrimental to electron mobility if scattered by ionized impurities.[73] In this study, no clear plateau of electrical properties is observed within the range of conditions, and possibly further modification can be achieved for doping cycle contributions exceeding 5%.The values obtained are comparable with those reported in the literature for planar substrates.[21,22,[74][75][76]

Concluding remarks
Conformal deposition of ZnO and AZO over complex 3D ZrO 2 microstructures has been achieved via atomic partial layer deposition using diethylzinc, trimethylaluminum, and water as reagents.Thin films of AZO at 4%, 4-5%, and 5 % Al doping cycle contributions have been conformally deposited.The crystallographic phase of 3D ZrO 2 microstructures (t-ZrO 2 and m-ZrO 2 ) has not affected film growth.No specific cleaning of the 3D-printed ZrO 2 structures is required, and structures can be applied as-annealed (1 h at 600 or 1200 • C) after storage.Qualitative chemical analysis confirmed homogenous elemental distributions throughout the films grown over the 3D ZrO 2 microstructures.No domains, undesired aluminum oxide islands, or areas differing from the desired film morphology grown over the 3D structure were found.The GPC is not significantly affected by the shape of 3D architectures.
Furthermore, the CL study confirms the emissions from the ZnO and AZO films associated with V O , V Zn , and O i .Interestingly, for thin films on planar substrates, two components of the NBE can be distinguished, attributed to the high aspect ratio of crystallites, confirmed by SEM.In CL spectra for ZrO 2 , we previously found components associated with V O and C i , which may overlap with defect-related emissions in thin films grown in this study, especially under the beam acceleration conditions (15 kV) required to analyze the 3D ZrO 2 structures.Only one NBE component can be distinguished for ZnO and AZO grown over ZrO 2 3D structures.The NBE peak position changes with the Al content and is respectively 3.18 eV, 3.30 eV, 3.30 eV, and 3.34 eV for ZnO, AZO 4%, AZO 4-5%, and AZO 5%.
All AZO films exhibit n-type semiconductor behavior, as expected for Al-doped ZnO.For AZO 4%, 4-5%, and 5%, the mobility is lower than for AZO of Al < 4% and reaches the minimum of 5.79 cm 2 V − 1 s − 1 for AZO 4-5%.[19] The electrical resistivity of AZO 5% reaches a minimum of 1.2 × 10 -3 Ω cm, an improvement from previous results using the supercycle (sc) approach by APLD.[19] Compared with ZnO and AZO 4%, the carrier concentration increments for the AZO 4-5% and 5% films with values near 7.7 × 10 20 cm 3 .The electrical parameters are not plateaued, and tuning might be possible by increasing the Al doping cycle contribution.The effective impurification methodology (APLD) for Al as a substitutional dopant in the wurtzite has resulted in defects leading to improved charge carrier concentration and reduced resistivity.The values obtained are comparable with the literature for planar substrates.
The study concludes that the deposition of thin conductive oxides using the APLD is feasible.The characteristics of the ZnO and AZO films deposited over 3D ZrO 2 microstructures resemble their typical optical and electrical properties.Supercycle methodology enables the development of an auspicious reproducible approach, addressing intrinsic and persistent issues emerging when doping is done via conventional ALD, showcased even for complex 3D microstructures.Further studies on APLD could result in the foundation of a reproducible approach permitting the manufacturing of functional thin transparent electrode materials for optoelectronics, integrated optics, or light-emitting diodes.[34,77,78] The ability to deposit TCOs over 3D-printed microstructured ceramics is a step towards achieving dimensionally refined electroluminescent diodes and other devices, utilizing lanthanide-doped phosphors, e.g., ZrO 2 :Eu 3+ , as emitters.[34,79] The advantages of such integration include improved heat dissipation and higher phosphor resolution, combined with reduced optical cross-talk and scattering through structural design and precise positioning.[34,80]

Materials
Diethylzinc and trimethylaluminum, of microelectronics grade, used as ALD precursors, were purchased from Sigma-Aldrich.The ACS reagent grade acetone, ethanol, isopropanol, sulphuric acid, and hydrogen peroxide were acquired from Fermont.Industrial-grade N 2 was purchased from Grupo INFRA MX.Deionized water was obtained from the ultrapure filtration system.The photoresin was prepared using the recipe established before.[34]

Additive manufacturing of ZrO 2 microstructures
The ZrO 2 microarchitectures were additively manufactured using a Zr-rich tailor-made photoresin.The custom recipe and methodology

Atomic layer deposition (ALD) of AZO and ZnO films
The depositions of AZO and ZnO were conducted in a custom-built TFS-200 Atomic Layer Deposition reactor (Beneq Oy), operated in thermal mode, following the recipe presented in the previous work.[19] In total, four different film compositions were studied: ZnO, ZnO:Al-4.0%(AZO 4%), ZnO:Al-4-5% (AZO 4-5%), and ZnO:Al-5% (AZO 5%), where Al % refers to the percentage of cycles involving the Al-rich precursor, trimethylaluminum.The films were deposited over 3Dprinted t-ZrO 2 and m-ZrO 2 structures and two types of control samples, i.e., Fisherfinest™ Premium plain glass microscope slides (Fisher Scientific) and commercial-grade silicon substrates.Before deposition, the Si substrates were sequentially sonicated for 10 min in baths of acetone, ethanol, isopropyl alcohol, and piranha solution (H 2 SO 4 -H 2 O 2 4:1 v/v) to be finally dried under an N 2 stream.The flow rate of ultrapure N 2 was 400 sccm for the process and 300 sccm for the vessel; a viscous flow regime was followed.The reactor temperature was set at 200 • C. The 3D-printed ZrO 2 microstructures and the control samples were placed in the process chamber 10 min before the deposition to equilibrate the temperature.
As a baseline, the ZnO deposition recipe consists of a fixed number of cycles.A typical individual cycle (tc) includes a diethylzinc (DEZ) dose of 0.240 s, a purge of 2.0 s, an H 2 O dose of 0.240 s, and a 2.0 s purge.The ALD regime for high-quality films (≈ 100 nm growth for 625 cycles) is maintained, as schematically depicted in (Figure S2).A supercycle recipe was followed for the deposition of AZO at all studied dopant concentrations (4.0%, 4-5%, and 5.0 Al doping cycle %).In such a recipe, a tc is replaced with a doping cycle (dc), corresponding to the desired dopant percentage.The dc includes an extra trimethylaluminum (TMA) pulse and proceeds as follows: i. DEZ dose, ii.purge, iii.TMA dose, iv.purge, v⋅H 2 O dose, vi.purge, as depicted in Figure S2.For instance, the ZnO:Al-4% (AZO 4%) sc contains 24 tcs, followed by a single dc.After 25 repetitions (625 cycles in total), the dopant agent constitutes 4% of the dc 25 cycles, as presented in Fig. 1 and Figure S2.Table S1 shows the specific conditions for each recipe for AZO 4%, AZO 4-5%, AZO 5%, and ZnO (without Al doping).A complete deposition was driven nonstop to obtain ≈ 100 nm thickness.

Field emission scanning electron microscopy (SEM)
The electron micrographs of the deposited films and the 3D ZrO 2 structures were obtained using a field emission scanning electron microscope (SEM, FEI Nova Nano SEM 200) equipped with an immersion lens secondary electron detector.An acceleration voltage of 15 kV was applied.

Crystallite size determination -Image analysis
The high-magnification SEM images of the ZnO, AZO 4%, AZO 4-5%, and AZO 5% films deposited over Si substrates earlier annealed at 600 and 1200 • C were analyzed using the Fiji image processing package (2.9.0 release) to determine the average crystallite sizes.Forty spots were analyzed in each case.

Scanning electron microscopy (SEM)
The images of the m-ZrO 2 microstructures after the deposition of AZO 4-5% were collected at 15 kV acceleration voltage using a JSM-IT500 InTouchScope™ scanning electron microscope.

Energy dispersive X-ray spectroscopy (EDX)
Chemical characterization was conducted using an energy dispersive X-ray microanalysis system (Oxford, INCA X-Sight) attached to the SEM.The data was collected using INCA Energy software.

Spectral reflectance analysis
The thickness of the deposited layers was determined in spectral reflectance measurements conducted on the glass and Si control samples.A commercial thin-film thickness measurement system (Filmetrics F20-UV, KLA Corporation) with a source operating in a 200 nm − 1100 nm spectral range was used.

X-ray photoelectron spectroscopy (XPS)
The high-resolution XPS of the deposited films were registered using a Thermo Scientific™ ESCALAB™ 250Xi XPS system.A monochromatic Al K α (hν = 1486.7 eV) XR6 X-ray source and the 180 • hemispherical energy analyzer were used.The photoelectrons were measured at the perpendicular (θT = 90 • ) orientation of the magnetic field to the sample surface.The spectra were acquired from a 650 μm spot at 10 eV pass energy, with compensation for potential charging effects using an electron flood gun.The position of the adventitious C 1 s peak (284.8 eV) was monitored, and specimens were scrupulously sputtered using an Ar gun, operated at 2000 eV and 10 μA until the disappearance of the surface C peak, thus resulting in a clean sampling area.The peaks were fitted with the AAnalyzer software version 1.20.[81] The background was subtracted using the active background approach function.Binding energy was normalized by individually deconvoluting the C 1s region and adjusting the BE of carbon sp 2 peak position to 284.1 eV.The spectra were then deconvoluted using Chi-square fitting.

Hall effect measurements
The electrical properties of the thin films were characterized via Hall Effect measurements using the control samples.The DX-50 (Dexing Magnet Tech.)The Hall effect system with a magnetic field of 0.5 T was used.The carrier concentration (n), resistivity (ρ), and mobility (µ n ) were obtained at room temperature by measuring the Hall voltage using the van der Pauw method to study the electrical properties of the deposited AZO.

Cathodoluminescence (CL)
The cathodoluminescence study was conducted with a Gatan Mon-oCl4 detector attached to a JEOL JIB-4500 SEM.During the measurements, a 15 kV acceleration voltage was used.

Ultraviolet-visible-near-infrared transmittance measurements (UV-Vis-NIR)
The UV-Vis-NIR transmittance spectra of the thin films deposited on glass were recorded using a Cary 5000 (Agilent Technologies) spectrophotometer in the 190 -1100 nm range.Pristine glass substrates were used for reference.

Confocal Raman spectroscopy (Raman)
The Confocal Raman Spectroscopy measurements were conducted at ambient conditions using an Alpha 300 (WiTec) with a 100x/0.9NAair objective (MPlan FL N, Olympus) and a 600 g/mm grating.The samples were excited with a 532 nm laser at 5.0 mW power.The presented data are an average of 60 acquisitions of 1 s.CrystalSleuth software

Fig. 1 .
Fig. 1.Schematic depiction of the proposed sequence of AZO growth by the APLD mode: (a) a substrate with terminal -OH groups, (b) DEZ non-saturating pulse, (c) TMA saturating TMA pulse, (d) H 2 O oxidation pulse.

Fig. 4 .
Fig. 4. Elemental mapping of the octet-truss lattice cross-section prepared via FIB milling: (a) secondary electron image of the inspected area, and (bh) EDX maps showing the distribution of (b) Zr, (c) Zn, (d) O, (e) Al, (f) C, (g) Si, and an overlayed image (h) of (b) and (c) at 75% opacity.The scale bar presented in (a) is shared across the (bh) images.

Fig. 9 .
Fig. 9. SEM and monochromatic cathodoluminescence images of the 3D m-ZrO 2 structure coated with AZO 4%: (a) SEM image of a gyroid, and CL image at (b) 2.58 eV, (c) 2.88 eV, and (d) 3.30 eV; (a' -d') close-up images of the areas labeled with boxes in images (ad).