1. Introduction
The global demand for sustainable and clean energy has significantly accelerated the growth of the solar energy sector, making it one of the most rapidly expanding industries of the 21st century. This expansion mirrors the early development trajectory of the microelectronics industry, with continuous innovation driving down production costs and improving efficiency. The performance of photovoltaic (PV) devices, particularly silicon-based solar cells, can be substantially enhanced through the optimization of cell architecture, surface passivation techniques, and the integration of advanced functional materials. A critical component in improving solar cell efficiency is the application of transparent conductive layers (TCLs), which serve a dual role: allowing light to pass through to the active layer while simultaneously acting as an efficient electrical contact [
1,
2,
3]. These layers are typically composed of transparent conductive oxides (TCOs)—materials that uniquely combine optical transparency in the visible spectrum with adequate electrical conductivity. TCOs have become indispensable in modern optoelectronic applications, including displays, LEDs, touch panels, and especially photovoltaic devices. The functionality of TCOs depends largely on the choice of base material, its doping strategy, and the method of deposition [
1,
2].
While tin-doped indium oxide (ITO) remains the benchmark due to its excellent electrical and optical performance, its high cost and the scarcity of indium have driven the search for alternative materials [
2,
3]. Fluorine-doped tin oxide (FTO) has emerged as a chemically robust and cost-effective substitute, though it suffers from lower carrier mobility compared to ITO. Among the most promising candidates for next-generation TCOs is aluminum-doped zinc oxide (AZO) [
3,
4,
5]. This material stands out due to the abundance and non-toxicity of its constituent elements, combined with desirable optoelectronic properties that rival those of ITO. In particular, The AZO exhibits a wide band gap and favorable n-type conductivity, which can be further tuned by controlling aluminum doping concentrations. Zinc oxide (ZnO) is a naturally n-type semiconductor with a wide bandgap (~3.3 eV), making it inherently transparent in the visible spectrum. However, its native conductivity is typically inadequate for TCO applications [
5,
6,
7]. Doping with trivalent elements such as aluminum significantly enhances the carrier concentration, allowing ZnO:Al thin films to reach electrical and optical properties comparable to those of ITO at a fraction of the cost. The ZnO:Al thin films typically crystallize in the hexagonal wurtzite structure, and their microstructure, conductivity, and transparency strongly depend on the deposition technique and process parameters (e.g., temperature, pressure, precursor chemistry, doping concentration) [
7,
8,
9,
10]. Below in
Table 1 is a comparison of selected literature on ZnO:Al thin-film fabrication and resulting properties.
Sputtering remains the most used industrial technique for TCO fabrication. While it offers low resistivity and high transparency, it often requires post-deposition annealing to enhance film crystallinity and reduce defects. On the other hand, sol–gel methods are attractive for their simplicity and low equipment costs but typically yield higher resistivities due to poor crystallinity and porosity in the films. In turn, spray pyrolysis allows for easy scaling and deposition on large substrates; however, controlling film stoichiometry and doping levels can be challenging, affecting optical quality. A method that offers good results is Pulsed Laser Deposition (PLD), which offers superior control over stoichiometry and produces dense films with excellent properties, but its high cost and low throughput limit industrial applicability. ALD, as demonstrated in this work, provides unmatched uniformity and dopant control at relatively low temperatures. The resulting ZnO:Al thin films show low resistivity and high optical transmittance even at nanometric thicknesses, making them ideal for use in high-efficiency solar cells where layer thickness and surface conformity are critical [
7,
8].
The ability of ALD to coat complex, nanostructured surfaces with precision enables further innovations in advanced photovoltaic architectures, such as passivated emitter and rear contact (PERC) cells or tandem solar cells. The focus of this article is to investigate the structural and electrical characteristics of ZnO:Al thin films and assess their suitability for application in silicon solar cells as antireflection and TCO component [
6,
7,
8,
9,
10].
2. Materials and Methods
The experimental procedure involved the use of substrates in the form of glass slides and monocrystalline silicon wafers, each coated with a thin ZnO:Al thin film deposited via ALD. All substrates were square-shaped with dimensions of 20 × 20 mm
2. Prior to deposition, the substrates underwent a three-stage cleaning process using an ultrasonic bath: first, immersion in a water-detergent solution for 20 min; next, rinsing in acetone for 15 min; and finally, cleaning with isopropyl alcohol for an additional 15 min. Glass substrates were used exclusively for transmittance measurements due to their high optical transparency in the visible range. All structural, morphological, and electrical characterizations were conducted on silicon substrates, which are the intended base material for photovoltaic integration in polycrystalline silicon solar cells. The choice reflects the application-oriented nature of the study, with silicon treated as the target substrate for evaluating ZnO:Al thin-film performance. Photovoltaic structures were fabricated using p-type polycrystalline silicon wafers characterized by the following parameters: approximately 330 µm in thickness, a surface area of 5 × 5 cm
2, boron doping, and a specific resistivity of 1 Ω·cm. The formation of the p–n junctions was achieved through phosphorus diffusion in an open-tube quartz furnace using POCl
3 (phosphorus oxychloride) as the dopant source. All solar cells featured a single busbar configuration. The ZnO:Al thin films were deposited using a Picosun R-200 Standard ALD system (Picosun, Espoo, Finland). The precursors for the deposition process included diethylzinc (DEZ) and trimethylaluminum (TMA), while deionized water was employed as the oxygen source for both ZnO and Al
2O
3 deposition reactions. Precursors were dosed for 0.1 s each, and nitrogen was used both as a carrier and purge gas. Nitrogen flow was maintained at 200 sccm, with purge durations of 4 s after each precursor pulse and 5 s after water exposure. The deposition temperature was set at 225 °C, based on process parameters reported in earlier studies as optimal for achieving high-quality ZnO and Al
2O
3 layers [
16,
17]. The ZnO:Al thin films were obtained through a supercycle process, consisting of alternating sequences of 20 cycles of ZnO followed by one cycle of Al
2O
3, forming a controlled multilayer architecture. The number of these bilayer supercycles was varied to tune the final film thickness and composition. A schematic representation of the multilayer AZO coating structure is provided in
Figure 1, illustrating the periodic sequence of ZnO and Al
2O
3 sublayers that form the final transparent conductive oxide film results. The 20:1 ZnO:Al
2O
3 supercycle ratio used during the ALD process was selected as a compromise between optimal electrical performance and high optical transmittance. This ratio ensures uniform and controllable Al incorporation while preserving the crystallinity and low roughness of ZnO layers. Similar approaches have been described in the literature [
14], where controlled Al doping was shown to enhance conductivity while maintaining transparency in ZnO-based thin films.
The surface morphology of the fabricated AZO films was examined using scanning electron microscopy (SEM), conducted with a Zeiss Supra 35 system. The imaging was performed at an accelerating voltage ranging from 3 to 5 kV. Surface features and topography were visualized using the in-lens secondary electron detector, which is particularly suitable for high-resolution imaging of fine surface structures. In addition to morphological evaluation, the chemical composition of the coatings was assessed qualitatively through energy-dispersive X-ray spectroscopy (EDS), allowing the identification of elemental constituents present within the layer. Detailed structural investigations of the ZnO:Al thin films were carried out using high-resolution transmission electron microscopy (HR-TEM). These measurements were conducted on a FEI Titan 80–300 S/TEM microscope (FEI, Hillsboro, OR, USA), operated at 300 kV. The instrument is equipped with a super-twin lens system and an annular dark-field detector, enabling observations in both conventional TEM and scanning transmission electron microscopy (STEM) modes. For STEM imaging, three detectors were utilized: bright field (BF), annular dark field (ADF), and high-angle annular dark field (HAADF), providing complementary contrast mechanisms. Crystallographic information was obtained via selected area electron diffraction (SAED), allowing the identification of phase structure and lattice ordering within the films. Complementary to electron microscopy, Raman spectroscopy was employed for further structural and phase characterization. A Renishaw inVia Reflex Raman spectrometer (New Mills, UK) equipped with a 514.5 nm green excitation laser was used for this purpose. Spectra were collected across a wide spectral range (50–3100 cm−1), providing insights into vibrational modes associated with crystalline and amorphous regions of the ZnO:Al thin films. This technique enables the identification of distinct phonon modes, offering spectral fingerprints of material phases and verifying the presence of specific structural configurations. Optical properties of the thin films, including their refractive index (n), absorption (A), transmission (T), and thickness (d), were determined via spectroscopic ellipsometry. Measurements were performed using a SENTECH SE850E ellipsometer, operating over the spectral range of 240–2500 nm, with analysis conducted using SpectraRay 3 software. Data acquisition was performed in variable angle mode, with Ψ and Δ parameters recorded at incidence angles between 50° and 70°, in 5° increments. The electrical characteristics of the ZnO:Al thin films were evaluated by measuring sheet resistance using a four-point probe system (Ossila, Sheffield, UK). This method, involving linear probe arrangement with equal spacing, allows for precise resistance measurements by applying current through the outer probes and recording the voltage drop between the inner probes. For the evaluation of photovoltaic performance, current-voltage (I–V) characteristics of polycrystalline silicon solar cells incorporating ZnO:Al were measured. The measurements were carried out using an SS150AAA solar simulator under standard test conditions (irradiance: 1000 W/m2, AM1.5G spectrum, temperature: 25 °C). The I–V characteristics were analyzed using I–V Curve Tracer software version 2.11.3, enabling the extraction of key performance parameters such as open-circuit voltage, short-circuit current, fill factor, and conversion efficiency. Comparative analysis was performed between samples fabricated under varying ALD conditions to assess the influence of deposition parameters on solar cell efficiency. Due to the limited availability of directly comparable literature data for ZnO:Al thin films applied to thick-wafer polycrystalline silicon cells, a direct quantitative comparison was not feasible. Most prior studies focus on thin-film solar technologies (e.g., CIGS, a-Si, perovskites), which involve different device architectures and material requirements. As such, this work fills an important knowledge gap and introduces ALD-grown ZnO:Al thin films as a viable and scalable alternative to traditional TCOs (such as ITO) in industrially relevant poly-Si photovoltaic devices. In conclusion, ALD-deposited ZnO:Al thin films demonstrate excellent structural, optical, and electrical properties suitable for solar cell applications. Their tunability, compatibility with large-area substrates, and conformal coverage make them highly promising for next-generation clean energy technologies. Future work will involve correlating changes in optical constants with structural features determined by XRD and further optimization of ALD parameters (precursor selection and pulse durations, critical for surface saturation and minimizing impurities; and carrier gas flow and purge times, which ensure complete precursor removal and uniform film growth) to enhance performance and extend applicability to other solar technologies. Although ALD is widely recognized for producing highly uniform and conformal thin films, its relatively low deposition rates and capital costs currently limit large-scale industrial adoption for photovoltaic manufacturing. However, emerging approaches such as spatial ALD and high-throughput batch reactors are promising for enhancing scalability. The exceptional film quality achievable with ALD remains particularly attractive for advanced solar cells where performance outweighs cost considerations. Future developments in ALD technology may bridge the gap between laboratory-scale precision and commercial viability.
3. Results and Discussion
SEM analysis revealed that the surface morphology of the obtained coatings exhibited a granular structure (
Figure 2), with elongated grains typical for zinc oxide, which constituted the topmost layer in all multilayer systems. The grains appeared randomly distributed, without any preferential orientation. Furthermore, an increase in the average grain size was noted with the rising number of deposited layers. The grain size distribution based on SEM images using ImageJ software version 1.54p was quantified, providing average grain sizes for selected samples. The grain size increases with the number of ALD cycles, from ~30 nm (420 cycles) to ~70 nm (1050 cycles). These findings are consistent with previous reports, which also describe the formation of elongated ZnO crystallites with a random spatial distribution and size growth correlated with film thickness or multilayer configuration [
18,
19,
20].
The elemental composition analysis confirmed the presence of zinc and oxygen within the deposited coatings (
Figure 3), consistent with the expected stoichiometry of the applied materials. Due to the relatively low thickness of the coatings compared to the penetration depth of characteristic X-rays, signals originating from the silicon substrate were also detected. The peak near 1.01 keV corresponds to zinc, and its intensity was observed to increase with the number of deposited cycles. This enhancement suggests a progressive accumulation of Zn with each additional layer. Simultaneously, the signals associated with the substrate components gradually diminished, indicating effective coverage of the surface. These trends align with findings reported by other researchers, where an increase in Zn signal intensity and attenuation of substrate peaks have been correlated with growing layer thickness and improved surface coverage [
21].
Structural analysis performed using STEM and TEM techniques confirmed the high degree of uniformity and integrity of the ZnO:Al thin films prepared via ALD. The bright-field (BF) image (
Figure 4a) and high-angle annular dark-field (HAADF) image (
Figure 4b) of the cross-sectional lamella illustrate a continuous, homogeneous film, free of microcracks, voids, or other apparent structural defects. Furthermore, the selected area electron diffraction (SAED) pattern (
Figure 5) displays distinct diffraction rings corresponding to the (224), (004), (113), and (022) planes of the spinel ZnAl
2O
4 phase, confirming the crystalline nature of the deposited coating. These findings are in strong agreement with previous studies, which report that ZnAl
2O
4 layers grown by ALD exhibit high crystallinity, excellent structural uniformity, and defect-free morphology [
22].
The Raman spectrum of ZnAl
2O
4 spinel, shown in
Figure 6, reveals characteristic vibrational features typical of the cubic spinel structure (space group Fd-3m). Based on group theory analysis, spinel-type structures are expected to exhibit one A
1g mode, one Eg mode, and three T
2g modes as Raman-active vibrations. In the measured spectrum, several distinct bands can be observed and assigned to specific phonon modes. The T
2g mode around 200 cm
−1 is associated with lattice vibrations involving Zn and Al cations. A band appearing between 400 and 500 cm
−1 corresponds to the Eg mode, representing symmetric displacements of oxygen atoms within both tetrahedral and octahedral sublattices. Furthermore, two T
2g modes located between 500 and 650 cm
−1 are attributed to oxygen vibrations primarily linked to the AlO
6 octahedra. The most intense feature, observed near 750–800 cm
−1, is assigned to the A
1g symmetric stretching vibration of Al–O bonds in the octahedral coordination sites. The obtained Raman spectrum is in good agreement with the literature for ZnAl
2O
4, confirming the formation of a pure spinel phase with clearly defined vibrational modes. The absence of extraneous peaks further indicates the high phase purity of the synthesized material, with no evidence of secondary phases such as ZnO or Al
2O
3.
The presence of ZnAl2O4 spinel phase, observed in SAED and Raman analysis, is attributed to Al diffusion and local chemical rearrangement during ALD growth. However, the dominant matrix of the film is still based on wurtzite ZnO, especially in samples with lower cycle numbers. The coexistence of ZnAl2O4 does not negatively affect the TCO functionality, as evidenced by the low sheet resistance and high optical transparency. In fact, spinel inclusions may improve microstructural stability, which is beneficial in long-term photovoltaic operation.
The spectra of refractive indexes as a functions of wavelength are shown in
Figure 7 The values of n for the wavelength λ = 1 μm are included in the range of 1.64 to 1.74 and are presented in
Table 2. The result obtained for the sample deposited after 210 cycles is clearly different from the others, and its n for the given wavelength is 1.64. The slope of the curve is also smaller compared to the course of other dispersions. The n value obtained for this sample suggests that it has a slightly different physical composition, which is probably related to the percentage ratio of Al
2O
3 to ZnO in the composite. The results were compared with those obtained by Balevicius et al. [
23], where the refractive indices for pure ZnO and Al
2O
3 were presented. The dispersion pattern n for the 210 sample is more similar to the curve for Al
2O
3 presented by the authors. This suggests a high influence of this ingredient in the composite layer. It is also intuitive that the decrease in optical transmittance is 100% consistent with the determined d values. This relationship is shown in
Figure 8.
The highest determined RMS value was recorded for the sample deposited after 1050 cycles. This is due to the presence of numerous granular structures on the sample surface.
This result is also fully consistent with the results of the generated optical absorption spectra (
Figure 9), where the spectrum of the mentioned sample has a higher level of absorption intensity than the other samples, over the entire λ wavelength range. This effect results from the greater scattering of the light beam on the surface of the tested sample. The generated optical absorption spectra have a typical course for this type of composite [
24]. In all spectra, there is one strong electronic band, the maximum of which is located in the UV region. Due to the measurement range of the ellipsometer (240–2500 nm), it was not possible to generate a waveform of the entire band. The area of the lowest absorption for all samples is in the visible range—300–750 nm. As in the work of A.A. Alnajjar [
24], absorption intensity increases significantly in the infrared region. In the abovementioned work [
24], layered structures were examined, where, similarly to our research, successive layers of ZnO and Al
2O
3 were alternately placed on top of each other. The thickness of the layers tested by Alnajjar was 0.04–2 μm. The spectra he presented show numerous oscillations, which are the result of numerous internal reflections of the electromagnetic beam inside quite thick layers. In our case, the oscillations are very mild and rare, which is related to the low value of sample thickness obtained (22–189 nm). For thicker AZO films (e.g., >150 nm), a noticeable increase in infrared absorption was observed, likely due to free-carrier absorption associated with higher electron concentrations. While this improves conductivity, it may reduce light transmission in the near-IR range, where silicon absorbs efficiently. Therefore, optimizing doping levels and maintaining film thickness below ~120 nm can help preserve IR transparency and improve overall solar cell performance.
The optical band gaps were determined by linear fitting to the absorption edge where the function increases most rapidly. The Tauc is revealed by the following equation:
where α is the absorption coefficient, hv = E is the photon energy, C is the constant, and Eg is the bandgap. In the graph of the relationship (αhv)
1/2 as a function of photon energy, a tangent line was drawn to the energy axis, and the Eg value for the absorption coefficient α = 0 was read from it. Typical values of the Al
2O
3 energy gap are in the range of 5.1–7.1 eV [
25], and its values depend mainly on the occurrence of various metastable polymorphic forms of aluminum [
25]. Al
2O
3 is treated as an insulator for various applications in thin-film electronics, e.g., for dielectric gates in MOS transistors. In the case of ZnO, the energy gap value is usually in the range of 3.1–3.3 eV and depends mainly on the morphology of the obtained ZnO layers [
26]. In the case of ZnO:Al thin films, the gap values usually vary in the range of 2.9–3.7 eV and depend mainly on the thickness of the composite layers [
24]. Our results are fully consistent with the literature data. Therefore, the lowest value of the energy gap for the sample deposited after 210 cycles—1.64 eV (
Figure 10)—may be partly due to the low value of the obtained sample thickness (22 nm). The reduced band gap of 2.6 eV observed for the 210-cycle (22 nm) sample may result from increased structural disorder, surface states, and limited crystallinity in the ultrathin film. The low doping level at this stage also results in a reduced Burstein–Moss effect. These factors can collectively lead to an apparent redshift in the optical absorption edge. With increasing film thickness and Al incorporation, the band gap values converge toward expected values (2.9–3.1 eV), consistent with prior reports on AZO thin films.
The data in
Table 2 suggest a clear correlation between surface morphology—specifically roughness—and the optical properties of ZnO:Al thin films. Samples with lower roughness (e.g., 4–8 nm) exhibited higher refractive indices (n ≈ 1.735–1.740 at λ = 1000 nm) and more stable optical band gaps, while rougher films (e.g., 60 nm) displayed greater variability in both parameters. Increased surface roughness can introduce light scattering and inhomogeneities, which in turn affect light absorption and refractive behavior. These findings underscore the importance of surface morphology control in optimizing the optical performance of AZO films for optoelectronic applications.
The correlation between the structural features and optical properties of the ZnO:Al thin films is evident across the samples deposited with varying ALD cycle numbers. TEM and SAED analyses confirm the development of a crystalline spinel phase (ZnAl2O4) with good structural uniformity, particularly in the sample deposited after 630 cycles. This phase identification is further supported by Raman spectroscopy, which reveals distinct vibrational modes (T2g, Eg, A1g) characteristic of a well-formed cubic spinel structure with no detectable secondary phases.
These structural findings are reflected in the optical properties, especially the band gap energy (Eg) and refractive index (n). The presence of a more ordered crystalline phase, along with relatively low surface roughness (8 nm), in the 630-cycle sample corresponds to a well-defined band gap of 2.94 eV and a high refractive index of 1.740. In contrast, samples with higher roughness or thinner layers tend to show greater variation in both Eg and n, likely due to increased defect density or interface scattering. This supports the conclusion that optical performance is strongly influenced by structural ordering and surface morphology, confirming the relevance of combined structural–optical analysis in assessing thin-film quality for optoelectronic applications.
The sheet resistance values decreased non-linearly with an increasing number of ALD cycles, as shown in
Figure 11. The lowest resistance value, 85 Ω/□, was achieved for films deposited using 1260 cycles, corresponding to approximately 189 nm thickness. This trend can be attributed to the enhanced crystallinity and improved carrier concentration with increasing film thickness, which aligns with observations reported in the literature [
27,
28]. These findings are consistent with previous studies. For instance, Cisneros-Contreras et al. reported a sheet resistance of 32 Ω/□ for ZnO:Al thin films with 80% transmittance, highlighting the potential of AZO as a transparent conductive oxide [
27]. Similarly, research by Kim et al. demonstrated that optimizing Al doping levels in ZnO:Al thin films can lead to resistivities as low as 7 × 10
−4 Ω·cm, emphasizing the importance of precise doping control in achieving desirable electrical properties [
28].
The photovoltaic performance of silicon solar cells with varying ZnO:Al thicknesses was assessed by analyzing key parameters: short-circuit current density (Jsc), open-circuit voltage (Voc), and power conversion efficiency (PCE) calculated based on current-voltage characteristics (
Figure 12). The ZnO:Al thin films were deposited using ALD at 200 °C, with supercycle counts ranging from 210 to 1260 cycles, corresponding to different layer thicknesses and sheet resistances. The electrical properties of the polycrystalline silicon solar cells with ZnO:Al thin films are summarized in
Table 3. As the number of ALD cycles increased, the sheet resistance of the ZnO:Al thin film decreased, leading to improved electrical conductivity. This enhancement in conductivity positively influenced the photovoltaic parameters up to a certain threshold. Notably, the solar cell with 840 ALD cycles exhibited optimal performance, achieving a Jsc of 36.0 mA/cm
2, Voc of 590 mV, and a PCE of 15.3%. However, further increasing the number of cycles to 1260 resulted in a decline in performance, with a Jsc of 34.8 mA/cm
2, Voc of 580 mV, and a PCE of 13.7%. This decline is attributed to increased optical absorption due to the excessive thickness of the ZnO:Al thin films, which hampers light transmission to the active region of the solar cell.
It is worth noting that most studies involving ZnO:Al (AZO) thin films as transparent conductive oxides (TCOs) are focused on solar cell architectures such as CIGS, perovskite, or amorphous silicon devices, which operate under different optical and electrical conditions compared to conventional polycrystalline silicon (poly-Si) cells. In contrast, the present study explores the integration of ALD-deposited AZO layers in thick-wafer poly-Si solar cells, which remains an underexplored area in the literature. The specific influence of AZO layer thickness, morphology, and deposition parameters on both optical transparency and electrical performance is systematically analyzed. These findings represent a novel contribution, particularly for industrial poly-Si solar cell technologies, where precise control of front-side coatings can significantly impact device efficiency.