Role of ZnO and MgO interfaces on the growth and optoelectronic properties of atomic layer deposited Zn_1−xMg_xO films

Zn$1−x$Mg$x$O films show excellent optical properties and a tunable bandgap, from 3.3 eV up to $∼4$ eV with the addition of Mg (up to $x=0.4$⁠), making them well-suited for optoelectronic applications. They have been widely used in UV light emitting diodes¹ as well as in photovoltaic devices as a replacement of toxic CdS buffer layers in copper-indium-gallium-sulfide (CIGS) based thin film solar cells.² In the use of Zn$1−x$Mg$x$O films as buffer layers, the change of the bandgap with the addition of Mg leads to a shift of the conduction band, providing suitable band alignment with CIGS absorbers, thus achieving better device efficiencies.³ 

Because of their popularity, Zn$1−x$Mg$x$O films have been synthesized by various methods, molecular beam epitaxy,⁴ solgel synthesis,⁵ RF sputtering,⁶ to grow MgO/ZnO superlattices,⁷ MgZnO/ZnO heterostructures,⁸ and Zn$1−x$Mg$x$O buffer layers.^3,9,10 Atomic layer deposition (ALD) of Zn$1−x$Mg$x$O offers advantages of low-temperature processing, high quality conformal film growth, and good control over the composition.

ALD relies on the self-limitation of surface reactions for its highly conformal and controlled growth. A usual way to reliably obtain a ternary material is to alternate two binary ALD processes in sequence in a so-called supercycle, where the composition can be simply varied by varying the relative number of binary processes in each supercycle. Ideally, the growth and properties of the resulting ternary material are expected to be a simple linear combination of the ones of pure materials. However, the interplay between ALD chemistries of the alternating binary processes results in complexities, causing the growth and properties of the ternary material to deviate largely from the ideal case.¹¹ Understanding the growth mechanism of a ternary ALD process becomes crucial to precisely control and grow materials with desired properties. In the case of Zn$1−x$Mg$x$O films, most previous studies on ALD growth focused on device performance³ and film characterization,^12,13 and only a few studies investigated the growth kinetics.^14,15

Although Törndhal et al. investigated the growth behavior of atomic layer deposited Zn$1−x$Mg$x$O, the interplay between ZnO and MgO ALD chemistries on the growth kinetics remained unexplored. Studies on ALD grown Zn$1−x$Mg$x$O using low-energy ion scattering and Auger electron spectroscopy addressed the growth behavior to an extent, suggesting the need for in-depth correlation between growth and structure-properties.^14,16 The only study to provide the detailed synthesis-structure-property relationship of ALD grown Zn$1−x$Mg$x$O (Ref. 15) investigated the interplay between the growth kinetics of ZnO and MgO using in situ quartz crystal microbalance analysis. Still, a clear model on optimizing the Zn$1−x$Mg$x$O ALD process to achieve optimal mixing of materials, efficient doping of Mg, tuning of bandgap, and electrical conductivity at lower temperatures is missing.

In this work, we demonstrate a reproducible thermal Zn$1−x$Mg$x$O (ZMO) ALD process using supercycles with bilayer periods (BPs) ranging from a few cycles to 80. We discuss the relationship between deposition parameters, the pulse ratio and bilayer period, and ZMO growth, composition, optical, and electrical properties. We investigate the growth kinetics of the ZMO ALD process by analyzing the effect of the bilayer period. We establish a suitable ALD supercycle sequence for optimal doping and tuning of bandgap at lower deposition temperatures as desired for photovoltaic applications.

The ZMO films were grown in a Beneq TFS 500 reactor with Diethylzinc min. 95$%$ [DEZ or Zn(C$2$H$5$⁠)$2$], bis(cyclopentadienyl) magnesium (99.99+$%$-Mg) PURATREM [MgCp$2$ or Mg (C$5$H$5$⁠)$2$], and de-ionized water as precursors for zinc, magnesium, and oxygen, respectively. The MgCp$2$ precursor was heated to 80 $°$C in a hot source. Argon was used as a purge gas with a flow of 250 sccm. The pulse-purge sequence was optimized as follows: for ZnO: 200 ms DEZ pulse, 5 s purge, 200 ms H$2$O pulse, 5 s purge; for MgO: 4 s of MgCp$2$ pulse, 5 s purge, 200 ms H$2$O pulse, 5 s purge. For MgO, we attain saturation at 130 $°$C with 4 s pulse time of MgCp$2$⁠. Owing to the insufficient vapor pressure within the canister, the MgCp$2$ canister is pressurized by first opening the carrier gas valve for 0.5 s, followed by a brief wait time (50 ms), then simultaneous opening of both the carrier gas valve and the precursor valve for 4 s to draw the precursor into the reaction chamber. The ZMO films were grown at different deposition temperatures following a supercycle approach, where a supercycle consists of z number of cycles of (DEZ + H$2$O) pulses (named ZnO cycles in the following) and m number of cycles of (MgCp$2$ + H$2$O) pulses (named MgO cycles in the following). The composition of Zn$1−x$Mg$x$O was varied by changing two supercycle parameters, namely, the pulse ratio (PR) and the BP.

In a Zn$z$Mg$m$O ALD supercycle process, we define the PR as the ratio of MgO cycles to the total number of cycles in a supercycle during the complete deposition, i.e., $PR=m/(z+m)$⁠. The BP is defined as the average number of MgO and ZnO individual cycles contained in a supercycle and repeated to deposit the mixed material, i.e., it is equal to the average value of $z+m$⁠. Supercycles with different BP are shown, for example in Fig. 1 for the PR of 0.2. The smallest BP for this PR of 0.2 is 5, consisting of 4 ZnO + 1 MgO cycle, which can be increased to 10 by doing 2 MgO cycles every 8 ZnO cycles, and to get a BP of 20 it requires cycles of 16 ZnO + 4 MgO, and so on. Targeting around 100 nm thick films, this supercycle is repeated a number of times, e.g., for PR of 0.2, around 500 cycles of 4 ZnO + 1 MgO were performed.

The ZMO films were grown at the two deposition temperatures of 130 and 150 $°$C. As our primary target is the use of ZMO films as a buffer layer for CIGS solar cells, we have, thus, restricted the study to low-temperature deposition and a maximum PR of 0.3, which is the most relevant range for said application. Although previous studies report the deposition of Zn$1−x$Mg$x$O films at 120 $°$C, we did not attain saturation at 120 $°$C in our setup. The ZMO films were grown on 300 mm diameter monitor grade Si with $⟨100⟩$ orientation, thickness $775±25$ $μ$m, and on 1 mm thick $25×25mm2$ optical quartz glass substrates with UV-quality, with transmission $T>80$% at 185 nm. All characterizations were performed on $∼100$ nm thick ZMO films. In all our ZMO films, MgO remains the topmost layer.

The thickness of deposited films was measured by ellipsometry (VASE Ellipsometer M2000) on films deposited on Si using a general oscillator (GENOSC) model for ZnO, a Sellmeier model for MgO, and combining the two into an effective medium approximation (EMA) model for ZMO. The deposition remains homogeneous and uniform across the 300 mm Si wafer, with the thicknesses of films near the inlet and outlet differing by less than 2%. The difference increases to 5% for ZMO with PR higher than 0.25 and for pure MgO, irrespective of the deposition temperature.

X-ray reflectivity was performed using PANalytical X’Pert Pro diffractometer (PANalytical) with Cu-K$α$ radiation to evaluate the thickness and density. The film density was obtained by fitting the XRR data using GenX (see Ref. 17 and Figs. 10 and 11 in the Appendix). The densities of our MgO and ZnO films were found to be 3.127 $g$/$cm3$ (0.0467 AU/$Å3$⁠) and 5.33 $g$/$cm3$ (0.0395 AU/$Å3$⁠), respectively. The values are in close agreement with densities reported in the literature, i.e., 3.07 $g$/$cm3$ for MgO and 5.61 $g$/$cm3$ for ZnO.¹⁸ 

The x-ray diffraction (XRD) data were collected with the grazing incidence technique with the 2$θ$ range from 20 to 60$°$ using Cu-K$α$ radiation in Bruker D8 diffractometer (Bruker, USA) and were analyzed according to the following PdF cards 04-009-7657 and 00-004-0829 for ZnO and MgO, respectively. The atomistic composition of the films was determined by energy dispersive x-ray spectroscopy using Xmax 50 $mm2$ EdX detector (Oxford Instruments) attached to Helios NanoLab 650 (FEI, USA) scanning electron microscope (SEM). At an acquisition angle of 35$°$⁠, an e-beam of 3 keV at 40 $μm$ magnification was sufficient to detect the $Kα$ of Mg at 1.253 keV. The analyses were made using inca software.

Optical spectra (transmission and reflection) were measured in the 250–2500 nm wavelength range using a Perkin Elmer Lambda 1050 spectrophotometer equipped with a 150 mm integrating sphere. The absorption coefficient $α$ of the films deposited on glass are computed based on the relation¹⁹ 

where $T$ and $R$ are the transmittance and reflectance, respectively, with $d$ being the thickness of the film. The bandgap of the deposited Zn$1−x$Mg$x$O films is calculated based on the relation for direct bandgap materials,

Electrical characterization was performed by measuring the sheet resistance of the thin films deposited on glass using an in-line four-point-probe system with a probe spacing of 1 mm and a diameter of 100 $μm$ (Jandel).

Growth: Individual growth per cycle (GPC) measured from pure ZnO and MgO samples is 1.95 and 1.40 Å/cycle at 130 $°$C, and 2.05 and 1.41 Å/cycle at 150 $°$C, respectively. The MgO GPC does not differ more than 10% in the temperature range of 130–150 $°$C and agrees with values previously reported.¹⁴ The ZnO GPC is observed to increase with temperature and only decreases after 150 $°$C.

Ideally, the growth rate of a ternary material is a linear combination of binary growth rates. The expected GPC of Zn$1−x$Mg$x$O, $gZMO$⁠, is, hence, calculated based on the GPC of ZnO (⁠$gZnO$⁠), MgO (⁠$gMgO$⁠), and the fraction in which they are pulsed (⁠$PR$⁠),

Figure 2 compares the expected GPC from Eq. (3) and the observed GPC of ZMO as a function of the pulse ratio, deposited at different temperatures with a similar bilayer period. As the GPC of pure ZnO is higher than pure MgO, the GPC of ZMO decreases relative to the increase in MgO pulses. The experimental data follow the expected trend, but it is below the expected GPC. It is usual for a ternary material to deviate from the ideal behavior, and nonideal growth behaviors of ternary from the respective binaries have been observed for zinc tin oxide (ZTO),²⁰ aluminum doped zinc oxide,²¹ and also in zinc magnesium oxide (ZMO).³ 

Composition: Herein, we derive the expected atomic composition that reflects the true growth behavior of a ternary material by taking into account the different GPCs and densities of individual binary materials. The expected atomic composition is calculated from the densities of ZnO and MgO obtained from XRR and the GPCs from ellipsometry, with the following formula:

From the above equation, one expects a higher Mg content at lower deposition temperatures, due to the decrease in growth of ZnO at lower temperatures.

Figure 3 shows the relative amount of Mg content ([Mg]/([Zn]+[Mg])) observed when varying the pulse ratio of ZMO films grown at 130 and 150 $°$C with a similar bilayer period. We observe an increase in magnesium content with a decrease in the deposition temperature of ZMO films, as predicted by Eq. (4), due to the decreasing GPC of ZnO with decreasing temperature, which also supports similar behaviors observed previously.³ Moreover, the Mg content in our ZMO films is always observed to be higher than the expected values calculated (shown as squares in Fig. 3). This higher Mg incorporation in the ZMO films suggests a possible reduction of ZnO growth or enhancement of MgO growth in the mixed material as compared to the bulk values, presumably due to interface effect between both materials. To understand further the interaction between ZnO and MgO deposition, we study the effect of the bilayer period.

We use the bilayer period to understand the growth properties, and, in particular, the interaction between ZnO and MgO cycles in the mixing of ZnO and MgO into ZMO films. The bilayer period governs whether we obtain a homogeneous or laminar film. In this supercycle approach, to understand the effect of the bilayer period, we keep the pulse ratio of our ZMO films constant and vary the bilayer period. For a given pulse ratio, we grow films at various periods from the smallest cycle possible, as explained in Fig. 1. Table I shows the pulsing sequence for different bilayer periods for a pulse ratio of 0.2 and the number of cycles the supercycle needs to be repeated to grow approximately 100 nm thick films.

It should be noted that the smallest bilayer period varies for different pulse ratios. In the case of 0.25, the smallest bilayer period is 4 with 3 ZnO and 1 MgO, for $PR=0.2857$⁠, the minimum BP is 7 with 5 ZnO and 2 MgO, and for $PR=0.3$⁠, the BP of 5 follows a fine mixture involving the following sequence: 3 ZnO, 1 MgO, 3 ZnO, 1 MgO, 1 ZnO, and 1 MgO (detailed Tables II and III in the  Appendix). The GPC is expected to remain the same irrespective of the bilayer period [see Eq. (3)] assuming the surfaces have the same number of reacting groups.

Growth: The GPCs of ZMO films observed at various bilayer periods for different pulse ratios (PR = 0.2 and 0.3) at 150 $°$C are shown in Fig. 4. We observe a deviation in GPC from the expected values, the growth of ZMO films decreases with the increase in bilayer period. Here, we find that, varying the bilayer period, the GPC does not follow the expected trend. This nonideal behavior indicates that the growth of at least one material is influenced by the other. Considering that the GPC obtained at the smallest bilayer period is the one mostly influenced by the interface and that the GPC starts falling at larger bilayer periods, it is evident from Fig. 4 that the growth of MgO on ZnO follows a “substrate-enhanced growth,” where the GPC is higher at the beginning of the growth (nucleation stage).²² Initially, when the bilayer period is small, always one MgO cycle is interspersed among a few successive ZnO cycles (1–4). This leads to a larger initial GPC, as seen in Fig. 4 for bilayer period 5. The smallest bilayer period or fine mixing leads to a transient regime of nonconstant growth. After that, however, the GPC of ZMO starts falling below the expected value with an increase in the bilayer period, where the number of reaction cycles of both ZnO and MgO increases within one supercycle.

The GPC at larger bilayer periods is expected to incline toward bulk values with an increase in the number of ZnO and MgO cycles within one supercycle, but the GPC falls below the expected estimate. This discrepancy can be attributed to the errors in the EMA model used in determining the thicknesses for the mixed ZMO material (see Sec. II). The strong approximation made in this model can result in the global shift of the values of GPC of ZMO films. We, however, still believe that the relative variation in thicknesses and, hence, GPC as a function of bilayer period is still reliable as n and k we obtain are consistent throughout.

Composition: Not only the growth but also the composition deviates from the expected values when varying the bilayer period. Figure 5 shows the relative Mg content observed when varying the bilayer period at 130 and 150 $°$C for different pulse ratios. The different pulse ratios are indicated in different colors and the corresponding dashed lines represent the expected composition, calculated using Eq. (4). We find that the Mg content decreases with increasing the bilayer period, which is consistent with the result presented in Fig. 3, where we observed the Mg content to be higher than the expected values at a low bilayer period.

Higher Mg concentration at lower bilayer period suggests that the incorporation of Mg is maximum in supercycles where ZnO and MgO are well-mixed. From the variation of composition with respect to the bilayer period, we conclude that one Mg cycle is enough to incorporate Mg in the Zn$1−x$Mg$x$O system, suggesting that the growth of MgO is not retarded either by the preceding or succeeding cycle of ZnO. Meanwhile, both the growth and Mg content decrease when more MgO cycles start to occur consecutively as the bilayer period increases. This means that, when MgO grows on MgO, there is a decrease in both GPC and Mg content of our ZMO films, but when MgO grows on ZnO, both the GPC and Mg concentration of ZMO films increase. This variation in growth and Mg content with bilayer period shows that the ZnO layer enhances the growth of MgO in the synthesis of ZMO by ALD.

In situ spectroscopic ellipsometry studies on atomic layer deposited MgO films on Si have shown that the initial growth of MgO is nonlinear and strongly dependent on the density of OH groups on the surface.²³ Studies on surface hydroxylation pointed out that dissociative water adsorption that enhances the surface hydroxylation only occurs at defect sites of MgO(100).²⁴ As a ZnO surface gets saturated with OH monolayer,²⁵ a ZnO layer can offer better hydroxyl density than an MgO layer, leading to increased GPC and Mg incorporation at smaller bilayer periods where ZnO and MgO are well mixed. In addition, the microstructure of the material could also influence the hydroxyl density, as the roughness of ZnO is higher compared to MgO.¹⁵ Therefore, having an MgO layer formed by successive cycles of MgO, neither promotes the growth of the next MgO layer nor the next ZnO layer due to the availability of less number of hydroxyl groups for the precursor’s half-reaction. Our conclusion that the ZnO layer enhances the growth of MgO in ZMO by ALD is further supported by QCM study, which reported the mass gain of MgO cycles to be higher than pure MgO cycles in the growth of ZMO at 120 $°$C.¹⁵ 

For an effective mixing of MgO and ZnO in ZMO films, the bilayer period should be kept minimal. Unlike zinc tin oxide, where mixing of a small number of ZnO cycles into SnO$x$ results in deposition of less Sn,²⁰ herein deposition of Mg is effective when mixing a small number of ZnO cycles with MgO. Thus, by optimizing the bilayer period, we achieve good growth and high Mg content in our ALD grown Zn$1−x$Mg$x$O films. We also demonstrate here that investigating the effect of the bilayer period can be an easy, potential alternative tool to QCM studies in understanding the mixing and interplay of binary materials in the synthesis of ternaries by ALD.

Diffractograms of deposited films show that ZMO films exhibit a single phase wurtzite ZnO type structure Fig. 6(a). With our pure MgO films being weakly crystalline, the ZMO films do not show any MgO type structure, and we do not find two-phase regions even with higher Mg content. Compared to pure ZnO films, the addition of Mg causes peak shifts in 2$θ$ to lower angles [see Fig. 6(b)]. For example, both the (100) and the (110)-reflections shift a maximum of 1.66$°$ to lower angles. The in-plane lattice constant a is found to increase from 3.243 Å (pure ZnO) to 3.286 Å in our deposited ZMO films resulting in a positive strain of 1.32$%$⁠. This is due to the tensile stress caused by the lattice mismatch between ZnO and MgO. Furthermore, we observe that the variation of bilayer period does not cause any additional peak shifts [Fig. 6(b)], showing that the ZnO layer does not relax even for the largest bilayer period, corresponding to a ZnO thickness up to 8 nm. However, the relative intensity of the peaks changes with varying bilayer periods. In particular, (101)-reflection is higher than (002)-reflection at the smallest bilayer period, while this ratio is inverted at the larger bilayer period, similar to the ratio obtained for relaxed ZnO films. This is most probably related to the columnar growth of ZnO, which stops at an earlier stage for the lowest bilayer period.

To establish a nexus between the growth behavior and material properties, we have investigated the optical properties of ZMO films grown with various pulse sequences. Understanding the optical characteristics of the ZMO films aids in the precise tuning of the bandgap and tailoring the ZMO films toward specific optoelectronic or photovoltaic applications. Figure 7(a) shows the UV-Vis optical transmittance of ZMO films deposited on glass at 130 and 150 $°$C with various pulse ratios denoted in the legend. We observe a shift in absorption edge toward lower wavelengths with the addition of Mg. We determine the bandgap from the Tauc plot, by fitting the absorption edge assuming a direct bandgap [see Eq. (2) and Fig. 12 in the Appendix]. For the ZMO films with larger Mg content (PR $>0.3$⁠), the determination of the bandgap is limited by absorption from the glass substrate.

The variation of the bandgap with respect to the measured Mg content in our ZMO films deposited at both 130 and 150 $°$C can be seen in Fig. 7(b) as squares and dots, respectively. With the addition of Mg, as the absorption edge shifts toward lower wavelengths, the bandgap of deposited ZMO films increases. This variation of bandgap with the addition of Mg agrees well with previously reported trends.^1,15 It is clear from Fig. 7(b) that the deposition temperature has no direct influence on the bandgap, which is exclusively dependent on the Mg composition. The bandgap can, hence, be tuned by changing the Mg composition, which can be precisely controlled by altering the pulse ratio of the ALD process during synthesis.

The effect of the bilayer period on the optical properties of ZMO films is shown in Fig. 7(b), with triangles representing the bandgap of ZMO (PR = 0.3) grown at 150 $°$C for different bilayer periods. The bandgap values of ZMO (PR = 0.3) films decrease toward the one of ZnO (3.29 eV) with increasing bilayer period, even though the Mg content does not change significantly. At larger periods of 40 and 60, where the bandgap is almost equivalent to ZnO, it is evident that the bandgap of ZMO films is dominated by the ZnO bandgap.

This strong effect of the bilayer period on the optical properties can be further understood from Fig. 8, which shows the variation of the bandgap as a function of the bilayer period for both 130 and 150 $°$C. The bandgap values of ZMO films for a given pulse ratio are high for ZMO films with finely mixed ZnO and MgO but start decreasing with an increase in the bilayer period. For a given pulse ratio, we observe the highest bandgaps at the shortest bilayer periods where one MgO layer is mixed with few ZnO cycles. The variation in bandgap within a small range of the bilayer period from 10 to 20 implies that there is a weak interdiffusion of MgO into ZnO at these periods.

Looking into the length scale, ZnO has an excitonic Bohr radius $aB$ of 1.8 nm with a dielectric constant $ε=6.56$⁠.²⁶ With the change in the bilayer period, the thickness of ZnO layer varies, e.g., for a PR of 0.2, it increases from 1.6 to 3.2 nm going from bilayer period 10 to 20. When the thickness of the ZnO layer is below the Bohr radius (i.e., $<1.8$ nm), the electrons experience a mixed material. As the bilayer period increases, the electrons encounter more of one material. The influence of ZnO on the carriers dominates as the thickness becomes greater than the Bohr radius. Also, in quantum well structures, when the size of the well is smaller than Fermi wavelength $λF$ (or de Broglie wavelength), the carriers are confined and experience quantum size effects. Treating the ZnO layer as a quantum well, the variation in bandgap with the bilayer period could be explained by the well thickness-dependent shift in the effective bandgap due to the quantum confined stark effect.²⁷ From the bulk concentration determined from Hall measurements, we find the Fermi wavelength of the ALD grown ZnO films as 4.32 nm. As the ZnO well thickness increases (⁠$∼λF$⁠) with the bilayer period, the quantum size effect decreases and, hence, the effective bandgap also decreases. As the bilayer period further increases (⁠$>λF$⁠), ZnO and MgO form a laminated structure rather than a mixed material, and the bandgap tends to the lowest value, which corresponds to ZnO.

Therefore, we obtain well-mixed nanolaminates of ZMO with a smaller bilayer period, while, as the bilayer period increases, we see more of a multilayer behavior of ZMO films with ZnO dominating the optical properties. This transition from mixed to multilayer material begins at a bilayer period greater than 10. This indicates a penetration depth of Mg into ZnO lower than typically 7–8 ZnO monolayers, which is the ZnO layer thickness at these pulse ratios. Additionally, the change in the bandgap, which continues until a bilayer period up to 20, can be exploited, making the bilayer period another supercycle parameter of significance in controlling the material properties.

ZMO films deposited at 150 $°$C starting from a bilayer period of 10 showed finite conductivity. Only ZMO films with the lowest pulse ratios, i.e., with the lowest number of MgO cycles, showed conductivity. ZMO films with a pulse ratio of 0.1 (9 ZnO + 1 MgO) and 0.2 (8 ZnO + 2 MgO) showed a conductivity of 1.445 and 0.266 S/cm, respectively, while the conductivity of films with PR of 0.2857 and 0.3 is nonmeasurable, meaning that the conductivity of ZMO films decreases with increasing MgO cycles. Since pure MgO is highly resistive, an increase in conductivity of ZMO films could be attributed to the ZnO content in the film. Also, the number of ZnO layers present might play a role in the increased conductivity of ZMO films, which is investigated by observing the effect of the bilayer period.

Figure 9 shows the variation of the conductivity as a function of the bilayer period for ZMO films deposited at 150 $°$C and at PR = 0.2. We see that the conductivity increases when increasing the bilayer period. This effect is clearly observed only for the case of low Mg doping (PR = 0.2). The variation of conductivity with respect to the bilayer period for higher pulse ratios (PR > 0.2) could not be demonstrated, either the errors in measurement became too large or because it was beyond the limit of the measurement. Similar behavior was observed in ZMO films deposited at 130 $°$C, which also exhibited poor conductivity.

From Fig. 9, we observe the conductivity of ZMO films starts improving after the bilayer period of 20. The ZMO films show low conductivity between the bilayer period of 20 and 10, and below 10, it is not conductive at all. This increase in conductivity could be attributed to the increase in the number of successive ZnO layers going from a low to high bilayer period within a supercycle. At lower bilayer periods, ZMO films behave as a well-mixed material, but as the bilayer period increases, ZnO starts to dominate the electrical properties of the ZMO films. The electrical characteristics do not demonstrate a clear transition of ZMO films from mixed material to multilayers with an increase in the bilayer period as observed in optical properties. Still, the increased conductivity at larger bilayer periods (40 and above) indicates a multilayered ZMO film with ZnO dominating the electrical properties.

We have demonstrated a reproducible thermal Zn$1−x$Mg$x$O ALD process using various supercycles parameters. We discussed the relationship between deposition parameters, the pulse ratio and bilayer period, on ZMO growth, composition, optical, and electrical properties. By studying the effect of the bilayer period on the growth and composition of the material, we showed that MgO is enhanced by ZnO in the growth of ZMO films by ALD, thus demonstrating the use of the bilayer period as an effective tool to understand the mixing and interplay of binary materials in the synthesis of ternary material by ALD. We discussed and established a suitable ALD supercycle sequence for the effective incorporation of Mg into ZnO and tuning of the bandgap of ZMO films at lower deposition temperatures, as desired for photovoltaic applications.

This work was supported by the Luxembourg National Research Fund (FNR) for the financial support (No. PRIDE17/12246511/PACE). The authors thank Tony Schenk for assisting with XRR measurements.

The authors have no conflicts to disclose.

Poorani Gnanasambandan: Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). Noureddine Adjeroud: Methodology (equal); Resources (equal). Renaud Leturcq: Funding acquisition (equal); Project administration (equal); Resources (equal); Supervision (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

The thicknesses measured in XRR agree with the ellipsometry thicknesses determined for the same samples (Figs. 10 and 11).