Tai
Nguyen
ab,
Nathalie
Valle
a,
Jérôme
Guillot
a,
Jérôme
Bour
a,
Noureddine
Adjeroud
a,
Yves
Fleming
a,
Mael
Guennou
b,
Jean-Nicolas
Audinot
a,
Brahime
El Adib
a,
Raoul
Joly
ab,
Didier
Arl
a,
Gilles
Frache
a and
Jérôme
Polesel-Maris
*a
aMaterials Research and Technology Department, Luxembourg Institute of Science and Technology, 41, rue du Brill, L-4422 Belvaux, Luxembourg. E-mail: jerome.polesel@list.lu
bDepartment of Physics and Materials Science, University of Luxembourg, 41 rue du Brill, L-4422 Belvaux, Luxembourg
First published on 15th March 2021
The growth process of zinc oxide (ZnO) thin films by atomic layer deposition (ALD) accompanied by the presence of oxygen gas pulsing is investigated by means of the isotopic tracking of oxygen 18O from the water precursor and oxygen 16O from the gas. In a previous study [T. Nguyen et al., Results Mater., 2020, 6, 100088, DOI: 10.1016/j.rinma.2020.100088], by means of structural, electrical, and optical characterizations, we identified key growth parameters of this unusual ALD process. Unexpectedly, the influence of molecular oxygen on the crystallography, microstructure, and morphology of the hundred-nanometer- to micrometer-thick ZnO films was significant. In this study, we present an unprecedented methodology by combining isotopic tracers with mass spectrometry to elucidate the role of the two different sources of oxygen atoms during the evolution of the growth. Notably, the use of in situ quartz crystal microbalance (QCM) and Secondary Ion Mass Spectrometry (SIMS) reveals new insights into the reaction mechanism for ZnO thin film growth. On the one hand, the non-negative mass change during the ZnO growth without O2 gas is attributed to the presence of bare zinc atoms on the surface due to the reaction between monoethyl zinc and hydroxyl groups of the water precursor after the diethyl zinc pulse. On the other hand, the detection of ZnxOyC2H5− ions by Time-of-Flight SIMS (TOF-SIMS) and the mass increase during the O2 pulse suggest a new reaction mechanism for the ZnO thin film growth in the presence of gaseous O2 where the ethyl ligand of the zinc precursor can react with O2 to form ethylperoxy radicals. The formations of the ethylperoxy zinc and/or zinc atoms lead to more adsorption of water to form ethylhydroperoxide during the water pulse, inducing the positive mass change. The use of an isotopic substitution allowed us to unambiguously associate the mass gain with the gradual incorporation of gaseous oxygen throughout the growth process and thereby support the chemical reaction.
Zn(C2H5)2 + H2O → ZnO + 2C2H6 | (R1) |
DEZ pulse: Zn(C2H5)2 + I–OH →I–O–Zn–C2H5 + C2H6 | (R2a) |
Water pulse: H2O + I–O–Zn–C2H5 → I–O–Zn–OH + C2H6 | (R3) |
I–O–Zn–C2H5 + I–OH → (I–O)2–Zn + C2H6 | (R2b) |
Here, we address these issues using an innovative approach based on a combination of isotopic substitution during the film growth and secondary ion mass spectrometry (SIMS) to monitor the contribution of each source of oxygen atoms. We used 18O-labelled water as a precursor in order to distinguish between the respective contributions of oxygen atoms from water precursor (H218O) and from gaseous oxygen (16O2). In situ quartz crystal microbalance (QCM) was implemented to study the effect of oxygen presence on the ALD growth characteristics of ZnO thin films. In addition, time-of-flight SIMS (ToF-SIMS) was carried out to identify the resulting chemical composition of the ZnO thin films.
Δm/S = −(μρ)0.5Δf/(2f02) | (1) |
X-ray diffractometry (Diffractometer Bruker D8 Discover with Cu Kα radiation and a 5-axis Eulerian cradle) was conducted in θ–2θ configuration to examine the crystalline quality of the ZnO thin films on all samples. The microstructure of samples was studied by scanning electron microscopy (SEM) on a Helios Nanolab 650 FIB-SEM instrument (FEI Company, USA). A four-point probe (Ecopia HMS-3000) was performed to measure the resistivity of the thin film on glass substrates. Elemental composition and chemical states were investigated by X-ray photoelectron spectroscopy (XPS) (Axis Ultra DLD, Kratos Analytical Ltd) over a surface area of 110 × 110 μm2 using an X-ray source (Al Kα monochromated, E = 1486.6 eV) at a power of 150 W, and an energy resolution of 1.5 eV for survey scans and 0.55 eV for narrow scans determined on a silver sample. The spectra were calibrated in energy from the Zn 2p peak in ZnO at 1022.0 eV. The surface contamination was removed by etching an area of 3 × 3 mm2 with an Ar+ ion beam operating at 2 kV and 2 μA.
The transmittance and reflection spectra of ZnO thin films on glass slides were collected by UV-visible measurements (LAMBDA 1050 UV/Vis Spectrometer, PerkinElmer) in a 250–2000 nm range. Room-temperature photoluminescence (PL) spectra were acquired by a Renishaw inVia confocal microRaman spectrometer using an excitation wavelength of 325 nm produced by an 8 mW He–Cd laser focused through a Thorlabs UV objective with 40× magnification and a numerical aperture of 0.5. A 300 g mm−1 grating enabled analysis in the 350–900 nm range. A single crystal ZnO (MTI Corporation) was used as a reference.
Dynamic-secondary ion mass spectrometry (D-SIMS) (CAMECA SC-Ultra) was used to analyze the elemental and isotopic composition of the films grown. First, depth profiles were acquired with a Cs+ bombardment operating at 1 keV and a 6 nA primary ion beam scanned over an area of 500 × 500 μm2. Secondary positive ions were detected as MCs+ clusters (M is the element of interest) from a 60 μm area centered on the scanned area. Secondly, the oxygen isotopic signatures were determined with the M− mode analysis over an area with a diameter of 8 μm. A mass resolution of 1000 allowed isobaric interference on 18O to be eliminated.
A Helium Ion Microscope (HIM, Nanofab, ZEISS, Peabody, USA) coupled with in-house SIMS system (HIM–SIMS)21,22 was used to analyze and map the 16O and 18O isotopes. SIMS images were acquired by collecting the secondary ions emitted from the film surface scanned by a 20 keV Ne+ ion beam of 2 pA over a matrix of 512 × 512 pixels with a counting time of 5 ms per pixel.
The time-of-flight secondary ion mass spectrometry experiments were performed using a commercial TOF-SIMS V time-of-flight mass spectrometer (ION-TOF GmbH, Münster, Germany) operating at a pressure of 10−9 mbar. Mass spectra were carried out with a 25 keV pulsed Bi3+ cluster ion source, delivering a 0.31 pA target current. The analyzed area was 500 mm × 500 mm. The analyses were performed using a primary ion dose density maintained at 1011 ions per cm2, which is below the so-called static SIMS limit. In these conditions, secondary ions are produced, reflecting the molecular and elemental composition of the first nanometer of a sample. The data were obtained in negative mode and the secondary ion mass spectra were calibrated using Cn– carbon clusters.
Fig. 1 XRD patterns and top-view SEM images of (a) the ZnO thin film grown without, (b) with gaseous oxygen. The scale bar in the SEM images is 300 nm. |
The optical bandgap of ZnO thin films is derived from the Tauc plot, as shown in Fig. 2(a). The reflectance and transmittance spectra of ZnO thin films and a glass slide were carried out to calculate the absorption spectra of ZnO thin films. For a direct bandgap semiconductor,23 the relationship between the absorption coefficient α and photon energy hν is given by: (αhν)2 = A(hν − Eg), where h is Planck's constant, A is a constant and Eg is the optical bandgap. The bandgap is determined by extrapolating the linear part of the Tauc plots to α = 0, which is 3.2 eV for ZnO thin films grown without oxygen gas and increases up to 3.3 eV for the film grown with the 1 s pulse of oxygen.
Fig. 2(b) presents the normalized photoluminescence spectra performed at room temperature. The strong energy peak located from 3.28 to 3.31 eV is assigned to the near band edge (NBE) emission. A broad emission band (1.5 to 2.7 eV) is observed that results from defect-related DLE and can be deconvoluted into green (2.58 to 2.25 eV), yellow-orange (2.25 to 2.03 eV), and red (2.03 to 1.64 eV) emission bands. Although the nature of the deep-level emission is controversial, there are three main arguments that are widely accepted:24,25
(i) The green emission is associated either with oxygen vacancies simply ionized , oxygen antisites, or even zinc vacancies .
(ii) The yellow-orange emission corresponds to oxygen vacancies doubly ionized (VO°°).
(iii) The red emission is attributed to excess oxygen on the ZnO surface.
Interestingly, by adding oxygen molecules during the ALD process, the DLE region became minor compared to the NBE emission peak, whereas the DLE region can be clearly observed for ZnO thin film processed without oxygen gas at the same deposition temperature.
The chemical state and environment of oxygen and zinc were studied by X-ray photoelectron spectroscopy, as shown in Fig. 2(c) and (d). The Zn 2p peak is not affected by the presence of oxygen gas (not shown). In contrast, an additional peak located above 530 eV is observed in the O 1s spectrum when oxygen gas is added during the ALD processes (Fig. 2(c)). In general, the O 1s peak in ZnO compounds can be split into two components: a main peak related to the O–Zn bonds in the hexagonal wurtzite ZnO structure (∼530.60 eV),26–28 and an additional peak located at ∼532.2 eV assigned to hydroxides29,30 or to chemisorbed oxygen.27,28 The O 1s spectra acquisition was completed in less than 300 s after the end of the Ar sputtering to limit the adsorption of hydroxides occurring even under ultra-high vacuum on Zn dangling bonds present at the surface of the film (Fig. 2(c)). It can clearly be seen that not only O but also hydroxides are inserted inside the bulk of the film during the ZnO growth only when oxygen gas is added. This is also confirmed by the higher hydrogen level found in the SIMS depth profiles (Fig. 5(a) and (b)). They represent around 10% of the oxygen in the ZnO network. The valence band shape, sensitive to the filling of the oxygen vacancies, is also modified when adding oxygen gas into the synthesis process (Fig. 2(d)). The two structures around 4 eV and 7.5 eV are mainly due to electrons in the O 2p orbitals and the O 2p–Zn 4sp hybridized state respectively.31–33 An increase of the O 2p–Zn 4sp states, meaning these orbitals are more populated, is observed when oxygen gas is added during the synthesis process and is linked to a higher number of zinc atoms surrounded by oxygen atoms (O and/or OH) in the final wurtzite structure, thus decreasing the vacancy density in the film. Furthermore, the valence band positions of the two samples (determined without any energy calibration) clearly show a shift of 0.5 eV. Without the O2 pulse in the synthesis process, the Fermi level is closer to the conduction band, due to the higher oxygen vacancies amount in this n-type material. Based on this result, it is worth noting that the defects density due to oxygen vacancies (sample without O2 pulse) is much higher than the defects related to the incorporation of carbon resulting of the incomplete decomposition of DEZ (sample with the O2 pulse). This is confirmed by the fact that a peak shape evolution is observed in the O 1s spectrum versus the synthesis condition whereas carbon has not been detected in any sample by XPS whose sensitivity is lower than dynamic or ToF-SIMS. The O/Zn ratio was found to be 0.96 ± 0.03 and 0.92 ± 0.02 for the ZnO films grown with and without O2.
The in situ QCM characteristics of the ZnO thin film synthesized in the presence of oxygen gas are shown in Fig. 4. The timing sequence is written as 0.1 s (DEZ pulsing)–5 s (Ar purging)–1 s (O2 pulsing)–20 s (Ar purging)–0.1 s (18O-labeled DI water pulsing)–10 s (Ar purging).
Here also, the whole growth processes present a linear dependence of the mass gain on deposition time, as shown in Fig. 4(a). Enlarged views of the QCM results that highlight the growth characteristics of the ZnO film at indicated periods are shown in Fig. 4(b) and (c). After the first 3000 s, corresponding to 80 cycles, the steady-state in mass gain is achieved during the ALD loop stages for DEZ pulsing, inserted oxygen exposures, and water pulsing. The mass gain for the DEZ, oxygen gas, and H2O exposures is ΔmDEZ ≈ 70–75 ng cm−2, ΔmO2 ≈ 30–47 ng cm−2, ΔmH2O ≈ 33–35 ng cm−2, respectively. The total mass gain per cycle is Δm ≈ 133–157 ng cm−2. It is worth mentioning that the mass gain after the oxygen pulse rises as the deposition time increases, as presented in Fig. 4(d). Moreover, a positive mass change during the water pulse was observed instead of a negative one for the ALD process without oxygen gas.
Fig. 5(c) and (d) show that the depth-resolved SIMS analysis of 18O originated predominantly from the 18O-enriched water (H218O), 16O stemmed mainly from oxygen gas, and the 18O/16O ratio for ZnO thin films grown with and without an additional presence of oxygen gas, respectively. On one hand, the 18O/16O ratio remains unchanged at around 4.7 versus the sputtering time for the film grown with DEZ and water only. The 18O/16O obtained value of 4.7 is significantly small compared to the 18O/16O ratio of 18O-labeled deionized water of 32. This is due to a dilution effect as the introduction of H218O is done in the reaction chamber with certain levels of residual gases and moisture (the chamber pressure of approximately 1.2 mbar) where 16O predominates (16O: 99.757% and 18O: 0.205%, i.e.18O/16O ∼2.05 × 10−3). Consequently, the 18O/16O ratio of the ZnO film drastically decreases. On the other hand, it decreases from a value of 1.8 to 1.5 as the film grows (i.e. towards the low value of sputtering time) with the additional O2.
This decrease in the 18O/16O ratio upon the sputtering time implies that the oxygen atoms mainly stem from water at the early stage of the growth, meaning that the contribution of the water precursor is predominant. However, the contribution of additional O2 gas in the growth process gradually becomes significant as the film gets thicker. This is in line with the observation from the QCM data (Fig. 4(d)), which shows the increase in the mass gain after the O2 pulse. It is also worth noticing that amounts of carbon and hydrogen present in bulk in the ZnO film grown with O2 are about 50% higher than those of the ZnO film grown without O2 (Fig. 5(a) and (b)). This will be commented on later in the discussion.
Fig. 6 presents the ToF-SIMS analysis of the ZnO thin films. ZnxOyH− ions are detected for both ZnO thin films due to the fact that the films were grown by using DEZ and DI water as precursors. Interestingly, ZnxOyC2H5− ions are only detected in the ZnO films grown in the presence of gaseous O2. This signature suggests a direct chemical reaction of the O2 gas molecules with the growing ZnO-based thin film during the ALD process. This is also supported by the lower 18O/16O ratios measured (Fig. 5c and d) through the whole thickness of the film with the additional presence of O2 during the ALD process. It is also noticing that the presences of Cl−, NO3−, SO42−, and fluor were observed. The presences of these ions can be due to contaminants on the surface, which were not detected by XPS because its sensitivity is lower than dynamic or ToF-SIMS.
Fig. 6 TOF-SIMS spectra with the m/z ranges of (a) 0–200, (b) 200–400, (c) 400–600, and (d) 600–800 acquired by negative mode of ZnO thin films grown with (red) and without (blue) O2 gas. |
A high resolution lateral mapping by HIM-SIMS of the 18O and 16O atoms, and respective overlaps of (18O + 16O), on Fig. S1 (in the ESI†) show a uniform distribution of both atoms among the structure of the grains, depicted in Fig. 1, of the polycrystalline ZnO thin films synthesized with and without 1 s of oxygen gas pulsing for each ALD loop.
For ZnO film growth using DEZ and water precursors, the reaction mechanism usually follows the reactions (R2a) and (R3). If all hydroxyl groups are linked to MEZ groups after the DEZ saturation, the mass change during the water pulse should be negative as the ethyl ligands are replaced by much lighter hydroxyl groups via the ligand-exchange reaction. For the first 266 cycles of growth (about 2700 s of growth duration), the reaction mechanism follows reactions (R2a) and (R3), as shown in Fig. 3(b) and (c). After that, the non-negative mass change during the water pulse was observed (Fig. 3(c) and (d)). This behavior was also reported by Yousfi and co-workers.20 It suggests that the reaction mechanism postulated in equations (R2a) and (R3) does not completely reflect the nature of the growth process. It is possible that after reactions (R2a) and (R2b) can also occur, thus leading to the presence of bare zinc atoms in addition to the monoethyl zinc on the surface.18,35 These bare zinc atoms can adsorb more water onto the surface and counterbalance the mass change due to the ligand-exchange reaction. T. Weckman and K. Laasonen proposed this explanation based on density functional theory study on the atomic layer deposition of zinc oxide and supported by experimental results from other publications.18,35 The surface ethyl-ligand elimination during the water pulse is incomplete, resulting in persisting ethyl ligands on the surface after the water pulse has ended. The fraction of these persisting ligands is strongly dependent on temperature, suggesting that there is a kinetic barrier to ligand elimination by water. At elevated temperatures the kinetic barrier for the ligand removal is overcome, and the number of persisting ligands can be expected to be reduced as a function of temperature. As the surface becomes rapidly saturated with monoethyl zinc, the adsorption of diethyl zinc slows and the monoethyl zinc slowly reacts with the remaining hydroxyl groups into bare zinc atoms. In addition, the effect of surface morphology on the ligand-exchange reactions have been theoretically studied.35 Their calculation suggested that the initial ligand-exchange reactions are preferred on the planar surface over the step surface. Yousfi et al.20 also discussed that a film crystal structure might have invoked the change of growth mechanisms due to polar (002) and non-polar (100) planes. Therefore, as the film grew the film structure and morphology can be changed. These may contribute to the change of the reaction mechanisms.
In the presence of oxygen gas during the ALD growth, the growth characteristic is completely different from the one grown by DEZ and water precursors only, as shown in Fig. 4. Ignatyev et al.36 and Rienstra-Kiracofe et al.37 demonstrated that the monoethyl radical could react to O2 in the event of the additional presence of gaseous O2 during ALD processes to form the ethylperoxy radical:
C2H5˙ + O2 → CH3CH2OO˙ | (R4) |
I–O–Zn–C2H5 + O2 → I–O–Zn–C2H5O2 | (R2c) |
A higher intensity of carbon and hydrogen by D-SIMS in the film grown with O2 compared to the film grown without O2 and particularly the significant detection of ZnxOyC2H5− ion by TOF-SIMS supports our hypothesis. As a result, the formation of ethylperoxy groups induces the positive mass change during the gaseous O2 pulse. Possibly, reactions (R2b) and (R2c) occur simultaneously, leading to the presence of zinc atoms and the formation of I–O–Zn–C2H5O2. As pointed out by D. H. Ehhalt et al.,38 the ethylperoxy radical is polar, relatively long-lived and water-reactive. This form of Criegee intermediate is known to have a very high reactivity with water vapor to form mainly alkylhydroperoxides.39 L. Sheps et al. also determined the additional minor formation of aldehydes and carboxylic acids.40 Hence, these ethylperoxy radical formations would cause more adsorption of water during the water pulse in the ALD cycle to form mainly ethylhydroperoxide chains, resulting in the positive mass change during the water pulse, which is consistent to the in situ QCM data (Fig. 4). A quantitative estimation of the incorporation of oxygen atoms from H2O and O2 by the D-SIMS measurements of the 18O/16O ratio in Fig. 5 would deserve a more complete analysis of the final end products reaction rates of the Criegee intermediates with water during growth of the film with steric interactions to consider and not treated in this work.
Footnote |
† Electronic supplementary information (ESI) available: Additional information with complementary data concerning XRD spectra, the ZnO thin film grown with and without oxygen gas at different deposition temperatures, electrical resistivity of ZnO thin films, and HIM–SIMS mapping. See DOI: 10.1039/d0tc05439a |
This journal is © The Royal Society of Chemistry 2021 |