Atomic layer deposition of zirconium oxide thin films

Abstract

In this work, we studied an atomic layer deposition (ALD) process of ZrO₂ with the precursors of tetrakis(dimethylamido)zirconium(IV) and water. We investigated the growth characteristics and mechanism of the ALD ZrO₂ in the temperature range of 50–275 °C. Furthermore, the evolutions of film thickness and morphology were studied and discussed. It was found that the growth rate of ZrO₂ decreased almost linearly with the increasing temperature from ∼1.81 Å/cycle at 50 °C to ∼0.8 Å/cycle at 225 °C. Interestingly, it was revealed that the growth of ZrO₂ films ceased after a certain number of ALD cycles at a temperature higher than 250 °C. We also verified that the crystallinity of ZrO₂ evolved with deposition temperature from amorphous to crystalline phase. In addition, the wettability of ZrO₂ films was studied, showing a hydrophobic nature.

Introduction

Zirconium oxide (ZrO₂) is an attractive material in many applications, like ceramics production and optical devices, because of its excellent thermal, mechanical, optical, electrical, and catalytic properties [1, 2, 3, 4]. ZrO₂ has three crystalline structures: monoclinic (below 1170 °C), tetragonal (1170–2370 °C), and cubic (above 2370 °C) [2]. The monoclinic phase remains stable at room temperature, and the monoclinic-to-tetragonal and tetragonal-to-cubic phase transitions may occur at about 1200 °C and 2375 °C, respectively [5]. The density of ZrO₂ also varies with different crystalline structures, i.e., 5.89, 6.16, and 6.27 g/cm³ for monoclinic, tetragonal, and cubic phases, respectively [3].

To date, ZrO₂ thin films have already been prepared by conventional physical vapor deposition (PVD) and chemical vapor deposition (CVD). A PVD process, such as laser pulse ablation [6] and sputtering [7], enables to control planar film growth. However, they are unable to produce conformal films and also cannot precisely control the thickness of the deposited films. A CVD process typically needs high temperature (>800 °C) and can generate uniform and high-quality films. However, CVD is also not skilled at producing conformal films. Previously, ZrO₂ films have been synthesized via CVD processes using various precursors, such as chloride precursor [8], alkoxide precursors [9], and β-diketonate precursors [10, 11, 12]. Compared with PVD and CVD processes, atomic layer deposition (ALD) is an alternative technique enabling excellent uniform and conformal films over any shaped substrates at low temperatures, featuring its film controllability at the atomic level [13, 14, 15]. Owing to the many applications of ZrO₂, it has been developed via some ALD processes. There were Zr[OC(CH₃)₃]₄ [16], ZrI₄ [17], and ZrCl₄ [18, 19] studied as Zr precursors. In these studies, H₂O was widely used as the oxygen precursor. Precursors take a critical role in determining the resultant ZrO₂ films such as growth per cycle (GPC), crystallinity, and temperature range for deposition. Each precursor has advantages and disadvantages. For example, ZrCl₄ is a typical Zr precursor and it has a high thermal stability (∼500 °C) [20], but it suffers from low volatility and needs high evaporation temperature (165 °C) for ALD processes [21]. In addition, there is HCl generated as the byproduct during ALD processes using ZrCl₄ and H₂O, posing risk of corrosion [21] and degradation of the film quality [22]. Among all the Zr precursors investigated so far, tetrakis(dimethylamido)zirconium(IV) [Zr(NMe₂)₄ or TDMA-Zr] has shown some benefits over ZrCl₄, including high volatility, good thermal stability, and high reactivity to surface ligands (e.g., –OH) [23, 24, 25]. Liu et al. deposited ZrO₂ thin films using TDMA-Zr and H₂O for depositing ZrO₂ films on nitrogen-doped carbon nanotubes (N-CNTs) [26] and graphene nanosheets (GNS) [27]. The processes in both studies are based on the following two half reactions, as proposed by Hausmann [23]:

(1A)$$\scale 90%{├{\rm{OH}} + {\rm{Zr}}{\left( {{\rm{NM}}{{\rm{e}}_2}} \right)_4}\left( {\rm{g}} \right) \to ├{\rm{O}}–{\rm{Zr}}{\left( {{\rm{NM}}{{\rm{e}}_2}} \right)_3} + {\rm{HNM}}{{\rm{e}}_{\rm{2}}}\left( {\rm{g}} \right)\quad ,$$

(1B)$$\eqalign{\scale 89%{├{\rm{O}}–{\rm{Ze}}{\left( {{\rm{NM}}{{\rm{e}}_2}} \right)_3} + 3{{\rm{H}}_{\rm{2}}}{\rm{O}}\left( {\rm{g}} \right) \to ├{\rm{O}}–{\rm{Zr}}–{\left( {{\rm{OH}}} \right)_{\rm{3}}} +\cr {\rm{3HNM}}_{{\rm{E}}{\rm{2}}}\left( {\rm{g}} \right)\quad ,$$

where “├” denotes the substrate surface and “(g)” refers to a gaseous phase. The results showed that uniform and conformal films can be developed at temperature as low as 100 °C, and the GPC and crystallinity of ZrO₂ thin film experienced a strong dependence on the deposition temperatures. However, the surface reactions between the precursors were not fully discussed, which are essential for a better understanding of the ALD process of ZrO₂.

Stimulated by the tremendous capability for uniform and conformal coatings at low temperature and aimed at further exploring the ALD process using TDMA-Zr and H₂O as precursors, we recently conducted a systematic study, featuring the utilization of an in situ quartz crystal microbalance (QCM) in a wide temperature range of 50–275 °C. The ALD process was further applied on nitrogen-doped graphene nanosheets (N-GNS) and N-CNTs for investigating film evolution with temperatures and ALD cycles. In addition, the wettability of ZrO₂ films was investigated for different film thicknesses and different deposition temperatures.

Results and discussion

We first conducted a series of studies to optimize ALD dosing and purging parameters at 100 °C, as shown in Figs. 1(a) and 1(b). With the fixed dosing time of TDMA-Zr (0.03 s) and purging time (10 s), the dosing time of H₂O was increased from 0 to 0.025 s, and it was observed that the mass gain would maintain the same value as the dosing time of H₂O reached 0.005 s or longer [Fig. 1(a)], which indicated the saturation growth of ZrO₂; thus, we selected a value of 0.015 s for the dosing time of H₂O. Similarly, we selected the optimal dosing time of TDMA-Zr as 0.03 s [Fig. 1(b)]. In addition, in situ QCM measurements were conducted to investigate the growth of ALD-ZrO₂ at various deposition temperatures. Figs. 1(c) and 1(d) illustrate the time-resolved mass changes measured by in situ QCM during 100 ALD cycles using TDMA-Zr and H₂O as precursors between 50 and 275 °C. Prior to the ZrO₂ ALD, an ALD Al₂O₃ film was deposited on the QCM crystal surface alternately using TMA and H₂O as precursors with a timing sequence of 0.015–5–0.015–5 s to establish an uniform starting surface. During the deposition process, the ALD ZrO₂ GPC gradually decreased and stabilized to a constant value after the first ∼20 cycles (0–400 s), and the larger GPC in the first 20 cycles were due to the effect of the predeposited Al₂O₃ film. In addition, the mass gain per cycle in the stable regime changes with temperatures, which decreased almost linearly from 100.28 ng/cm² at 50 °C to 40.66 ng/cm² at 225 °C. However, no mass gain can be observed after certain cycles at 250 and 275 °C. This might be related to the precursor decomposition of TDMA-Zr and the exactly underlying reason will be investigated in future studies.

Figure 1: Optimization of the dosing time of precursors at 100 °C: (a) the effect of dosing time of H₂O on the mass gain, (b) the effect of dosing time of TDMA-Zr on the mass gain. In situ QCM measurements of ALD ZrO₂ with 100 cycles: (c) using the timing sequence of 0.03–30–0.015–30 s at 50 and 75 °C, (d) using the timing sequence of 0.03–10–0.015–10 s in the range of 100–275 °C.

Fig. 2(a) demonstrates the thickness variation of ZrO₂ films during 100 ALD cycles from 50 to 275 °C. Supposed that the deposited films are crystalline, having a density of 6.16 g/cm³, we then could calculate the thickness using the QCM data (ng/cm²). As shown in Fig. 2(a), the thickness of ZrO₂ films increases almost as a straight line within 100 cycles at temperatures lower than 225 °C, indicating that at each temperature the GPC is constant. However, the thickness of ZrO₂ films at 250 and 275 °C remains unchanged after 80 and 20 cycles, respectively, which means the growth of the films terminated. It was reported that the maximum ALD temperature of TDMA-Zr is 250 °C [23]. Therefore, decomposition of TDMA-Zr might be responsible to the growth cessation at the high temperatures of 250 and 275 °C. The GPC of ALD-ZrO₂ changes with temperature, as shown in Fig. 2(b). The GPC is calculated to be 1.81, 1.84, 1.60, 1.21, 0.67, and 0.66 Å/cycle at the deposition temperatures of 50, 75, 100, 150, 200, and 225 °C, respectively. At low temperatures (i.e., 50 and 75 °C), the GPCs are comparable and then significantly decreases with the increasing temperatures. The underlying reasons will be discussed later.

Figure 2: In situ QCM measurements of ALD ZrO₂ at various temperatures using the timing sequence of 0.03–30–0.015–30 s at 50 and 75 °C, and 0.03–10–0.015–10 s in the range of 100–275 °C: (a) The thicknesses of ZrO₂ film versus ALD cycle number during 100 ALD cycles and (b) the GPCs of ZrO₂ films versus temperature.

Fig. 3 shows an enlarged view of consecutive ZrO₂ ALD cycles in the stable growth regime. Inspired by our previous ALD studies [28, 29, 30], the two half reactions can be rewritten as follow:

(2A)$$╟{\left( {{\rm{OH}}} \right)_x} + {\rm{Zr}}{\left( {{\rm{NM}}{{\rm{e}}_{\rm{2}}}} \right)_{\rm{4}}}\left( {\rm{g}} \right) \to ╟{{\rm{O}}_x}–{\rm{Zr}}{\left( {{\rm{NM}}{{\rm{e}}_{\rm{2}}}} \right)_{{\rm{(4}} - x)}} + x{\rm{H}}\left( {{\rm{NM}}{{\rm{e}}_{\rm{2}}}} \right)\left( {\rm{g}} \right)\quad ,$$

(2B)$$╟{{\rm{O}}_x}–{\rm{Zr}}{\left( {{\rm{NM}}{{\rm{e}}_{\rm{2}}}} \right)_{(4 - x)}} + 2{{\rm{H}}_{\rm{2}}}{\rm{O}}\left( {\rm{g}} \right) \to ╟{{\rm{O}}_2}–{\rm{Zr}}–{\left( {{\rm{OH}}} \right)_x} + \left( {4 - x} \right){\rm{H}}\left( {{\rm{NM}}{{\rm{e}}_{\rm{2}}}} \right)\left( {\rm{g}} \right)\quad ,$$

where x is between 0 and 4, indicating the number of reactive surface (–OH) groups. In Eq. (2A), TDMA-Zr reacts with x –OH while liberating x of (–NMe₂) into gaseous C₂H₇N, and the surface mass gain can be expressed as m ₁ = 267 − 45x. In Eq. (2B), H₂O reacts with (4 − x) (–NMe₂) surface ligands which are then replaced by x –OH groups, and the surface mass gain can be expressed as m ₂ = 45x − 144. Thus, the total mass gain for one ZrO₂ ALD cycle is the sum of the mass gain in the two equations, i.e., Δm = m ₁ + m ₂ = 123, which is exactly the molar mass of ZrO₂. These equations assume that the –OH groups are the reactive species for the chemisorption of Zr precursor. Considering the mass gain m ₁ in Eq. (2A) and the total mass gain Δm, we can combine them in Eq. (3):

(3)$$R = {{\Delta m} / {{m_1} = {{123} / {\left( {267 - 45x} \right)}}}}$$

where the ratio R reveals the number x of surface (–OH) groups reacting with each TDMA-Zr molecule. For instance, at 75 °C, R = 1.34, so that x = 3.9, implying that 3.9/4.0 (or ∼97.5%) of the (–NMe₂) ligands are released as gaseous C₂H₇N, leaving only ∼2.5% on ZrO_x surface following TDMA-Zr reaction [Eq. (2A)]. The R and x values decrease with increasing temperatures from 75 to 275 °C, which means that a greater fraction of the (–NMe₂) ligands from the same TDMA-Zr were chemisorbed on the surface or, conversely, that the coverage of the surface (–OH) groups was lower at the higher temperatures, either due to the desorption of TDMA-Zr or the dehydration of –OH sites. The variation of x with deposition temperature is illustrated in Fig. 3. It can be seen that R and x values of ALD ZrO₂ at 50 °C are lower than that at 75 °C (Fig. 4). The abnormal behavior may originate from the low value of m ₂, i.e., the low mass gain after H₂O dosing, as shown in Fig. 3(a), which may be due to the low reactivity of H₂O or the physical absorption of H₂O on many reactive sites at 50 °C, resulting in a lower GPC than that of 75 °C.

Figure 3: Enlarged view of consecutive ALD cycles of ZrO₂ deposition in the regime of steady-state growth at (a) 50, (b) 75, (c) 100, (d) 150, (e) 200, (f) 225, (g) 250, and (h) 275 °C (m ₁, m ₂, and Δm are illustrated in the text).

Figure 4: The number x (0 ≤ x ≤ 4) of surface (–OH) groups reacting with TDMA-Zr at different temperatures.

To verify the GPC of ZrO₂ at different temperatures, ZrO₂ films were deposited on N-GNS at 75, 150, and 250 °C with various numbers of ALD cycles, and the ZrO₂–N-GNS composites were characterized by scanning electron microscopy (SEM) (Fig. 5). Low-magnification SEM images were inset in Figs. 5(a)–5(i), which indicated that ZrO₂ thin films were uniformly deposited onto the N-GNS. At 75 °C, the graphene wrinkles were coated by ZrO₂ with 100, 200, and 300 ALD cycles [see Figs. 5(a)–5(c)]. It can be seen that the ZrO₂ coatings are very conformal. After 100 ALD cycles, the thickness of the coated N-GNS is measured to be 40 ± 0.5 nm. With the increasing ALD cycles, the coatings become thicker, i.e., the thickness rises to 77 ± 0.5 nm, 115 ± 0.5 nm after 200 and 300 cycles, respectively. Furthermore, in our previous study [30], the pristine N-GNS was measured to be ∼3 nm in thickness; therefore, the GPCs of the ALD-ZrO₂ were calculated as 1.9 Å/cycle at 75 °C, which is a little bit higher but very close to that of QCM data. The calculation is based on the following equation: GPC = (wrinkle thickness of coated N-GNS—wrinkle thickness of pristine N-NGS)/(2 × cycle numbers) [31]. The growth of ALD-ZrO₂ on N-GNS with different cycles at 150 °C exhibits the similar trend as that at 75 °C, as shown in Figs. 5(d)–5(f), and the GPC is calculated as ∼1.3 Å/cycle. To confirm the terminating growth of ALD-ZrO₂ after 80 cycles at 250 °C, the coating was also performed for 25, 75, and 100 cycles, as shown in Figs. 5(g)–5(i). From the thickness variation of coated N-GNS with 25 and 75 cycles, the GPC is estimated as ∼0.5 Å/cycle. Compared with that of 75 cycles, however, the coated N-GNS of 100 cycles almost maintains the same thickness. This corroborates the QCM data, indicating that the growth of ALD-ZrO₂ indeed terminated after 80 cycles at 250 °C.

Figure 5: SEM images of ZrO₂–N-GNS composites prepared at (a, b, c) 75 °C (d, e, f) 150 °C, and (g, h, i) 250 °C with (a, d) 100 ALD cycles (b, e) 200 ALD cycles, (c, f) 300 ALD cycles, (g) 25 ALD cycles, (h) 75 ALD cycles, and (i) 100 ALD cycles.

Atomic force microscope (AFM) images of ZrO₂ coating on glass substrate are shown in Fig. 6. According to the QCM results, the GPCs are ∼1.8, 1.2, and 0.7 Å/cycle at 50, 150, and 225 °C, respectively. To facilitate the comparison of films with a similar thickness, we deposited ZrO₂ films for 10 cycles at 50 °C, 15 cycles at 150 °C, and 25 cycles at 225 °C, having a similar thickness of ∼1.8 nm. We also deposited films for 300 cycles at 50 °C, 400 cycles at 150 °C, and 700 cycles at 225 °C, having a comparable thickness of ∼50 nm. It was observed that deposition temperature and film thickness have some remarkable effects on the film topography. At each temperature, high ALD cycles (e.g., ∼50 nm-thick films) resulted in more uniform films. In the cases of ∼1.8 nm-thick films, the 10-cycle film at 50 °C might not have fully covered the substrate. Interestingly, the 15-cycle film deposited at 150 °C showed many nanorod-like structures, while the 25-cycle film deposited at 250 °C showed a more uniform structure. The nanorod-like structure might be related to phase transition, i.e., the existence of sparse nanoparticles before the full formation of nanofilms, which can be observed in others’ ALD studies [26, 27, 31, 32, 33]. One can also find that at each temperature, the thicker coating generally has a higher roughness than the thin one.

Figure 6: AFM images of ALD ZrO₂ films deposited on glass substrates with (a) 10 cycles at 50 °C, (b) 15 cycles at 150 °C, and (c) 25 cycles at 225 °C, (d) 300 cycles at 50 °C, (e) 400 cycles at 150 °C, and (f) 700 cycles at 225 °C. The average roughness is also presented in each figure. The scanned area is 5 × 5 µm.

We further clarified the crystallinity of ZrO₂ films with deposition temperature using synchrotron-based X-ray diffraction (XRD). ZrO₂ films were deposited on N-CNTs for 100 ALD cycle at 75, 100, 150, 200, and 250 °C. The XRD patterns of the pristine N-CNTs and the resultant ZrO₂–N-CNTs composites are shown in Fig. 7. It can be seen that all samples at these temperatures exhibit three peaks at 30°, 50°, and 62°, which are indexed as (002), (100), and (004) planes of graphite, respectively (JCPDS PDF No. 08-0415). For the ZrO₂-CNTs prepared at 150, 200, 250 °C, the peaks located at 36° correspond well to the plane (101) of tetragonal ZrO₂ (JCPDS PDF No. 50-1089). This is consistent to Liu’s work [26]. Because the deposited ZrO₂ films were pretty thin, other XRD peaks of ZrO₂ are not identifiable. At low deposition temperatures of 75 and 100 °C, the (101) peak becomes much broader. This implies the amorphous structure of ZrO₂.

Figure 7: XRD patterns of (a) the pristine N-CNTs, and ZrO₂-coated N-CNTs with 100 ALD cycles prepared at (b) 75, (c) 100, (d) 150, (e) 200, and (f) 250 °C.

The tetragonal ZrO₂ typically exists at high temperature (1170–2370 °C) [2], whereas in our study, this phase was not only prepared between 150 and 250 °C but also was stable at room temperature. The existence of tetragonal ZrO₂ at low temperatures was found in previous reports [34, 35, 36, 37]. As pointed out in previous studies, this phenomenon may be related to the grain size. Small ZrO₂ size and high specific surface area might have provided sufficient surface energy to stabilize the metastable tetragonal phase at low temperatures.

Furthermore, we investigated water contact angles of ZrO₂ films, as shown in Fig. 8. As we can see, all the water contact angles of the samples are larger than 90° regardless of the thickness and deposition temperatures, indicating that the ZrO₂ thin films are hydrophobic. For the same thickness of ZrO₂ thin films, the water contact angles decreased with the increasing temperatures. In other words, the ZrO₂ thin films are more hydrophobic at lower temperatures. At each temperature, the thicker films would result in larger water contact angles, that is, the thicker films are more hydrophobic than the thinner ones at 50, 150, and 225 °C. The surface wettability is related to surface microstructures and surface functional groups [38, 39].

Figure 8: Water contact angles of ZrO₂ films deposited with various ALD cycles at different temperatures.

Conclusions

In this work, we systematically studied the growth characteristics of ALD ZrO₂ using TDMA-Zr and H₂O as precursors. The study confirmed that there has a phase evolution for ALD ZrO₂ films with temperature, changing from an amorphous phase at temperatures lower than 100 °C to a crystalline phase at temperatures higher than 150 °C. At the same time, it was found that the GPCs of ALD ZrO₂ films are dependent to deposition temperatures. The GPC could be precisely controlled at the atomic level for various growth temperatures, changing from ∼1.8 at 50 °C to 0.7 Å/cycle at 225 °C. Furthermore, it was found that the deposited ZrO₂ films showed a hydrophobic nature.

Experimental section

QCM of ZrO₂

In situ QCM can be used to track the mass variation of deposited ZrO₂ films. The measurements were conducted at 50, 75, 100, 150, 200, 225, 250, and 275 °C using TDMA-Zr and H₂O as precursors in an ALD reactor (Savannah 200, Cambridge Nanotech Inc., Cambridge, MA). 6 MHz quartz crystals (#750-1059-G10, Inficon, East Syracuse, NY) were used. Prior to the QCM measurements at each deposition temperature, 50-cycle ALD of Al₂O₃ was performed using trimethylaluminum (Al(CH₃)₃ or TMA) and H₂O as precursors to maintain a consistent starting surface. Prior to ALD deposition, TDMA-Zr was heated to 75 °C for supplying sufficient precursor vapor, whereas H₂O was kept at room temperature. The delivery pipelines and valve manifold were kept at 90 °C to prevent the condensation of Zr precursor, and the exhaust pipelines was heated to 150 °C for removing any excessive precursor vapors and other by-products. High purity Argon (Ar) gas was used as the carrier gas with a flow rate of 20 sccm, and the ALD reactor was kept at a low-level base pressure of 0.3–0.4 torr by a vacuum pump (Pascal 2021C2, Adixen, Jefferson Hills, PA). One ALD cycle was performed with four steps in the temperature range of 100–275 °C: (i) a 0.03-s dose of Zr(NMe₂)₄, (ii) a 10-s purge of Ar gas to remove excessive Zr(NMe₂)₄ and by-products, (iii) a 0.015 s dose of H₂O, and (iv) a 10-s purge of Ar gas to remove excessive H₂O and by-products. At low temperatures of 50 and 75 °C, the purging time was increased to 30 s to fully purge the ALD system while the dosing times were kept as the same.

Preparation and characterization of ZrO₂ films

ZrO₂ films were deposited on N-CNTs, N-GNS, and glass substrates for characterization. Synchrotron-based XRD was performed to characterize the crystallinity of the ZrO₂–N-CNTs composites, prepared at 75, 100, 150, 200, and 250 °C. The XRD measurements were conducted using an X-ray wavelength of 0.4336 Å at the beamline 13 BM-C of the Advanced Photon Source (APS) at Argonne National Laboratory (ANL; Illinois). Scanning electron microscopy (SEM) (XL30, Philips FEI, Hillsboro, OR) was used to characterize the thickness changes of ZrO₂-coated N-GNS with various ALD cycles at different temperatures. AFM (Dimension Icon, Bruker Nano Surfaces, Billerica, MA) was used to measure the roughness of the thin films. Water contact angle test was employed to determine the wettability of ZrO₂ films on glass substrates via Contact Angle Measuring System (OCA 15, DataPhysics Instruments GmbH, Filderstadt, Germany).