Research Article www.acsami.org
Comparison of the Atomic Layer Deposition of Tantalum Oxide Thin Films Using Ta(NtBu)(NEt2)3, Ta(NtBu)(NEt2)2Cp, and H2O Seul Ji Song,† Taehyung Park,† Kyung Jean Yoon,† Jung Ho Yoon,† Dae Eun Kwon,† Wontae Noh,‡ Clement Lansalot-Matras,‡ Satoko Gatineau,‡ Han-Koo Lee,§ Sanjeev Gautam,∥ Deok-Yong Cho,⊥ Sang Woon Lee,*,# and Cheol Seong Hwang*,† †
Department of Materials Science and Engineering and Inter-University Semiconductor Research Center, Seoul National University, Seoul 151-744, Korea ‡ Air Liquide Laboratories Korea, Suite 176, Yonsei Engineering Research Park, 50 Yonsei-ro, Seodaemun-gu, Seoul 120-749, Korea § Pohang Accelerator Laboratory, Pohang 37673, Korea ∥ Dr. S.S.Bhatnagar University Institute of Chemical Engineering and Technology, Panjab University, Chandigarh 160 014, India ⊥ Department of Physics, Chonbuk National University, Jeonju 54896, Korea # Department of Physics and Division of Energy Systems Research, Ajou University, Suwon 16499, Korea S Supporting Information *
ABSTRACT: The growth characteristics of Ta2O5 thin films by atomic layer deposition (ALD) were examined using Ta(NtBu)(NEt2)3 (TBTDET) and Ta(NtBu)(NEt2)2Cp (TBDETCp) as Ta-precursors, where tBu, Et, and Cp represent tert-butyl, ethyl, and cyclopentadienyl groups, respectively, along with water vapor as oxygen source. The grown Ta2O5 films were amorphous with very smooth surface morphology for both the Ta-precursors. The saturated ALD growth rates of Ta2O5 films were 0.77 Å cycle−1 at 250 °C and 0.67 Å cycle−1 at 300 °C using TBTDET and TBDETCp precursors, respectively. The thermal decomposition of the amido ligand (NEt2) limited the ALD process temperature below 275 °C for TBTDET precursor. However, the ALD temperature window could be extended up to 325 °C due to a strong Ta−Cp bond for the TBDETCp precursor. Because of the improved thermal stability of TBDETCp precursor, excellent nonuniformity of ∼2% in 200 mm wafer could be achieved with a step coverage of ∼90% in a deep hole structure (aspect ratio 5:1) which is promising for 3-dimensional architecture to form high density memories. Nonetheless, a rather high concentration (∼7 at. %) of carbon impurities was incorporated into the Ta2O5 film using TBDETCp, which was possibly due to readsorption of dissociated ligands as small organic molecules in the growth of Ta2O5 film by ALD. Despite the presence of high carbon concentration which might be an origin of large leakage current under electric fields, the Ta2O5 film using TBDETCp showed a promising resistive switching performance with an endurance cycle as high as ∼17 500 for resistance switching random access memory application. The optical refractive index of the deposited Ta2O5 films was 2.1−2.2 at 632.8 nm using both the Taprecursors, and indirect optical band gap was estimated to be ∼4.1 eV for both the cases. KEYWORDS: atomic layer deposition (ALD), Ta2O5, Ta(NtBu)(NEt2)3 (TBTDET), Ta(NtBu)(NEt2)2Cp (TBDETCp), resistive switching
■
INTRODUCTION In advanced semiconductor device technologies, adoption of three-dimensional (3-D) structures appears to be the only feasible solution to many challenging issues which have arisen due to the continued shrinkage in device geometry. Stacked capacitor in dynamic random access memory (DRAM) represents a conventional example of such a trend, and very recent development of vertical NAND (V-NAND) flash memory is another representative example of 3-D structure of highly advanced semiconductor memory technology.1,2 Vertical resistance switching random access memory (V-ReRAM) is © XXXX American Chemical Society
also now actively researched as a possible replacement or successor of V-NAND.3,4 In this regard, atomic layer deposition (ALD) process is also gaining appreciable attention in semiconductor device fabrication because it appears to be the only feasible method to grow various ultrathin (≪10 nm) films with high quality, atomic-scale thickness controllability, and sufficient conformality over a large wafer (300 mm diameter) Received: September 14, 2016 Accepted: December 12, 2016 Published: December 12, 2016 A
DOI: 10.1021/acsami.6b11613 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
ACS Applied Materials & Interfaces
■
having extreme 3-D structures.5−7 A typical ALD process involves exposure of the substrate surface to two or more different precursors alternatingly, where the precursors are kept separated by purging to prevent unwanted gas phase reactions. The precursor molecules are then chemically adsorbed on the surface via a chemisorption mechanism in a self-limiting manner. This indicates that the precursor chemistry plays decisive roles in determining not only the growth behavior but also electrical and chemical properties of the grown films. Therefore, adopting suitable precursors is of utmost importance for feasible ALD surface reactions and film quality. Tantalum oxide (Ta2O5) has been used extensively as high-k dielectrics in the capacitive element of DRAM as well as antireflection coatings in solar cell because of its high dielectric constant (∼15 for amorphous, ∼50 for crystalline film) and high refractive index (>2).8,9 Besides, Ta-based oxide has also received considerable attention recently for ReRAM application due to its superior device performance, endurance, and switching speed in comparison to other transition metal oxides.10−12 The ALD processes of Ta2O5 thin film have been developed using several precursors, such as halides,13,14 alkoxides,15,16 and alkylamide17−19 type materials. The metal halide precursor is widely used in conjunction with H2O as the oxygen source; however, due to formation of corrosive byproducts (e.g., HCl) in such a process, damage to hardware of the ALD tool became a major issue. On the other hand, the other metal−organic Ta-precursors, such as alkoxides (Ta(OEt)5) and the alkylamides (Ta(NMe2)5, Ta(NtPn) (NEt2)3), where Et, Me, and tPn refer to ethyl, methyl, and tert-pentyl, respectively, show limited thermal stability and low reactivity toward H2O. The lower reactivity toward H2O can be improved by increasing the deposition temperature, but such increase leads to thermal decomposition of the metal−organic Taprecursors at higher temperature. Adopting other oxygen sources, such as O3, which has a higher oxidation activity than H2O may be an option, but the ALD mechanism involving O3 is less clear compared with that for H2O-based process in general. In addition, O3 is an excessively strong oxidizer for Si substrate and result in a thick interfacial oxide layer, which is undesirable if the Ta2O5 film is intended to be used as a high-k gate dielectric layer in highly scaled field-effect transistor. Therefore, it is necessary to develop a new Ta-precursor, which has high thermal stability as well as high reactivity toward H2O for the growth of Ta2O5 films by ALD. In this work, thermal ALD reactions of Ta2O5 film using two types of the imido−amido type precursors, viz. tert-butylimidotris(diethylamido)tantalum (Ta(NtBu)(NEt2)3, TBTDET) and tert-butylimidobis(diethylamido)cyclopentadienyltantalum (Ta(NtBu)(NEt2)2Cp, TBDETCp), along with H2O were studied in detail to elucidate the ALD reaction mechanisms and their influences on the film growth behavior and properties. The growth behavior and mechanism of Ta2O5 ALD film using TBTDET and H2O as the ALD precursors have been extensively studied.15 Nevertheless, detailed growth result can be dependent on the specific reactor geometry and precursor conditions. In this work, the ALD behaviors of the films with the TBTDET and H2O were carefully examined using the identical reactor as for the new Ta-precursor, TBDETCp, which serve as the reference for evaluating the ALD performance of the new process.
Research Article
EXPERIMENTAL SECTION
Ta2O5 films were grown by a shower-head-type ALD system with a 200 mm scale wafer reactor (CN-1 Co., Plus-200). The deposition was carried out at growth temperatures (Tg) between 150 and 350 °C, which was estimated by a thermocouple-embedded Si wafer at a typical ALD pressure of 2.3 Torr. The substrates used were Si wafers containing native oxide (∼1 nm thickness, called native Si) on the surface and hydrofluoric acid (HF, 5%) dipped Si wafer (called HFdipped Si). The ALD reactor walls were heated to 150 °C, and all injection gas lines were maintained at 100 °C to prevent precursor condensation. TBTDET (UP Chemical Co.) and TBDETCp (Air Liquide Co.) were used as the Ta-precursors. The precursors were kept in a bubbler whose temperature was maintained at 100 °C, and the vaporized precursor molecules were introduced into the reaction chamber with the help of argon (Ar) carrier gas at a flow rate of 100 standard cubic centimeters per minute (sccm). The process pressure was fixed at 2.3 Torr during the ALD sequences. H2O was used as the oxygen source and was vaporized at 4 °C and supplied into the ALD reactor via Ar carrier gas at a flow rate of 1000 sccm in order to maintain a constant chamber pressure. Energy dispersive X-ray fluorescent spectroscopy (XRF, Themoscientific ARL Quant’X) was used to estimate the layer density (LD) of Ta atoms in Ta2O5 films. The optical properties of the deposited Ta2O5 films such as refractive index and optical band gap (Eg) were estimated by the spectroscopic ellipsometer (SE, J.A. Woollam ESM300). The film thickness and density were evaluated by X-ray reflectivity (XRR), and the crystal structure was investigated by selected area electron diffraction in transmission electron microscopy (SAED/TEM, JEOL, JEM-2100F) and grazing-angle incidence X-ray diffraction (GIXRD, PANalytical X’pert Pro) with Cu Kα X-ray radiation. The surface morphology of the film was imaged by atomic force microscopy (AFM, JEOL, JSPM-5200) and field-emission scanning electron microscopy (SEM, Hitachi, S-4800). To investigate the chemical binding state of Ta2O5 films, X-ray photoelectron spectroscopy (XPS, ThermoVG SIGMAPROBE) was employed using monochromatic Al Kα radiation. The depth profiling of the grown film was carried out using Auger electron spectroscopy (AES, PerkinElmer model 660). To examine the bonding structures in the film samples, soft X-ray absorption spectroscopy (XAS) at C K-, N K-, and O Kedges was performed at the 2A and 10D beamlines in Pohang Light Source. The electrical properties of the Ta2O5 film was examined using the Pt/Ta2O5/TiN structure, where the Pt (top) and TiN (bottom) electrodes were deposited by physical vapor deposition processes [electron beam evaporation through metal shadow mask with a hole diameter of 0.3 mm for top electrode (50 nm thickness), and dc magnetron sputtering for bottom electrode (50 nm thickness)] on SiO2/Si substrate. Resistive switching characteristics were tested with dc sweep mode using HP 4155B semiconductor parameter analyzer, and pulse switching was also performed with a closed-loop pulse switching (CLPS) mode by an in-house built gate array board for the field programming (minimum pulse width of 2 μs), where a feedback of each resistance value to the following pulse application allowed to achieve a targeted resistance value precisely. Detailed experimental conditions for the electrical tests were reported from the same group.20
■
RESULTS AND DISCUSSION Deposition Characteristics of Ta2O5 Films Grown by Using TBTDET and TBDETCp Precursors. Figures 1a and 1b show the chemical formulas of the two Ta-precursors. The chemical structure of TBDETCp precursor is similar to that of TBTDET, where one amido ligand of TBTDET is replaced by a Cp ring. TBDETCp precursor is reported to be thermally stable up to a higher temperature (400 °C) than that of TBTDET, which is ascribed to a stronger bond between Ta and Cp ring than that between the Ta and amido/imido ligands.21 The Ta2O5 ALD process using TBDETCp is likely to show similar growth characteristics to that of ALD for TBTDET, but a temperature window of self-limited ALD growth extended up B
DOI: 10.1021/acsami.6b11613 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
precursor increased with increasing pressure in the chamber until the GPC became saturated at ∼2.3 Torr. The precursor molecules and H2O concentration in the ALD reactor increased with the increasing chamber pressure. The constant GPC for the case of TBTDET suggested that the amounts of the Taprecursor and H2O vapors were sufficiently high to make the GPC saturated even at the lowest chamber pressure. However, the increasing GPC with increasing chamber pressure up to 2.3 Torr for the case of TBDETCp suggested that the amount of Ta-precursor was not sufficient to ensure saturation of GPC, which could be either due to lower vapor pressure of TBDETCp (0.4 Torr at 100 °C) than TBTDET (0.5 Torr at 100 °C) or relatively slower reaction speed of the TBDETCp. The vapor pressure of each precursor was provided by the material suppliers. Hence, considering the stability and reproducibility of ALD process, further ALD reactions were carried out at ∼2.3 Torr with both the Ta-precursors. Figures 2a−d show the variations of GPC of Ta2O5 films as functions of Ta-precursor pulse, Ta-precursor purge, H2O
Figure 1. Chemical formulas of (a) TBTDET and (b) TBDETCp. Changes in GPC of Ta2O5 films grown by using TBTDET (black, square) and TBDETCp (red, circle), with respect to (c) the growth temperature (Tg) and (d) the process chamber pressure with 100 sccm of Ar carrier gas.
to a higher value. Thus, using TBDETCp could be beneficial in achieving higher film quality than TBTDET. Figure 1c illustrates the changes in the growth per cycle (GPC) of Ta2O5 films deposited on native Si substrate as a function of Tg between 150 and 350 °C using TBTDET and TBDETCp precursors. The ALD sequence consisted of Ta-precursor pulse/purge/H2O pulse/purge (5 s−7 s−10 s−20 s) steps. The GPC was estimated by dividing the LD of Ta atoms by deposition cycle numbers (ncy), which was 100 for each sample. The GPC of Ta2O5 films deposited by using TBTDET precursor slightly decreased with increasing Tg from 150 to 275 °C, but the GPC was approximately maintained at a constant value of 44.5 ng cm−2 cycle−1 (0.78 Å cycle−1) between Tg of 200 and 275 °C. The Tg region, where a constant GPC was obtained, was regarded the self-limiting ALD regime. When the Tg increased further, a significant increase in the GPC of Ta2O5 films was observed which was attributed to the thermal decomposition of TBTDET precursor, which started around 250−275 °C.22,23 By contrast, the GPC of Ta2O5 films grown by TBDETCp precursor increased monotonously from 5.2 to 40.6 ng cm−2 cycle−1 with increasing Tg from 150 to 275 °C. This increase of GPC using TBDETCp precursor might be related to the stronger binding energy between Ta and the Cp ring compared with Ta-amido ligand, whose dissociation by ligand exchange reaction in ALD could be thermally activated. Although the Tg range of the constant GPC could be determined only between 275 and 325 °C, the average GPC of TBDETCp at this range was 42.3 ng cm−2 cycle−1 (0.74 Å cycle−1), which was comparable to the GPC at Tg’s of 200−275 °C using TBTDET precursor. Therefore, the appropriate temperature regimes for Ta2O5 ALD using TBTDET and TBDETCp could be 175−275 and 275−325 °C, respectively. The growth behavior with respect to changes in process pressure was investigated, as shown in Figure 1d. Each point in Figure 1d represents average GPC of Ta2O5 films (process repeated 5 times under same deposition conditions) at Tg of 275 °C to ensure that the deposition processes proceeded via ALD mechanism for both the cases. The process pressure was varied from 0.7 to 3 Torr by controlling the degree of opening of the throttle valve which was located between the ALD chamber and pump. In the case of TBTDET, the GPC of Ta2O5 film almost remained constant regardless of the process pressure. By contrast, the GPC of Ta2O5 films using TBDETCp
Figure 2. Variation of GPC of Ta2O5 films as a function of Taprecursor pulse/purge and H2O pulse/purge at substrate temperatures of (a, b) 250 °C and (c, d) 300 °C.
vapor pulse, and H2O purge time at two different Tg’s of 250 °C (Figures 2a,b) and 300 °C (Figures 2c,d). For these experiments, the ncy was fixed at 50. The error bars in Figures 2a−d indicate standard deviations of the GPC of Ta2O5 films by repeating the ALD process for five times. The very small error bars indicate that the process repeatability was very high. When the Ta-precursor pulse or purge time was varied, the H2O pulse and H2O purge time remained fixed at 10 and 15 s, respectively, at both the Tg’s. Also, the purge time was fixed at 20 s to ensure complete removal of residual precursor and the byproducts when the Ta-precursor pulse time was varied. Upon varying the purge time, on the contrary, the Ta-precursor was injected for 10 s in order to allow completely saturated surface reactions to occur. As illustrated in Figure 2a, the GPC’s of Ta2O5 films using both the Ta-precursors were saturated with >5 s of the precursor pulse time, and the GPC of Ta2O5 films changed negligibly with source purge time, which were the indications of self-limiting surface chemisorption. The GPC of the Ta2O5 films was lower when the TBDETCp precursor was used in the ALD instead of TBTDET at Tg of 250 °C due to the difference in reactivity and steric hindrance effect for the Cp ligand. The H2O pulse was also changed while the Ta-precursor pulse, and the purge time were fixed at 5 s at the Tg of 250 °C, as shown in Figure 2b. The H2O pulse and the purge time C
DOI: 10.1021/acsami.6b11613 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
TBTDET and TBDETCp precursors, respectively, from the slopes of the best-linear-fitted graphs of each data set. In order to investigate the initial growth characteristics of Ta2O5 films grown on the native Si and HF-dipped Si substrates, the changes in LD and its derivatives with respect to ncy (ncy values from 1 to 30) for both the Ta-precursors are shown in Figures 3c−f. The derivative of the LD indicated the increment in mass gain with increasing ncy. Lines in Figures 3c,d are the bestlinear-fitted graphs shown in Figures 3a,b, which represent the steady-state growth behavior of the films free from any influence of the substrates. For both the Ta-precursors, the yintercepts of LD of the best-linear fitted graphs were almost 0, as shown in Figures 3c,d, indicating that there were no significant incubation retardation or growth enhancement effects from the substrates for both the Ta-precursors. However, there were subtle differences during the first ∼5 cycles depending on the type of substrates, which can be understood from Figures 3e,f. From Figures 3c,d, it was found that the LD of the films on HF-dipped Si was generally lower than that on native Si for both the precursors up to ncy of 30. This was due to the initial retardation of nucleation on the HFdipped substrate up to ∼5 cycles, as can be understood from Figures 3e,f. By contrast, the native Si substrate showed an enhanced GPC up to ∼5 cycles for both the Ta-precursors. The initial mass gain for TBTDET was ∼110 ng cm−2 cycle−1, which was almost triple as that of the saturated GPC (∼44 ng cm−2 cycle−1). On the other hand, for TBDETCp the initial mass gain was also high (∼80 ng cm−2 cycle−1) but was only approximately double as that of its saturated value (∼38 ng cm−2 cycle−1). The higher initial enhancement of GPC on the native Si for TBTDET indicated that the amido ligand tended to be more reactive to the surface reaction sites, such as −OH, than the Cp ligand, and more amido ligand might be released during the adsorption of TBTDET with the interaction of surface −OH group than TBDETCp because one more amido ligand is included in the TBTDET. The derivative of LD became lower and saturated after five cycles. These findings are consistent with the general expectation that the surface of the native Si substrate has sufficiently high density of OH groups, which could work as the anchoring sites for the chemisorption of amido ligands within the Ta-precursors.24,25 This also suggested that the ALD mechanism using H2O as the oxygen source was related to the OH-group-mediated ligand-exchange reaction for both the precursors. The H-terminated structure of HF-dipped Si substrate retarded nucleation of the growing films. A higher Tg was necessary to achieve a saturated ALD GPC for TBDETCp due to the strong bond of Ta ligand (Ta− Cp) compared to that of TBTDET, as shown in Figures 1 and 2, and these results demonstrated that the precursor having Cp ligand was less influenced by the interaction of substrate, compared to the amido ligand, which in turn might be helpful in realizing uniformity and repeatability in the ALD process. The Cp group in TBDETCp can be retained without thermal decomposition at higher temperature, which allowed the saturated ALD behavior at a higher Tg than that for TBTDET. Nonetheless, other parts of the ligand can be thermally decomposed at such higher Tg, making the carbon impurity concentration in the film increased. This was indeed the case, as discussed in the following section. Properties of Ta2O5 Films from TBTDET and TBDETCp Precursors. Microstructures and crystal structures of the deposited Ta2O5 films were examined. Figure 4a shows the GIXRD pattern of as-deposited 20 nm thick Ta2O5 films for
needed for a saturated ALD reaction using TBTDET precursor were determined to be >7 and >15 s, respectively. Similar behaviors were observed for the H2O pulse and purge time (>10 and >15 s, respectively) for the case of TBDETCp. The slightly longer H2O pulse time was needed for the saturated ALD reaction in the case of TBDETCp than that of TBTDET. At a Tg of 300 °C, the GPC for TBTDET precursor increased continuously with the precursor injection time, as shown in Figure 2c. Considering the Tg-dependent behavior of GPC for TBTDET precursor, shown in Figure 1a, this behavior indicated that the self-limited ALD regime was disrupted at 300 °C. The saturated GPC for TBDETCp increased with increasing Tg, but still remained constant with >5 s of Taprecursor pulse and >10 s of purge time, as shown in Figure 2c. This result indicated that saturated ALD reaction can be achieved from the TBDETCp at this temperature. Figure 2d shows the ALD saturation behavior with respect to H2O process time, where the Ta-precursor pulse and purge time were fixed at 5 s. The H2O pulse and purge time for the saturated ALD were >10 and >20 s, respectively, for both the Ta-precursors. The experimental data for TBTDET precursor were comparable to the previous report,15 suggesting that they can serve as the reasonable reference to evaluate the performance of TBDETCp as Ta-precursor of Ta2O5 ALD. Figures 3a,b represent the change in LD of Ta atoms and film thickness as a function of ncy grown on native Si substrates. To
Figure 3. Changes in the layer density (LD) and thickness of Ta2O5 film deposited on native Si substrates using (a) TBTDET and (b) TBDETCp, with increase in ALD cycle number (ncy). The changes in LD and its derivatives with respect to ncy (ncy values from 1 to 30) using (c, e) TBTDET and (d, f) TBDETCp on native and HF-dipped Si substrates.
prevent any thermal decomposition of the precursor molecules, the Tg for TBTDET and TBDETCp precursors were selected to be 250 and 300 °C, respectively, hereafter. Thickness (LD) of Ta2O5 films increased linearly with ncy, and the saturated growth rates of film, viz. 0.77 Å cycle−1 (43.7 ng cm−2 cycle−1) and 0.67 Å cycle−1 (37.9 ng cm−2 cycle−1), were obtained for D
DOI: 10.1021/acsami.6b11613 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
Figure 5. Cross-sectional SEM image of Ta2O5 film deposited by (a) TBTDET and (b) TBDETCp on the contact hole patterned substrate, giving an aspect ratio of 5:1 and the opening size of 65 nm. The LD mapping profiles on 200 mm scale wafer for (c) TBTDET at Tg of 250 °C and (d) TBDETCp at Tg of 300 °C.
Figure 4. (a) GIXRD patterns and (b) XRR measurement results of 20 nm thick as-deposited Ta2O5 films grown with TBTDET (black line) and TBDETCp (red line). Inset of (a) shows the selected area diffraction pattern of the as-deposited Ta2O5 film. Cross-sectional view SEM images and AFM topography images (inset) of 60 nm thick Ta2O5 films grown by (c) TBTDET and (d) TBDETCp.
both the precursors were evaluated using the XRF, as shown in Figures 5c,d. From LD mapping of Ta atoms, the nonuniformity of Ta2O5 films was calculated as [(maximum − minimum of LD)/average of LD] × 100%. For TBTDET, the film nonuniformity was quite high (∼23%) on a 200 mm wafer at Tg of 250 °C, as shown in Figure 5c. The high nonuniformity did not originate from the short purge time (7 and 20 s for TBTDET and H2O purge) that can cause chemical vapor deposition reaction because the nonuniformity was not decreased by a prolonged purge time. However, the nonuniformity was reduced significantly (6% in Figure S1) by using TBTDET and H2O plasma (instead of H2O) at the same growth temperature even under short purge time of H2O plasma (5 s). Thus, it was regarded that the TBTDET precursor adsorbed on the wafer uniformly; however, the low reactivity of the TBTDET precursor toward the H2O was the origin of the nonuniformity because the uniformity was highly enhanced by using more reactive oxygen source, i.e., H2O plasma. On the other hand, the Ta2O5 film from TBDETCp precursor exhibited an excellent LD nonuniformity of 2.3% even at a higher Tg of 300 °C. XPS and AES analyses were performed to determine the binding states and chemical composition of 20 nm thick Ta2O5 films. Figures 6a,b illustrate the Ta 4f and O 1s XPS spectra for Ta2O5 films grown using both the precursors. The binding energy (BE) was calibrated by using the C−C peak at BE of 284.8 eV. In Figure 6a, the doublet of Ta 4f7/2 and 4f5/2 peaks, separated by ∼2 eV, were observed at ∼26 and ∼28 eV for both the precursors. However, the whole spectra for TBTDET precursor slightly shifted toward lower BE by ∼0.4 eV compared to that of TBDETCp precursor. Prior reports mentioned that the position of Ta 4f7/2 of +5 oxidation state were in the BE range from 26.3 to 27 eV,9,28−30 which was slightly higher than that of Ta 4f in this work. Because Ta−C and Ta−N binding state had been observed in the range of 22− 24 eV of BE,31,32 it is unlikely that the binding state of Ta was shifted by the incorporation of C and N impurity in the Ta2O5 films. Therefore, it could be suggested that both Ta2O5 films have an oxygen deficient stoichiometry, involving the +4 oxidation state of Ta. The lower BE of the film from the
both TBTDET and TBDETCp. No peaks corresponding to the deposited film (sharp and broad peak at 2θ of 54° and 56° are from substrate and GIXRD target contamination) could be observed in the spectrum. The inset figure shows the selected area electron diffraction patterns from the plan-view TEM of the same film. These structural analysis data show that the asdeposited films with thickness up to 20 nm were amorphous in nature, which was consistent with previous reports for Ta2O5 films grown by ALD at low temperature.14,26 The density of Ta2O5 films deposited by both the precursors was analyzed by using XRR, as shown in Figure 4b. From the best-fitting results using simple model of Ta2O5/Si stacked structure, a film density of 7.698 g cm−3 was achieved for TBTDET at Tg of 250 °C, and 7.998 g cm−3 was obtained with TBDETCp at Tg of 300 °C. The density of bulk crystalline α-Ta2O5 is 8.37 g cm−3, suggesting that the present ALD method enabled the growth of Ta2O5 films with high density.27 Figures 4c and 4d show crosssectional SEM images for 60 nm thick Ta2O5 films grown with TBTDET and TBDETCp, respectively, confirming very smooth surface morphology with no discernible grain or cluster structures across the cross section. The insets in Figures 4c and 4d show the surface morphology image analyzed by AFM from 500 × 500 nm2 scan areas, where the root-meansquare (rms) roughness for both the Ta2O5 films was confirmed to be
