Bin Huang‡
ab,
Kun Cao‡ab,
Xiao Liuab,
Lihua Qianc,
Bin Shan*b and
Rong Chen*a
aState Key Laboratory of Digital Manufacturing Equipment and Technology, School of Mechanical Science and Engineering, Huazhong University of Science and Technology, Wuhan 430074, People's Republic of China. E-mail: rongchen@mail.hust.edu.cn; Fax: +86-27-87558744; Tel: +86-27-87558744
bState Key Laboratory of Material Processing and Die and Mould Technology, School of Materials Science and Engineering, Huazhong University of Science and Technology, Wuhan 430074, People's Republic of China. E-mail: bshan@mail.hust.edu.cn
cSchool of Physics, Huazhong University of Science and Technology, Huazhong University of Science and Technology, Wuhan 430074, People's Republic of China
First published on 19th August 2015
Ultrathin cobalt oxide (CoOx) films (<10 nm) have been prepared on both planar and three dimensional substrates by atomic layer deposition (ALD) using Co(Cp)2 and O3 as precursors. The optimal temperature window was 150–250 °C with a saturated growth rate of ∼0.37 Å per cycle. While the main composition of the thicker film consists of Co3O4 as verified by Raman spectroscopy, the initial few layers grown by ALD show a mixture of Co3O4 and CoO as revealed by X-ray photoelectron spectrum (XPS). The surface morphology of the film is greatly influenced by the deposition temperature as atomic force microscopy (AFM) and high-resolution transmission electron microscopy (HR-TEM) characterization show the formation of crystalline islands and uneven coating on porous structures at high temperatures, while a smooth and uniform coating can be obtained at lower temperatures. The role of nucleation sites and crystallization speed on film morphology is discussed. Furthermore, we show that tuning of deposition temperatures could lead to improved catalytic activities as demonstrated in the CO oxidation light off test.
Many methods have been developed for thin film fabrication, including spray pyrolysis, chemical vapor deposition (CVD), physical vapor deposition (PVD), pulse laser deposition (PLD), and atomic layer deposition (ALD).14–20 Among these methods, ALD attracts great interest due to its self-limiting nature and exceptional coating conformality on three dimensional structures. While ALD generally exhibit a linear growth in the steady state, its growth behavior during the initial nucleation period could deviate considerably from the ideal layer-by-layer growth model, due to the complexity of nucleation and crystallinity of the deposited film.21–24 While ALD growth of Co3O4 film on silicon-based substrates has been reported,18,25,26 the ultrathin film characteristics during the initial growth stage has rarely reported in previous studies. Since the difference in the ultrathin film's thickness, composition, or morphology could lead to considerable changes in system properties,27,28 it is thus of essential importance to understand the growth behavior of ALD grown cobalt oxide at the initial growth stage for the further optimization of CoOx ultrathin films.
We report here a comprehensive study on the growth behavior of CoOx ultrathin films (<10 nm) by ALD using Co(Cp)2 and O3 as precursors focusing on their composition and surface morphology. The coating conformality of CoOx was evaluated on both planar and three dimensional substrates. At low deposition temperature of 150 °C, films are uniform and smooth, while island and uneven films are obtained after raising deposition temperature to 250 °C. The ultrathin film's composition is a mixture of CoO and Co3O4, which is different from its steady state counterpart in the bulk film. The obtained films show good CO oxidation catalytic performance which strongly correlates to the deposition temperature. Based on these observations, the growth behavior of ALD-CoOx was discussed.
The film thickness was measured by spectroscopic ellipsometer (SE, J. A. Woollam Co. M2000). The structure of thicker films was evaluated by Raman spectroscopy (LabRAM HR800) with 532 nm laser excitation. The chemical composition of ultrathin films was analyzed by X-ray photoelectron spectra (XPS, VG Multilab 2000) using a monochromatic Al Kα X-ray radiation. XPS data were fitted with XPSPEAK4.1 software and the binding energy was calibrated with respect of C 1s (284.6 eV) peak position. Surface morphology of freshly prepared samples was measured by a tapping mode atomic force microscopy (AFM, Agilent 5500) to avoid potential air-contacted contamination. Transmission electron microscope (TEM, JEOL-2010FEF) was operated with a 200 kV acceleration voltage to study the film morphology as well as crystallinity. To avoid the diffraction from substrate silicon and obtain stronger signals from ultrathin films, some samples were directly deposited on the TEM grid as well.
The catalytic activity toward oxidation of CO was tested in a home-built stainless reactor. All samples were deposited on Si wafer with area of 1 cm*2 cm at different deposition temperature. For the evaluation of catalytic activity, the flow rate of inlet gas is set to 5 sccm controlled through a mass flow controller (Sevenstar, Beijing Sevenstar Electronics Co., Ltd), containing 1% CO, 10% O2 diluted in argon gas. Temperature is raised up from room temperature to 310 °C with a ramped rate of approx. 2 °C min−1. Data was recorded with a thermocouple attached at the bottom of the samples. A quadrupole mass spectrometers (QMS) was connected to the outlet to monitor the exhaust gas. All these setup of this planar catalyst system gives high sensitivity and reproducibility for CO conversion.
![]() | ||
Fig. 1 (a) The growth rate of cobalt oxide as a function of deposition temperature, (b) dependence of film thickness on the number of ALD cycles at different temperature. |
The film's thickness as a function of deposition cycles was plotted in Fig. 1(b). Ideally linear growth as a function of ALD cycles is shown at low deposition temperature (150 °C). However, films deposited at 250 °C show inhibited growth in the initial cycles as indicated by the thinner film thickness and nonlinear growth curve for the first 100 cycles. This phenomenon implies a relative slow nucleation rate at 250 °C as evidence by the slower growth rate in the initial 100 cycles.
However, the situation is considerably different when ultrathin CoOx film is deposited. X-ray photoelectron spectroscopy (XPS), a surface sensitive analytical technique, was conducted to investigate the film's composition at the initial growth stage with different film thickness and deposition temperature. High resolution Co 2p spectra shown in Fig. 3(a) and (b) splits into two peaks, which can be attributed to Co 2p1/2 and Co 2p3/2. No metallic cobalt impurity exists in the film since there is no peak at the binding energy (Eb) for metallic cobalt located at ∼778 eV.31 For films deposited at 150 °C with thickness larger than 6 nm, the Co binding energy (Eb(Co 2p3/2) is located at ∼780 eV and Eb(Co 2p1/2) at ∼795 eV) is in a good agreement with that of Co3O4.31–34 However, when the film is thin enough (for example, thickness ≤ 1 nm), the binding energy for main Co 2p3/2 and Co 2p1/2 shifts to ∼780.4 eV and ∼796.1 eV, which corresponds to the signatures of CoO.31,35 What's more, the satellite peak located at 5.6 eV higher than the main peak gradually appears. This is strong evidence for the existence of CoO at the initial grown stage, and the film composition gradually changes to Co3O4 for the steady growth.10,31–33,35–38 The composition evolution of film deposited at 250 °C shows a similar trend. The satellite peak disappears gradually with the increase of the film thickness. The only noticeable difference from that of 150 °C is the relatively slower conversion speed, thus thicker films are required to show such transition behavior (Fig. 3(b)). The O 1s XPS spectrum is given in Fig. 3(c) and (d). The peak located at ∼531.5 eV can be attributed to near surface chemisorbed oxygen species to oxygen vacancies (OH−, O2− et al.), which is indicative of cobalt oxide surface39,40 and was not from surface contaminations. This peak could be observed at initial growth stage of cobalt oxide deposited at both 150 °C and 250 °C. The peak locates at ∼529.5 eV can be ascribed to lattice oxygen of Co3O4.41 When the thickness reaches a turning point, film composition gradually becomes pure Co3O4 which may leads to the peak appearance at 529.5 eV. Compared with deposition done at 250 °C, this peak for lattice oxygen of Co3O4 was observed for thinner film at deposition temperature of 150 °C. The phenomenon indicates a quicker composition transformation at lower deposition temperature. Our results show that the composition is the mixture of Co3O4 and CoO for ultrathin CoOx film, and the content of Co3O4 phase increases with the increase of ALD cycles. After a threshold of ALD cycles the CoO ratio would fall below the detection limit and the whole film appear to be pure Co3O4. The speed of the conversion is faster at lower deposition temperature. Our Raman and XPS results reveal that CoO forms during the initial growth stage within the ALD temperature window, while the subsequent layer by layer growth results in Co3O4 films.
![]() | ||
Fig. 3 XPS spectra evolution for (a) Co 2p at 150 °C, (b) Co 2p at 250 °C, (c) O 1s at 150 °C and (d) O 1s at 250 °C. |
![]() | ||
Fig. 4 Surface morphology evolution for cobalt oxide deposited at (a) 150 °C, (b) 200 °C and (c) 250 °C for 100, 200 and 300 cycles. |
Fig. 5 shows the data for the dependence of film's RMS evolution on deposition temperature and growth cycles. The film's roughness keeps about the same at 150 °C with the increase of ALD cycles. As a comparison, roughness of films grown at 250 °C is much larger and keeps increasing at the initial 300 cycles. The RMS for 400 cycles cobalt oxide at 250 °C is lower than 300 cycles film, due to the merging of neighboring grains. This result shows a strong relationship between the deposition parameters and the resulting feature of the ultrathin cobalt oxide film.
To study the crystallinity of ALD ultrathin cobalt oxide, films were directly deposited on the TEM grid to avoid the signals from silicon substrate and obtain stronger diffractions from ultrathin films. Fig. 6 shows the TEM images and diffraction patterns for samples after 100, 200 and 300 ALD cycles. The difference of films' feature is consistent with the AFM observation that low density of islands and large crystal size were observed at 250 °C. Besides, the particles tend to crystallize more easily at 250 °C than that at 150 °C as indicated by the selected area electron diffraction (SAED) patterns (Fig. 6, inset). From the SAED patterns, cobalt oxide synthesis at 250 °C have crystalline structure from initial 100 cycles while cobalt oxide synthesis at 150 °C is amorphous. As ALD cycles increase, both samples have preferred crystalline orientation of (220), (311) and (331) as three nearest lattice planes.
![]() | ||
Fig. 6 TEM images and diffraction patterns of cobalt oxide grown for (a and d) 100 cycles, (b and e) 200 cycles and (c and f) 300 cycles at (a–c) 150 °C and (d–f) 250 °C. |
![]() | ||
Fig. 7 TEM observation for nanoporous gold samples coated by 100 cycles ALD cobalt oxide at (a), (b) 150 °C and (c and d) 250 °C. |
Side view HRTEM characterization for cobalt oxide grown on porous gold at both 150 °C and 250 °C with increasing ALD cycles are shown in Fig. 8. From HRTEM images, it can be seen that deposited cobalt oxide is amorphous in the initial stage at 150 °C, and change into crystalline structure as ALD cycles increase. At 250 °C, crystalline structure was observed in the initial stage which is consistent as SAED pattern indicate. (311) and (220) directions are preferred growth orientations for both deposition temperature.
![]() | ||
Fig. 8 HRTEM images cobalt oxide deposited on nanoporous gold with 50, 100, 200 cycles ALD at (a–c) 150 °C and (d–e) 250 °C respectively. |
With regard to the nucleation site density, the chemisorption of the precursors on the surface of the substrate is the dominant factor in the initial nucleation process. The nucleation events generally occur more easily on the surface active sites (defects or surface function groups). As indicated by the initial growth curve (Fig. 1(b)), AFM and TEM observations, the cobalt oxide shows poor nucleation under 250 °C. Literatures indicated that the density of surface hydroxyl group on silicon substrate was temperature-dependent and decreases at higher temperatures due to the dehydroxylation process.42–46 Since the loss of surface hydroxyl would retard the nucleation process, it is reasonable that cobalt oxide film shows poorer nucleation, longer incubation cycles, and lower density of grown islands at 250 °C. The same conclusion holds on nanoporous gold substrate. The surface of nanoporous gold was reported to have strong ability for the adsorption of molecular water, resulting in the formation of hydroxyl species on the gold surface.47–49 The concentration of adsorbed hydroxyl species on the gold surface would also decrease with rising deposition temperature.50,51 Therefore, the nucleation sites of cobalt oxide grown on porous gold at 250 °C are limited due to the reduction of hydroxyl group, inducing the lower density of nuclei. The effect of surface hydroxyl groups on the film growth has been further confirmed by CoOx deposition experiment on piranha treated Si samples (Fig. S2†).
With regard to the film crystallinity, the deposition temperature also plays a crucial role. The trend of crystallinity as function of temperature has been observed in the previous study of ALD metal oxide (TiO2, HfO2, ZrO2) that high temperature can promote the surface mobility of the absorbed species and enhance the ordering of the structure with more energy available for crystallization.46,52–54 Therefore, films deposited at relatively high temperatures normally include more crystalline phase. The crystallites occur more easily at higher temperatures inducing the very fast crystal growth along specific orientations (in our case, (220) and (311) for cobalt oxide growth).44,46 On the other hand, lower nucleation density decreases the steric hindrance on adsorption of precursors. Therefore, the precursors can easily adsorb on the boundary of the islands and promote the growth of the individual oxide particles rapidly, leading to the rough surface formation. The AFM (Fig. 4) and TEM (Fig. 6) observation confirms the growth and crystallization of the large islands. As discussed above, the relatively small number of nucleation sites and the fast crystallization speed function coherently and cause the rough and discontinuous CoOx film morphology at 250 °C.
Finally, CO oxidation activity was studied to illustrate the effect of deposition parameters on its catalytic activity. Samples were deposited at 150 °C, 200 °C and 250 °C (designated as T150 °C, T200 °C and T250 °C) with thickness of ∼10 nm. Bare Si wafer with the same area was used for comparison. Obviously, all cobalt oxide films show negative temperature shift relative to bare Si wafer (see Fig. 9). A noticeable tendency can be observed that the catalytic ability is in the following order: T150 °C > T200 °C > T250 °C. This result is reasonable since it is generally accepted that the oxidation of CO is attributed to the existence of Co3+ cation. The CO can be more easily absorbed on the Co3+ sites and react with the oxygen to form CO2.50,51 Thus, higher content of Co3O4 would in general lead to better CO oxidation catalytic activity. For cobalt oxide films deposited at lower temperatures, the composition is primarily Co3O4, while there still exists a larger portion of CoO in the film deposited at higher temperatures. Controlled experiments were further carried out to determine the influence of composition and morphology to the catalyst activity (detail results are given in ESI, Fig. S4–6†), which reveals that surface oxygen and cobalt oxide composition are two main factors influencing the catalytic performance. Those results confirm the importance for composition control of ALD CoOx ultrathin film.
![]() | ||
Fig. 9 CO oxidation light off test of cobalt oxide thin film catalyst grown under different temperature. |
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra09782g |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2015 |