Recently, gate insulator materials of downscaling MOSFET devices and insulator materials for metal-insulator-metal (MIM) capacitor have become key issues in semiconductor memory application field. The existence of gate dielectric suffers from increased gate leakage [1], and the insulator of MIM also cannot meet the requirement of high capacitance density and low leakage current [24]. To solve these challenges, high-k materials are needed for gate insulator and insulator of MIM capacitor. Until now, high-k materials including TiO2, TiN, HfAlO3, BaSeTiO3 and BaTiO3 have been widely studied [59]. Among these materials, BaTiO3 is emerging as a promising material due to the merits of high dielectric constant, low leakage current and excellent piezoelectric and ferroelectric properties [1012]. Using BaTiO3 thin film as the gate insulator and insulator of MIM capacitor can greatly improve the performance and the density of integrated circuit. So far, although a great deal of researchers devoted to researching the characteristics of BaTiO3 thin film for using different applications, there has been little study on micropatterning properties of BaTiO3. A research presents an investigation of the chemical mechanical polishing (CMP) process [13]. However, this CMP method has a significant limitation and complicated fabrication process. With regard to the etching technology, only in [14], a study on characterization of dry etching process is presented, but the authors just give a simple presentation about the relationship between plasma etch rate and applied RF power and mixture gas mixing ratio; there is no deep and systematic characterization for etching mechanism. To date, there is no feasible technology known for the etching of BaTiO3 thin film. These obstacles hinder understanding the properties of the BaTiO3 thin film etching process and further impede the related optimization of process. Therefore, it is necessary to study on how obtain a high etch rate and a good etch profile for dry etching mechanism of BaTiO3 thin film.

In this research, BaTiO3 thin films were etched using inductively coupled plasma (ICP) system with different fluorine-based plasmas. The etch rates of BaTiO3 thin films etched in different fluorine-based (CF4/O2, C4F8/O2 and SF6/O2) plasmas were compared. A comparative study of etch characteristics of the BaTiO3 thin films in these plasmas was conducted. The surface morphology of BaTiO3 thin films was examined by atomic force microscopy (AFM). Also, the chemical compositions and the binding states of the corresponding elements on the surface for each etched films were analysed by X-ray photoelectron spectroscopy (XPS).


The BaTiO3 thin films were deposited by the aerosol deposition (AD) process [15]. The source material for deposition was commercial BaTiO3 powder with a particle size of 300 nm. The total thickness of the deposited BaTiO3 thin film was approximately 300 nm, which starts from Pt/Ti/SiO2/silicon substrate. A Ti (10 nm)/Cr (790 nm) metal shadow mask fabricated by e-beam evaporation is used for the BaTiO3 thin films etching. The dry etching process was performed in an ICP system as shown in Figure 1a,b. The etching properties of BaTiO3 thin films were investigated in CF4/O2, C4F8/O2 and SF6/O2 mixture gas, respectively. The ratios of the three mixture gas between fluorine-based gas and O2 are all fixed to 50:5 sccm. The base conditions of the RF power, ICP power, gas pressure and chamber temperature were 150 W, 1,000 W, 7.5 mTorr and 293 K, respectively. The etch rates were measured using focused ion beam (FIB) in order to ensure the high accuracy of measurement. Finally, shadow mask was stripped by hydrofluoric acid and Cr etchant using wet etching to measure the etch rate of the BaTiO3 thin film after etching process. The surface morphology of BaTiO3 thin film was characterized using AFM. The composition after chemical reaction on the surface of BaTiO3 thin film was investigated using XPS. The Al Kα source provides non-monochromatic X-rays at 1,486.6 eV. The survey spectra are taken at a base pressure of 1.1 × 10−7 Pa, and a binding energy scan range from 0 to 1,000 eV is sufficient to identify all of the detectable elements. Narrow-scan spectra of all regions of interest are recorded with 23.5 eV pass energy to quantify the surface composition and identify the chemical binding state. The peak of C 1 s at 285 eV is assigned to carbon from hydrocarbon contamination, and it is used as the criterion to correct the energy of the spectra. The PHI MultiPakTM software (PHI, Chanhassen, MN, USA) is used to fit the narrow-scan spectra of Ba 3d, Ti 2p, O 1 s and F 1 s for as-deposited and etched BaTiO3 films under Shirley-type background subtraction [16]. All of the BaTiO3 thin film samples for analysis were set as 1 × 1 cm2. The cross-sectional view of the patterned BaTiO3 thin film measured by FIB is shown in Figure 1c.

Figure 1
figure 1

Schematic diagram and entity of ICP system (a, b) and cross-sectional view of SF 6 -based etched BaTiO 3 film obtained by FIB (c).

Results and discussion

Etching rate and surface morphology

Before analysing the etching rate of the BaTiO3 thin films using fluorine-based plasmas, the basic etching behaviour characterizations have to be presented firstly. Actually, the mechanism of the ICP process uses both chemical reaction and physical sputtering. In the CF4/O2 and C4F8/O2 mixing gas experiment, F ions from the fluorocarbon (CF4 or C4F8) has strong chemical reactivity. It reacts with BaTiO3 thin film to form the low volatile reaction byproducts which include BaF x and C x F y . Because of the charging effect, these byproducts are adhered to the etched surface. Meanwhile, the various detached CF m + ions originating from plasma sputter the reaction product from the surface and keep fluoride free to make further chemical reaction [17]. Under the SF6/O2 plasma environment, F ions from sulphur fluoride react with BaTiO3 thin film. The reacted byproducts such as BaF x passivate the surface. In this case, SF n + ions sputter the reaction product to stimulate the chemical reaction. During etching process, lots of volatile carbonmonoxide, carbondioxide and gaseous sulphur were pumped off by vacuum pumps. In this research, the introduced O2 played a role of catalyst, which can enhance the etch rate effectively.

The etch rate of the BaTiO3 thin film and the etch selectivity of BaTiO3 over Ti/Cr metal shadow mask as a function of three different types of mixing of plasmas are shown in Figure 2. Data show that the maximum etch rate is about 46.7 nm/min in SF6/O2 plasmas. The selectivity achieved was 2.53. As changing the CF4 and C4F8, the etch rate of BaTiO3 thin film decreases, which has an etch rate of 41.8 and 27.0 nm/min, while the selectivity achieved was 4.4 and 6.25. Based on the above experimental result, it is disclosed that higher BaTiO3 thin film etch rates can be achieved with SF6/O2 mixture gas compared with CF4/O2 and C4F8/O2 mixture gas and C4F8/O2 mixture gas is the worst one for BaTiO3 thin film etching. Analysing the reasons of the abovementioned result, it is possible to consider for the two following explanations. The first is a minimum amount of ion energy is necessary for SF6. The SF6-based plasmas can get higher kinetic energy in same condition with CF4-based and C4F8-based plasmas, which accelerate the chemical reactions as well as physical ion bombardment [18]. The second is the lower volatility of fluorocarbon polymers impede the further etching process in C4F8-based environment as proved by subsequent XPS experiment.

Figure 2
figure 2

Etch rates and selectivity of BaTiO 3 films. Etch rates of BaTiO3 films (black solid line) and selectivity of BaTiO3 to Ti/Cr shadow mask (red dashed line) as a function of the different fluorine-based mixture gas.

Figure 3 demonstrates the surface morphologies of the same BaTiO3 films which are under the unetched and etched by each fluorine-based plasmas. In each sample, the surface morphologies are investigated by root-mean-square (RMS) roughness and cross-sectional surface line profiles. Figure 3a,b,c,d shows 10 × 10 μm2 AFM images of 3-D views. It can be seen that the RMS roughness value of the as-deposited BaTiO3 film is 25.69 nm. After the BaTiO3 films etched in three fluorine-based plasmas, a better surface morphology can be achieved compared to the unetched examined sample, while there is no obvious difference that appears between the CF4/O2 etched BaTiO3 film surface and SF6/O2 etched surface. They have a RMS roughness value of 19.03 and 19.43 nm, respectively. However, in the case of C4F8/O2 etched BaTiO3 film surface, surface morphology is worse than two others with a RMS roughness value of 23.12 nm. This may attribute to the re-deposition and growth of C x F y polymer during the C4F8/O2 ICP etch [19]. Obviously, the quality of surface morphology of BaTiO3 film is deteriorated after etching in C4F8/O2 in comparison with those etched by CF4/O2 and SF6/O2. Figure 4a,b,c,d shows the AFM top 2-D views of the selected areas, and the cross-sectional surface line profiles are shown in Figure 4 (a-1 to d-1) corresponding to Figure 4a,b,c,d). The cross-sectional surface line profiles indicate the change in both the diameter and depth of the craters on the surface, which follow the trend in Figure 3a,b,c,d). Figure 4 (a-1) shows the craters on the as-deposited BaTiO3 films, which have a diameter of 1.85 μm and a depth of 60.7 nm. After three different fluorine-based plasma etching treatment, the smaller craters can be observed. Two relative high-quality BaTiO3 films can be found in CF4/O2 and SF6/O2 plasmas as shown in Figure 4 (b-1) and (d-1), which have diameters of 0.7 and 0.6 μm and depths of 29.3 and 35.9 nm, respectively. However, Figure 4 (c-1) reveals that a relative larger craters with a diameter of 1.7 μm and a depth of 55.9 nm appeared on the surface of the thin film.

Figure 3
figure 3

3-D view of surface morphologies of the BaTiO 3 thin films by AFM. (a) As-deposited, (b) etched in the CF4/O2 plasma, (c) etched in the C4F8 plasma, and (d) etched in the SF6 plasma. The etching time is set to 3 min, and the scanned areas are all 10 × 10 μm2.

Figure 4
figure 4

The surface morphologies of the BaTiO 3 films, which are under the unetched and etched by each fluorine-based plasmas. (a, b, c, d) The AFM top views of the selected areas and (a-1 to d-1) the corresponding cross-sectional surface line profiles of BaTiO3 thin films under different conditions.

XPS analysis

In order to know the more detailed surface chemical composition, an XPS analysis was performed. The XPS survey spectra obtained among the as-deposited and etched BaTiO3 films by three different mixture gas are shown in Figure 5a. In Figure 5a (1), the photoelectron lines of Ba, Ti, O and C elements exist on the as-deposited BaTiO3 films surface. C 1 s is used for the criterion to rectify the energy of spectra that has a peak at 285.0 eV from contaminated hydrocarbon [20]. There are Ba, Ti, O, C and F XPS photoelectron lines, where Ba 4d (89.7 eV) (CF4/O2 etched), Ba 4p (178.8 eV) (CF4/O2 etched), Ba 3d 5/2 (780.16 eV) (CF4/O2 etched), Ba 3d 3/2 (795.75 eV) (CF4/O2 etched), Ti 3p (73.5 eV) (CF4/O2 etched), Ti 2p (458.1 eV) (CF4/O2 etched), C 1 s (285.0 eV) (CF4/O2 etched), O 1 s (529.5 eV) (CF4/O2 etched) and F 1 s (684.1 eV) (CF4/O2 etched); Ba 4d (89.4 eV) (C4F8/O2 etched), Ba 4p (178.2 eV) (C4F8/O2 etched), Ba 3d 5/2 (780.55 eV) (C4F8/O2 etched), Ba 3d 3/2 (795.8 eV) (C4F8/O2 etched), Ti 3p (74.1 eV) (C4F8/O2 etched), Ti 2p (459.1 eV) (C4F8/O2 etched), C 1 s (285.0 eV) (C4F8/O2 etched), O 1 s (531.4 eV) (C4F8/O2 etched) and F 1 s (683.8 eV) (C4F8/O2 etched); Ba 4d (89.4 eV) (SF6/O2 etched), Ba 4p (178.7 eV) (SF6/O2 etched), Ba 3d 5/2 (779.1 eV) (SF6/O2 etched), Ba 3d 3/2 (795.35 eV) (SF6/O2 etched), Ti 3p (73.2 eV) (SF6/O2 etched), Ti 2p (458.0 eV) (SF6/O2 etched), C 1 s (285.0 eV) (SF6/O2 etched), O 1 s (529.85 eV) (SF6/O2 etched) and F 1 s (684.1 eV) (SF6/O2 etched) and the valence-type Auger lines for F (KLL) (838.2 eV), Ba (MNN) (902.7 eV) and O (KLL) (990.3 eV) can be confirmed on the three etched BaTiO3 film surfaces in Figure 5a (2, 3, 4). Figure 5b shows the XPS narrow-scan spectra of F 1 s obtained from the BaTiO3 films surface in as-deposited and etched by different mixture gas. There is no photoelectron line of the element F in the as-deposited BaTiO3 films specimen. After etching in CF4/O2, C4F8/O2 and SF6/O2 mixing gas environment, each F 1 s XPS spectrum shows a wide peak in the region of 682 to 686 eV with a maximum corresponding to a binding energy of 684.1, 683.86 and 684.02 eV, respectively. The fact that the XPS survey spectra of BaTiO3 films in Figure 5a is higher consistent with the F 1 s narrow-scan spectra shown in Figure 5b indicates that chemical reaction occurred when the fluorine-based plasmas were applied into etching process.

Figure 5
figure 5

XPS survey spectrum of BaTiO 3 thin films and F 1 s narrow-scan spectrum for each BaTiO 3 thin films. (a) XPS survey spectrum of BaTiO3 thin films: (1) as-deposited, (2) etched in the CF4/O2 plasma, (3) etched in the C4F8/O2 plasma, and (4) etched in the SF6/O2 plasma. (b) F 1 s narrow-scan spectrum for each BaTiO3 thin films. The vertical dashed line indicates each maximal position of the F 1 s singlet.

Figure 6 shows the peaks of the XPS narrow-scan spectra of (a) Ba 3d, (b) Ti 2p, (c) O 1 s and (d) F 1 s, which were obtained from the BaTiO3 film in as-deposited and each different fluorine-based plasma etched environment. Figure 6a shows the photoelectron peaks of Ba 3d. It can be seen that the unetched doublet consists of two peaks which are observed at 779.9 and 795.1 eV, which are mainly identified as signals from Ba-O bonds. The deconvoluted sub-peaks of Ba 3d 5/2 and Ba 3d 3/2 are related to BaCO3[21] or a relaxed Ba phase because of the O vacancies and the cation defects [22]. After the BaTiO3 thin films were exposed to the CF4/O2 and C4F8/O2 plasma severally, the peaks of Ba 3d 5/2 and Ba 3d 3/2 were chemically shifted to a higher binding energy and the maximum deviation are about 0.26/0.255 and 0.65/0.70 eV in comparison with the unetched counterparts. After the treatment in SF6/O2 plasma, Ba 3d 5/2 and Ba 3d 3/2 peaks show higher binding energy shifts of 0.1 and 0.25 eV, respectively. The shift of peaks indicates that Ba chemically reacted with F-component species, which some Ba-O bonds are broken and a few Ba-F bonds are generated. Because the bonding energies of the Ba-F bonds are higher than Ba-O bonds [23], peaks of the BaTiO3 film shift towards higher binding energy.

Figure 6
figure 6

The narrow-scan spectra of (a) Ba 3d , (b) Ti 2p , (c) O 1 s and (d) F 1 s peaks for each BaTiO 3 film. (1) As-deposited, (2) etched in the CF4/O2 plasma, (3) etched in the C4F8/O2 plasma, and (4) etched in the SF6/O2 plasma. Red bold solid lines represent the fitted experimental results after subtracting Shirley-type background. Peaks with dashed line are fitted sub-peaks. Peak positions of the Ba 3d, Ti 2p, O 1 s and F 1 s are the average peak positions for the corresponding sub-peaks. The broken lines reveal the approximate chemical shifts.

Figure 6b shows the photoelectron peaks of Ti 2p from the as-deposited and etched BaTiO3 films surface. In Figure 6b (1), the unetched Ti 2p consists of two wide peaks of Ti 2p 3/2 (457.8 eV) and Ti 2p 1/2 (463.57 eV) due to Ti-O bonds. After etching in CF4/O2, C4F8/O2 and SF6/O2 plasma, the peaks of Ti 2p 3/2 and Ti 2p 1/2 shift towards higher binding energy regions by 0.05 and 0.23, 1.35 and 0.98, and 0.2 and 0.43 eV, respectively, which is shown in Figure 6b (2, 3, 4). When BaTiO3 film is etched in C4F8/O2 plasma, the intensity of the Ti 2p 3/2 and Ti 2p 1/2 peaks decreased obviously because of the higher volatility of byproduct TiF x . The byproduct TiF x can be partly removed from the film surface as the thermal desorption process. The reason why the chemical shifts towards higher binding energy can be explained by the theory of bond shift compensation scheme between TiF x and the etched BaTiO3 film [24].

The fitted O 1 s narrow scan spectra of each BaTiO3 sample is shown Figure 6c. An O 1 s (531.24 eV) peak of the as-deposited BaTiO3 film which consists of three sub-peaks located at 529.65, 531.2 and 532.4 eV is shown in Figure 6c (1). The three sub-peaks are mainly affected by Ba-(O 1 s) (780 eV), Ti-(O 1 s) (529 eV) and C-(O 1 s) (532.3 eV) bonds [20]. The two oxides of Ba are made up of BaO and TiO2 in the BaTiO3 film, the surface contamination introduced the C-O bonds. The shoulder located at 532.4 eV is ascribed to the surface water vapour and carbon dioxide. In this research, the BaTiO3 film was deposited by AD method, the surface phase was formed with water vapour and carbon dioxide inevitably. After etching in each fluorine-based plasma, the etched film shows a chemical shift towards higher binding energy region, which is demonstrated in Figure 6c. It is revealed that the disconnection between Ba-O and Ti-O and re-connection between Ba-F and Ti-F happened through the physical sputtering of CF m + and SF n + ions and chemical reactions with reactive fluorides. A phenomenon can be observed that the sub-peaks at 532.4 eV is disappeared after etching in different fluorine-based plasmas. The reason of the decrease of sub-peaks in Figure 6c (2, 3, 4) compared with Figure 6c (1) is that the physical bump of ions removed the surface contamination (carbon dioxide) and the etching process is in the vacuum conditions, which would not introduce secondary contamination. Therefore, the sub-peaks at 532.4 eV in Figure 6c (2, 3, 4) cannot be found anymore.

Figure 6d shows the F 1 s narrow-scan spectra of the as-deposited and each etched BaTiO3 film surface. As shown in Figure 6d (1), there is no signal from a fluorine-contained compound. While adding the etching reaction CF4/O2 and SF6/O2 plasma for each sample, F 1 s appear at the binding energy of 684.1 and 684.02 eV, as revealed in Figure 6d (2 and 4). The sub-peaks are situated at 684.1/686.1 and 684.02/686.2 eV, respectively, which are assigned to the product of the etching reaction of Ba-F and a residue of Ti-F [25]. After etching in C4F8/O2 plasma, the F 1 s signal emerged and consisted of three sub-peaks (683.86, 686.01 and 688.15 eV). Unlike the CF4/O2 and SF6/O2 plasma, the main contributions of these three sub-peaks result from Ba-F, Ti-F and a residue of C-F compounds [26].


In this present work, an investigation of dry etching mechanisms for BaTiO3 thin films in ICP system using different fluorine-based plasmas was carried out. Experimental results indicate that a higher BaTiO3 thin film etch rates were achieved with SF6/O2 plasmas. The etch rate of SF6/O2 plasmas is over than 46.7 nm/min at RF power/ICP power of 150/1,000 W under gas pressure of 7.5 mTorr. The result of AFM reveals that the roughness of all etched surfaces by fluorine-based plasmas ameliorated in comparison with the as-deposited surface. Moreover, a better etched surface morphology can be achieved using SF6/O2 plasmas. Chemical compositions and bonding states on as-deposited and each etched BaTiO3 thin films were investigated by XPS. The XPS analysis indicated the accumulation of reaction products. According to the comprehensive analysis and comparison, SF6-based plasmas showed higher etch rates and excellent surface morphology. In addition, in terms of recent severe environment, SF6 gas is not a potent greenhouse gas compared with other two greenhouse effect gas CF4 and C4F8. SF6-based plasmas can be recommended to be an ideal candidate gas for BaTiO3 dry etching.