- Email: [email protected]
Ar inductively coupled plasma
Vacuum 93 (2013) 50e55
Contents lists available at SciVerse ScienceDirect
Vacuum journal homepage: www.elsevier.com/locate/vacuum
Etch characteristics of HfO2 thin films by using CF4/Ar inductively coupled plasma Pil-Seung Kang, Jong-Chang Woo, Young-Hee Joo, Chang-Il Kim* School of Electrical and Electronics Engineering, Chung-Ang University, 221 Heukseok-Dong, Dongjak-Gu, Seoul 156-756, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 November 2012 Received in revised form 24 December 2012 Accepted 27 December 2012
In this study, we carried out an investigation of the etching characteristics (etch rate, selectivity) of HfO2 thin films in the CF4/Ar inductively coupled plasma (ICP). The maximum etch rate of 54.48 nm/min for HfO2 thin films was obtained at CF4/Ar (¼20:80%) gas mixing ratio. At the same time, the etch rate was measured as function of the etching parameters such as ICP RF power, DC-bias voltage, and process pressure. The X-ray photoelectron spectroscopy analysis showed an efficient destruction of the oxide bonds by the ion bombardment as well as an accumulation of low volatile reaction products on the etched surface. Based on these data, the chemical reaction was proposed as the main etch mechanism for the CF4-containing plasmas. Ó 2013 Elsevier Ltd. All rights reserved.
Keywords: Etching HfO2 XPS ICP CF4
1. Introduction The complementary metal oxide semiconductor transistor scaling beyond the present 45 nm technology nodes makes it difficult to grow high quality ultra-thin oxide. The thickness of gate dielectrics, SiO2, should be reduced down to 2 nm or less. The thickness reduction of SiO2 brings many serious problems such as increased gate leakage current and reduced oxide reliability. To overcome this drawback, many metal oxides with high dielectric constant materials have been reported, such as TiO2, Ta2O5, SrTiO3 (STO), Al2O3, Y2O3, ZrO2, and BaSrTiO3 (BST) [1e4]. Although these materials have high dielectric constant, some of these fail one or more of the criteria. Al2O3 and Y2O3 did not provide sufficient advantages over SiO2 or Si3N4. TiO2 and Ta2O5 were observed to react with Si substrate. STO or BST may cause fringing field-induced barrier lowering effect. Among high-k materials compatible with silicon, oxides of Zr and Hf are attracting much attention recently because Hf forms the most stable oxide with the highest heat of formation. Additionally, Hf can reduce the native oxide layer to form HfO2. However, new gate oxide candidates must satisfy a standard processing procedure. In order to solve this problem, an understanding of the relations between the etch characteristics of HfO2 thin film and plasma properties is required for the low damaged removal process. * Corresponding author. Tel.: þ82 2 820 5334; fax: þ82 2 812 9651. E-mail address: [email protected] (C.-I. Kim). 0042-207X/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.vacuum.2012.12.007
Until now, there are only few works devoted to the investigations of etching characteristics of the HfO2 thin films using both fluorine- and chlorine-based plasma chemistries. Refs. [5e8] show the dependences of HfO2 etch rate on operating conditions for O2/BCl3, Cl2/BCl3, XeF/Ar, BCl3/C4F8/Ar, and C12/SF6/Ar plasmas, but do not discuss the etch mechanisms as well as the relationships between process parameters, and chemistry. The others are our works where the mechanism of HfO2 in Cl2/Ar, CH4/Ar, HBr/Ar, Cl2/BCl3/Ar and BCl3/Ar plasmas was analyzed using the combination of modeling and diagnostics tools [9e15]. In this study, we investigated the etch characteristics of HfO2 thin films in inductively coupled plasma (ICP) system with CF4/Ar gas chemistries. Etching characteristics on HfO2 thin films have been investigated in terms of etch rate and selectivity. Also, the etch rate and selectivity were measured as function of the etching parameters such as inductively coupled plasma RF power, DC-bias voltage, and process pressure. The chemical binding states in the surface of the etched HfO2 thin films were investigate with X-ray photoelectron spectroscopy (XPS). Scanning electron microscopy (SEM) was used to investigate the surface morphology of Hf-based high-k dielectrics exposed in plasma and etching profile. 2. Experimental The HfO2 thin films were prepared on an 8-inch Si (100) substrate by using plasma enhanced atomic layer deposition (PEALD). The PEALD system (ASM Genitech Korea Ltd. Stella 3000) can
P.-S. Kang et al. / Vacuum 93 (2013) 50e55
51
Fig. 1. Etch rates of HfO2 thin films as a function of the CF4/Ar gas mixing ratio. (a) The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C. (b) The DC-bias voltage was maintained at 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C. (c) The RF power was maintained at 600 W, the process pressure was 15 mTorr, and the substrate temperature was 40 C. (d) The RF power was maintained at 600 W, the DCbias voltage was 250 V, and the substrate temperature was 40 C.
deposit three kinds of binary oxides and their mixed materials. Three canisters of Ti, Al, and Hf metal source are contained in the system. The canister for Hf metal source (TEMAHf) with the volume of 1000 cc can be heated up to 65 C and delivery line for Hf metal source can be heated to 100 C surrounded with heating tapes to prevent condensation of precursor in the delivery line. The source in this canister is delivered to the main reactor chamber with Ar carrier gas instead of conventional bubbler system. The oxygen source used in this experiment is the O2 plasma, which is generated by RF plasma generator. The upper electrode could apply RF power with 13.56 MHz RF plasma source to the substrate that was grounded with capacitive. The line for Hf precursor was heated at the temperature 100 C by heating tapes. The flow rate of Ar purge was fixed at 200 sccm and the chamber pressure was maintained at 3 Torr during deposition process. The final thickness of HfO2 thin films was about 100 nm. This is schematically shown in Ref. [9]. The reactor consists of cylindrical aluminum anodized chamber with diameter of 26 cm. A 13.56 MHz RF power generator was connected through a matching network to the coil and generate plasma on the top of the process chamber. Another 13.56 MHz RF power generator was applied to the substrate to induce the DC-bias voltage to the wafer. Wafers were placed on a chuck, which is used as the bottom electrode. The chamber was evacuated to 106 Torr using a mechanical pump and a turbo-molecular pump. The HfO2 thin films were etched with CF4/Ar gas. The gas mixing ratio was varied to find the characteristics of etching. For these experiments, process pressure, RF power, bottom DC-bias voltage, and substrate temperature were 15 mTorr, 600 W, 250 V and 40 C, respectively. In
addition, plasma etching of HfO2 thin films was investigated by changing the etching parameter including top RF power of 500e 800 W, DC-bias voltage to the substrate of 150 to e300 V, and process pressure of 5e20 mTorr, and the gas flow rate of 20 sccm in the CF4/Ar gas mixing ratio, respectively. The etch rate was measured by the surface profiler (KLA Tencor, a-step 500). The compositional changes on the etched HfO2 surface were investigated using X-ray photoelectron spectroscopy (SIGMA PROBE, Thermo VG Scientific). The spectra were plotted by counting the photoelectrons at kinetic energy intervals of 0.1 eV. The detailed information on the inner region of the film was provided by the spectra recorded at the electron take-off angles of 90 . All of the samples for the XPS analysis were bare HfO2 thin films that did not have any photoresist patterns, and the size of the samples was 1 1 cm2. The etching time was 10 s. The etching profile of the cross-section was characterized using FE-SEM (Sirion 400, FEI). The HfO2 thin films used for measuring the etch rate and etching profile did have a photoresist pattern (SS03A9). The width and thickness of the PR pattern were 1.5 and 1.02 mm, respectively. 3. Result and discussion It is well known that the etch mechanism is not only influenced by the processing parameters for any plasma etching process but also depended strongly on the types of both chemically active species and surface atoms determining the volatility of reaction products. It can be understood that the magnesium fluorides are very low volatile compounds, so that the thermal desorption of etch
52
P.-S. Kang et al. / Vacuum 93 (2013) 50e55
products, and thus, the spontaneous etching can be neglected. Therefore, the HfO2 etch properties in the CF4/Ar plasma can be preliminary proposed as the chemical reaction where the role of ion bombardment includes three main effects: 1) sputtering of the main material; 2) sputtering (ion-stimulated desorption) of the etch products to provide the access for the F atoms to the etched surface; and 3) destruction of the HfeO bonds to provide the chemical interaction of F atoms with Hf. The evidence of the last pathway follows from the fact that the HfeF bond is weaker than the HfeO, so that the direct interaction between HfO2 and F atoms seems to be impossible. 3.1. Effect of gas mixing ratio Fig. 1(a) represents the etch rate of HfO2 thin film and the selectivity of HfO2 to SiO2 as a function of CF4/Ar gas mixing ratio. The etch rates of HfO2 thin films in pure Ar and CF4 were 35.78 and 14.75 nm/min, respectively. The maximum etch rate was 54.48 nm/min when 80% Ar was added to CF4/Ar plasma. The etch rate of SiO2 were also increased with increasing CF4 content in the CF4/Ar plasma mixture, so that the selectivity of HfO2 to SiO2 is about 0.16. Generally, the non-monotonic etch rate of HfO2 can result from at least two effects connected with both volume and surface chemistries. The first effect is the corresponding nonmonotonic behaviors of volume densities and fluxes of active species due to the influence of gas mixing ratio on their formatione decay kinetics. However, such explanation contradicts to a numerous experimental and modeling results for the CF4/Ar ICP (for example, Refs. [17,18]) demonstrating monotonic changes in gas phase composition when the gas mixing ratio is varied at constant pressure and input power. That is why, from the point of view of the etch mechanism proposed above and taking into account the results of Refs. [19,20], we assume that the maximum etch rate is connected with the effects of surface chemistry and can be explained by the concurrence of chemical and physical pathway in the chemical reaction. Particularly, when the CF4 mixing ratio in the CF4/Ar mixture is increased, there are two factors working in opposite directions, which are the increasing flux of F atoms on the etched surface and the decreasing fraction of free surface acceptable for chemical reaction [16]. The last fact is connected with a decrease in the efficiency of ion-stimulated desorption of reaction products and can be understood from the data on ion densities and fluxes in the CF4/Ar plasma reported in Ref. [17]. Considering the etch rate of HfO2, all subsequent experiments for the etching of HfO2 films were carried out with the etch gas of 20% CF4 in CF4/Ar plasma. This assumption is confirmed by the direct measurements of volume densities of particles in CF4/Ar plasma [18]. Fig. 1(b) shows the effect of RF power on the etch rates of HfO2 thin film under 80% Ar in CF4/Ar plasma. As RF power applied to the ICP coil was raised from 500 to 800 W, the etch rates of the HfO2 films increased from 32.8 to 68.05 nm/min, and the selectivity of HfO2 to SiO2 increased from 0.07 to 0.29. Such behavior of all etch rates with increasing input power may be explained by the acceleration of both physical and chemical etching pathways through the growth of volume densities and fluxes of ions and fluorine atoms [21]. The etch rates of HfO2 thin film and the selectivity of HfO2 to SiO2 are shown in Fig. 1(c) as functions of DC-bias voltage. As the DC-bias voltage increases from 150 to 300 V, the etch rate of HfO2 increases from 16.43 to 110.15 nm/min, respectively. The selectivity of HfO2 to SiO2 was slightly increased. This result is also related to the variation in etch rate dependence on the material. Therefore, the increase in the etch rate is thought to be associated with the increase in the mean ion energy, resulting in an increase in the sputtering yields for both HfO2 thin film and the reaction products. The effect of process pressure on etch rate is shown in Fig. 1(d). As
Fig. 2. The Hf 4f XPS narrow scan spectra of the etched HfO2 thin film surface. The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C. (a) As-deposited, (b) CF4/Ar gas mixing ratio, (c) Pure CF4.
process pressures increases from 5 to 20 mTorr, the etch rates of HfO2 decreases from 58.08 to 41.45 nm/min. However, the further increase in gas pressure up to 20 mTorr results in decreasing HfO2 etch rate in the range of 58.08e41.45 nm/min, so that we obtain a similar non-monotonic behavior as it was mentioned for the effect of gas mixing ratio. The selectivity of HfO2 to SiO2 decreases from 0.24 to 0.09. In our opinion, the effect of gas pressure may be explained as follows. An increase in gas pressure at fixed CF4/Ar mixing ratio leads to an increase in both density and flux of fluorine atoms on the etched surface, but causes a decrease in ion flux and mean ion energy [17]. As a result, with increasing gas pressure, we have a tendency to accelerate in chemical etch pathway, but a worse condition for ion-stimulated desorption of reaction products resulting in decreasing fraction of free surface acceptable for
P.-S. Kang et al. / Vacuum 93 (2013) 50e55
53
peaks corresponding to the HfeOx bond (17.7 0.1 eV), HfeO bonds (17 0.1 eV), and HfeHf bond (16.2 0.1 eV). These binding energies are well agreed with the reported values [22e24]. From these observations, the XPS results can explain the proposed etching mechanism shown in Fig. 1(a)e(d). For the films etched in the fluorine-containing plasmas, the one peak, observed in Fig. 2(b) and (c), did not appear in the binding energy region of Hf 4f. This means that the Hf fluorides are removed during the etching process. It is worthy to note that the peak intensity of HfeO is increased compensating HfeOx regardless etching parameters during etching
Fig. 3. The C 1s XPS narrow scan spectra of the etched HfO2 thin film surface. The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C.
chemical reaction. Similar to the effect of gas mixing ratio discussed above, these two factors are working in opposite direction and produce a non-monotonic behavior of the etch rate. 3.2. Analysis using XPS To analyze the etching mechanism of the HfO2 thin films in detail, the etched surfaces were examined by XPS. Fig. 2 shows the deconvoluted Hf 4f spectra XPS for the as-deposited and etched HfO2 thin films. The peak deconvolution was carried with XPSPEAK after subtraction of the background applying a Shirley model. The XPS spectra show the presence of Hf 4f peaks for Hf 4f5/2 and Hf 4f7/2. Fig. 2(a) shows the XPS narrow spectra obtained from the surface of the HfO2 thin film before etching. Fig. 2(b) and (c) shows the narrow XPS spectra obtained from the surface of the etched HfO2 thin films. Fig. 2(b) shows the XPS spectra after etching HfO2 thin films with the CF4/Ar (¼20:80%) gas at an RF power of 600 W, a DC-bias voltage of 250 V, and a process pressure of 15 mTorr, and a substrate temperature of 40 C. We note that for the asdeposited film, the total Hf 4f peak can be decomposed into three
Fig. 4. The F 1s XPS narrow scan spectra of the etched HfO2 thin film surface. The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C.
Fig. 5. The O 1s XPS narrow scan spectra of the etched HfO2 thin film surface. The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C. (a) As-deposited, (b) CF4/Ar gas mixing ratio, (c) Pure CF4.
54
P.-S. Kang et al. / Vacuum 93 (2013) 50e55
removed by etching process as well. Therefore, the XPS data gave clear confirmation that, in the CF4-containing plasmas, the surface of the etched HfO2 thin film was covered by volatile reaction products such as HfeF. In this situation, the ion-stimulated desorption of the reaction products is assumed to be the limiting stage for the etch process in the CF4-rich plasmas, whereas the etch rate is expected to be rather sensitive to the factors influencing the intensity of ion bombardment. This last assumption is in good agreement with the data shown in Fig. 1. 3.3. FE-SEM analysis
Fig. 6. The cross-sectional SEM image of the HfO2 thin film etched in the CF4/Ar plasma. The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C.
process. These results suggested that the vertical etching profile of the HfO2 thin film was obtained by ICP using F radicals and Ar ions, due to the low volatility of hafnium fluorides, as shown in Fig. 4. The binding energies of the peaks for the HfeO, HfeOx, and HfeHf bonds shifted for high energy with increasing amount of CF4 gas. Therefore, we can relate them to the signal from HfeFx. Fig. 3 shows the high resolution of C 1s XPS spectra obtained from the surface of the HfO2 thin film before etching and from the surface after etching as a function of the CF4/Ar (20:80%) gas mixing ratio at a fixed RF power of 600 W, a DC-bias voltage of 250 V, a process pressure of 2 Pa, and a substrate temperature of 40 C from top to bottom in the figure. After etching in CF4/Ar and CF4 only plasma, the peak intensity of C 1s decreased. This indicated that the polymer was accumulated on the HfO2 surface. Carbon and oxygen peaks were detected because the XPS samples had been exposed in air when they were transferred from etching system to XPS system. The peak at 285 eV was corresponded to CeC, CeO, CeOH or CeF bonds [25,26]. But, as the partial pressure of fluorine increased, the intensity of F 1s increased as shown in Fig. 4. Also, we could identify the F 1s peaks of 684.9 eV. This means that the Hf fluorides are removed during the etching process. But, after etching in CF4/Ar and CF4 only plasma, the peak intensity of F 1s increased and shifted to the higher energy. Therefore, we could consider that the chemical etching mechanism of the HfO2 thin film was changed by CF4 gas. Fig. 5 shows the high resolution O 1s XPS spectra obtained from the surface of HfO2 thin films which are as-deposited and etched at CF4/Ar and pure CF4 plasma. Fig. 5(a) shows the XPS narrow scan spectra obtained from the surface of the HfO2 thin film before etching. Fig. 5(b) shows the O 1s XPS spectra obtained from the HfO2 surface after etching as a function of CF4/Ar (20:80%) gas mixing ratio at an RF power of 600 W, a DC-bias voltage of 250 V, a process pressure of 15 mTorr, and a substrate temperature of 40 C. Fig. 5(c) shows the high resolution of O 1s XPS spectra obtained after etching the HfO2 surface by pure CF4 under the same conditions as those shown in Fig. 5(b). To investigate the chemical states of the surface, the O 1s peaks were deconvoluted. The O 1s peak was deconvoluted into four peaks which were deemed to correspond to the oxygen and/or hydroxyl group on the surface: SurfeO bond (SurfeO, 532.7 0.1 eV), OeO bond (531.7 0.1 eV), OxeHf bond (530.5 0.1 eV), and OeHf bond (529.7 0.1 eV) [27e 29]. For the etched films, the Hf 4f peaks at 17.7 eV and 17 eV dominated over their neighbors, while the intensity of the SurfeO peak at 532.7 eV showed the significant decrease for CF4/Ar (20:80%) etched one. The surface oxygen species were effectively
Fig. 6 shows the cross-sectional SEM images of the HfO2 thin film after etching in CF4/Ar (20:80%) plasma at a fixed RF power of 600 W, the DC-bias voltage of 250 V, the process pressure of 15 mTorr, and the substrate temperature of 40 C. As shown in Fig. 6, the cross-sections of the PR and HfO2 thin films were obtained with a PR mask. After etching in CF4/Ar for 60 s, the sidewall on the photoresist and HfO2 thin film was manufactured using a PR mask. However, a sloped angle of HfO2 was also obtained due to the low etch rate of HfO2 thin film. The experimental results showed that the etching profile of the plasma etched HfO2 thin film was associated with F radicals, due to the low etch rate of the HfO2 thin film. 4. Conclusions Etching characteristics of HfO2 thin films were investigated in terms of etch rate and selectivity using CF4/Ar plasma. Experiments were performed with variations of CF4/Ar gas mixing ratio, DC-bias voltage, and the chamber pressure. It was found that addition of Ar contents up to 80% leads the etch rate of HfO2 to increase in comparison with that of CF4 only. The maximum etch rate of HfO2 films was 54.48 nm/min under 20% CF4/(CF4 þ Ar) in 600 W, 250 V, 15 mTorr, and the substrate temperature of 40 C. The selectivity of HfO2 to SiO2 was 0.16. It showed that the increase of Ar addition enhanced ion bombardment and made the etch rate increase. The chemical states of etched HfO2 films were investigated using XPS and the etching mechanism of HfO2 thin films can be explained as follow. Hf interacted with the F radicals by adding CF4, but it remained at the surface due to non-volatility of HfFx. References [1] Rittersma ZM, Loo JJGP, Ponomarev YV, Verheijem MA, Kaiser M, Roozeboom F, et al. J Electrochem Soc 2004;151:G870. [2] Niwa M, Buturi Oyo. Jpn Soc Appl Phys 2003;72:1143. [3] Kang H, Roh Y, Bae G, Jun D, Yang CW. J Vac Sci Technol B 2002;20:1360. [4] Quevedo-Lopez MA, El-Bouanani M, Wallace RM, Gnade BE. J Vac Sci Technol 2002;A 60:1891. [5] Hamada D, Osari K, Nakamura K, Eriguchi K, Ono K, Oosawa M, et al, 6th Proc Int Symp Dry Process, (2006) p. 11. [6] Gevers PM, Beijerinck HCW, Van De Sanden MCM, Kessels WMM. J Appl Phys 2008;103:083304. [7] Ko JH, Kim DY, Park MS, Lee NE, Lee SS, Ahn J, et al. J Vac Sci Technol A 2007; 25:990. [8] Shin MH, Na SW, Lee NE, Oh TK, Kim JY, Lee TH, et al. Jpn J Appl Phys 2005;44: 5811. [9] Kim GH, Kim KT, Woo JC, Kim CI. Ferroelectrics 2007;357:41. [10] Kim DP, Kim GH, Woo JC, Kim HJ, Kim CI, Lee CI, et al. J Kor Phys Soc 2009; 54:934. [11] Kim DP, Kim GH, Woo JC, Yang X, Um DS, Kim CI. Ferroelectrics 2009;381:30. [12] Norasetthekul S, Park PY, Baik KH, Lee KP, Shin JH, Jeong BS, et al. Appl Surf Sci 2002;187:75. [13] Kim MK, Efremov AM, Lee HW, Park HH, Hong MP, Min NK, et al. Thin Solid Films 2011;519:6708. [14] Sungauer E, Mellhaoui X, Pargon E, Joubert O. Microelectro Eng 2009;86:965. [15] Nakamura K, Kitagawa T, Osari K, Takahashi K, Ono K. Vacuum 2006;80:761. [16] Vugts MJM, Hermans LJF, Beijerinck HCW. J Vac Sci Technol A 1996;14:2138. [17] Efremov AM, Kim DP, Kim CI. J Vac Sci Technol B 2004;75:133. [18] Choi CJ, Kown OS, Seol YS, Kim YW, Choi IH. J Vac Sci Technol B 2002;18:811.
P.-S. Kang et al. / Vacuum 93 (2013) 50e55 [19] Efremov AM, Koo SM, Kim DP, Kim KT, Kim CI. J Vac Sci Technol A 2004;22: 2101. [20] Efremov AM, Kim DP, Kim CI. IEEE Trans Plasma Sci 2004;32:1344. [21] Wuu DS, Horng RH, Huang SH, Chung CR. J Vac Sci Technol A 2002;20:766. [22] Jud E, Tang M, Chiang Y-M. J Appl Phys 2008;103:114108. [23] Jin H, Oh UK, Kang HJ, Lee SW, Lee YS, Lim KY. J Kor Phys Soc 2005;46:S52. [24] Tan R, Azuma Y, Kojima I. Appl Surf Sci 2005;241:135.
55
[25] Bertóti I. Surf Coat Technol 2002;151e152:194. [26] Kirchner CN, Hallmeier KH, Szargan R, Raschke T, Radehaus C, Wittstock G. Electroanalysis 2007;19:1023. [27] Feng L-P, Liu Z-T, Shen Y-M. Vacuum 2009;83:902. [28] Zhang L, Terauchi S-Y, Azuma Y, Fujimoto T. Surf Interface Anal 2008; 40:1701. [29] Cheng X, Wan L, Song Z, Yu Y, Shen D. Appl Phys Lett 2007;90:152910.
Contents lists available at SciVerse ScienceDirect
Vacuum journal homepage: www.elsevier.com/locate/vacuum
Etch characteristics of HfO2 thin films by using CF4/Ar inductively coupled plasma Pil-Seung Kang, Jong-Chang Woo, Young-Hee Joo, Chang-Il Kim* School of Electrical and Electronics Engineering, Chung-Ang University, 221 Heukseok-Dong, Dongjak-Gu, Seoul 156-756, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 November 2012 Received in revised form 24 December 2012 Accepted 27 December 2012
In this study, we carried out an investigation of the etching characteristics (etch rate, selectivity) of HfO2 thin films in the CF4/Ar inductively coupled plasma (ICP). The maximum etch rate of 54.48 nm/min for HfO2 thin films was obtained at CF4/Ar (¼20:80%) gas mixing ratio. At the same time, the etch rate was measured as function of the etching parameters such as ICP RF power, DC-bias voltage, and process pressure. The X-ray photoelectron spectroscopy analysis showed an efficient destruction of the oxide bonds by the ion bombardment as well as an accumulation of low volatile reaction products on the etched surface. Based on these data, the chemical reaction was proposed as the main etch mechanism for the CF4-containing plasmas. Ó 2013 Elsevier Ltd. All rights reserved.
Keywords: Etching HfO2 XPS ICP CF4
1. Introduction The complementary metal oxide semiconductor transistor scaling beyond the present 45 nm technology nodes makes it difficult to grow high quality ultra-thin oxide. The thickness of gate dielectrics, SiO2, should be reduced down to 2 nm or less. The thickness reduction of SiO2 brings many serious problems such as increased gate leakage current and reduced oxide reliability. To overcome this drawback, many metal oxides with high dielectric constant materials have been reported, such as TiO2, Ta2O5, SrTiO3 (STO), Al2O3, Y2O3, ZrO2, and BaSrTiO3 (BST) [1e4]. Although these materials have high dielectric constant, some of these fail one or more of the criteria. Al2O3 and Y2O3 did not provide sufficient advantages over SiO2 or Si3N4. TiO2 and Ta2O5 were observed to react with Si substrate. STO or BST may cause fringing field-induced barrier lowering effect. Among high-k materials compatible with silicon, oxides of Zr and Hf are attracting much attention recently because Hf forms the most stable oxide with the highest heat of formation. Additionally, Hf can reduce the native oxide layer to form HfO2. However, new gate oxide candidates must satisfy a standard processing procedure. In order to solve this problem, an understanding of the relations between the etch characteristics of HfO2 thin film and plasma properties is required for the low damaged removal process. * Corresponding author. Tel.: þ82 2 820 5334; fax: þ82 2 812 9651. E-mail address: [email protected] (C.-I. Kim). 0042-207X/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.vacuum.2012.12.007
Until now, there are only few works devoted to the investigations of etching characteristics of the HfO2 thin films using both fluorine- and chlorine-based plasma chemistries. Refs. [5e8] show the dependences of HfO2 etch rate on operating conditions for O2/BCl3, Cl2/BCl3, XeF/Ar, BCl3/C4F8/Ar, and C12/SF6/Ar plasmas, but do not discuss the etch mechanisms as well as the relationships between process parameters, and chemistry. The others are our works where the mechanism of HfO2 in Cl2/Ar, CH4/Ar, HBr/Ar, Cl2/BCl3/Ar and BCl3/Ar plasmas was analyzed using the combination of modeling and diagnostics tools [9e15]. In this study, we investigated the etch characteristics of HfO2 thin films in inductively coupled plasma (ICP) system with CF4/Ar gas chemistries. Etching characteristics on HfO2 thin films have been investigated in terms of etch rate and selectivity. Also, the etch rate and selectivity were measured as function of the etching parameters such as inductively coupled plasma RF power, DC-bias voltage, and process pressure. The chemical binding states in the surface of the etched HfO2 thin films were investigate with X-ray photoelectron spectroscopy (XPS). Scanning electron microscopy (SEM) was used to investigate the surface morphology of Hf-based high-k dielectrics exposed in plasma and etching profile. 2. Experimental The HfO2 thin films were prepared on an 8-inch Si (100) substrate by using plasma enhanced atomic layer deposition (PEALD). The PEALD system (ASM Genitech Korea Ltd. Stella 3000) can
P.-S. Kang et al. / Vacuum 93 (2013) 50e55
51
Fig. 1. Etch rates of HfO2 thin films as a function of the CF4/Ar gas mixing ratio. (a) The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C. (b) The DC-bias voltage was maintained at 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C. (c) The RF power was maintained at 600 W, the process pressure was 15 mTorr, and the substrate temperature was 40 C. (d) The RF power was maintained at 600 W, the DCbias voltage was 250 V, and the substrate temperature was 40 C.
deposit three kinds of binary oxides and their mixed materials. Three canisters of Ti, Al, and Hf metal source are contained in the system. The canister for Hf metal source (TEMAHf) with the volume of 1000 cc can be heated up to 65 C and delivery line for Hf metal source can be heated to 100 C surrounded with heating tapes to prevent condensation of precursor in the delivery line. The source in this canister is delivered to the main reactor chamber with Ar carrier gas instead of conventional bubbler system. The oxygen source used in this experiment is the O2 plasma, which is generated by RF plasma generator. The upper electrode could apply RF power with 13.56 MHz RF plasma source to the substrate that was grounded with capacitive. The line for Hf precursor was heated at the temperature 100 C by heating tapes. The flow rate of Ar purge was fixed at 200 sccm and the chamber pressure was maintained at 3 Torr during deposition process. The final thickness of HfO2 thin films was about 100 nm. This is schematically shown in Ref. [9]. The reactor consists of cylindrical aluminum anodized chamber with diameter of 26 cm. A 13.56 MHz RF power generator was connected through a matching network to the coil and generate plasma on the top of the process chamber. Another 13.56 MHz RF power generator was applied to the substrate to induce the DC-bias voltage to the wafer. Wafers were placed on a chuck, which is used as the bottom electrode. The chamber was evacuated to 106 Torr using a mechanical pump and a turbo-molecular pump. The HfO2 thin films were etched with CF4/Ar gas. The gas mixing ratio was varied to find the characteristics of etching. For these experiments, process pressure, RF power, bottom DC-bias voltage, and substrate temperature were 15 mTorr, 600 W, 250 V and 40 C, respectively. In
addition, plasma etching of HfO2 thin films was investigated by changing the etching parameter including top RF power of 500e 800 W, DC-bias voltage to the substrate of 150 to e300 V, and process pressure of 5e20 mTorr, and the gas flow rate of 20 sccm in the CF4/Ar gas mixing ratio, respectively. The etch rate was measured by the surface profiler (KLA Tencor, a-step 500). The compositional changes on the etched HfO2 surface were investigated using X-ray photoelectron spectroscopy (SIGMA PROBE, Thermo VG Scientific). The spectra were plotted by counting the photoelectrons at kinetic energy intervals of 0.1 eV. The detailed information on the inner region of the film was provided by the spectra recorded at the electron take-off angles of 90 . All of the samples for the XPS analysis were bare HfO2 thin films that did not have any photoresist patterns, and the size of the samples was 1 1 cm2. The etching time was 10 s. The etching profile of the cross-section was characterized using FE-SEM (Sirion 400, FEI). The HfO2 thin films used for measuring the etch rate and etching profile did have a photoresist pattern (SS03A9). The width and thickness of the PR pattern were 1.5 and 1.02 mm, respectively. 3. Result and discussion It is well known that the etch mechanism is not only influenced by the processing parameters for any plasma etching process but also depended strongly on the types of both chemically active species and surface atoms determining the volatility of reaction products. It can be understood that the magnesium fluorides are very low volatile compounds, so that the thermal desorption of etch
52
P.-S. Kang et al. / Vacuum 93 (2013) 50e55
products, and thus, the spontaneous etching can be neglected. Therefore, the HfO2 etch properties in the CF4/Ar plasma can be preliminary proposed as the chemical reaction where the role of ion bombardment includes three main effects: 1) sputtering of the main material; 2) sputtering (ion-stimulated desorption) of the etch products to provide the access for the F atoms to the etched surface; and 3) destruction of the HfeO bonds to provide the chemical interaction of F atoms with Hf. The evidence of the last pathway follows from the fact that the HfeF bond is weaker than the HfeO, so that the direct interaction between HfO2 and F atoms seems to be impossible. 3.1. Effect of gas mixing ratio Fig. 1(a) represents the etch rate of HfO2 thin film and the selectivity of HfO2 to SiO2 as a function of CF4/Ar gas mixing ratio. The etch rates of HfO2 thin films in pure Ar and CF4 were 35.78 and 14.75 nm/min, respectively. The maximum etch rate was 54.48 nm/min when 80% Ar was added to CF4/Ar plasma. The etch rate of SiO2 were also increased with increasing CF4 content in the CF4/Ar plasma mixture, so that the selectivity of HfO2 to SiO2 is about 0.16. Generally, the non-monotonic etch rate of HfO2 can result from at least two effects connected with both volume and surface chemistries. The first effect is the corresponding nonmonotonic behaviors of volume densities and fluxes of active species due to the influence of gas mixing ratio on their formatione decay kinetics. However, such explanation contradicts to a numerous experimental and modeling results for the CF4/Ar ICP (for example, Refs. [17,18]) demonstrating monotonic changes in gas phase composition when the gas mixing ratio is varied at constant pressure and input power. That is why, from the point of view of the etch mechanism proposed above and taking into account the results of Refs. [19,20], we assume that the maximum etch rate is connected with the effects of surface chemistry and can be explained by the concurrence of chemical and physical pathway in the chemical reaction. Particularly, when the CF4 mixing ratio in the CF4/Ar mixture is increased, there are two factors working in opposite directions, which are the increasing flux of F atoms on the etched surface and the decreasing fraction of free surface acceptable for chemical reaction [16]. The last fact is connected with a decrease in the efficiency of ion-stimulated desorption of reaction products and can be understood from the data on ion densities and fluxes in the CF4/Ar plasma reported in Ref. [17]. Considering the etch rate of HfO2, all subsequent experiments for the etching of HfO2 films were carried out with the etch gas of 20% CF4 in CF4/Ar plasma. This assumption is confirmed by the direct measurements of volume densities of particles in CF4/Ar plasma [18]. Fig. 1(b) shows the effect of RF power on the etch rates of HfO2 thin film under 80% Ar in CF4/Ar plasma. As RF power applied to the ICP coil was raised from 500 to 800 W, the etch rates of the HfO2 films increased from 32.8 to 68.05 nm/min, and the selectivity of HfO2 to SiO2 increased from 0.07 to 0.29. Such behavior of all etch rates with increasing input power may be explained by the acceleration of both physical and chemical etching pathways through the growth of volume densities and fluxes of ions and fluorine atoms [21]. The etch rates of HfO2 thin film and the selectivity of HfO2 to SiO2 are shown in Fig. 1(c) as functions of DC-bias voltage. As the DC-bias voltage increases from 150 to 300 V, the etch rate of HfO2 increases from 16.43 to 110.15 nm/min, respectively. The selectivity of HfO2 to SiO2 was slightly increased. This result is also related to the variation in etch rate dependence on the material. Therefore, the increase in the etch rate is thought to be associated with the increase in the mean ion energy, resulting in an increase in the sputtering yields for both HfO2 thin film and the reaction products. The effect of process pressure on etch rate is shown in Fig. 1(d). As
Fig. 2. The Hf 4f XPS narrow scan spectra of the etched HfO2 thin film surface. The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C. (a) As-deposited, (b) CF4/Ar gas mixing ratio, (c) Pure CF4.
process pressures increases from 5 to 20 mTorr, the etch rates of HfO2 decreases from 58.08 to 41.45 nm/min. However, the further increase in gas pressure up to 20 mTorr results in decreasing HfO2 etch rate in the range of 58.08e41.45 nm/min, so that we obtain a similar non-monotonic behavior as it was mentioned for the effect of gas mixing ratio. The selectivity of HfO2 to SiO2 decreases from 0.24 to 0.09. In our opinion, the effect of gas pressure may be explained as follows. An increase in gas pressure at fixed CF4/Ar mixing ratio leads to an increase in both density and flux of fluorine atoms on the etched surface, but causes a decrease in ion flux and mean ion energy [17]. As a result, with increasing gas pressure, we have a tendency to accelerate in chemical etch pathway, but a worse condition for ion-stimulated desorption of reaction products resulting in decreasing fraction of free surface acceptable for
P.-S. Kang et al. / Vacuum 93 (2013) 50e55
53
peaks corresponding to the HfeOx bond (17.7 0.1 eV), HfeO bonds (17 0.1 eV), and HfeHf bond (16.2 0.1 eV). These binding energies are well agreed with the reported values [22e24]. From these observations, the XPS results can explain the proposed etching mechanism shown in Fig. 1(a)e(d). For the films etched in the fluorine-containing plasmas, the one peak, observed in Fig. 2(b) and (c), did not appear in the binding energy region of Hf 4f. This means that the Hf fluorides are removed during the etching process. It is worthy to note that the peak intensity of HfeO is increased compensating HfeOx regardless etching parameters during etching
Fig. 3. The C 1s XPS narrow scan spectra of the etched HfO2 thin film surface. The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C.
chemical reaction. Similar to the effect of gas mixing ratio discussed above, these two factors are working in opposite direction and produce a non-monotonic behavior of the etch rate. 3.2. Analysis using XPS To analyze the etching mechanism of the HfO2 thin films in detail, the etched surfaces were examined by XPS. Fig. 2 shows the deconvoluted Hf 4f spectra XPS for the as-deposited and etched HfO2 thin films. The peak deconvolution was carried with XPSPEAK after subtraction of the background applying a Shirley model. The XPS spectra show the presence of Hf 4f peaks for Hf 4f5/2 and Hf 4f7/2. Fig. 2(a) shows the XPS narrow spectra obtained from the surface of the HfO2 thin film before etching. Fig. 2(b) and (c) shows the narrow XPS spectra obtained from the surface of the etched HfO2 thin films. Fig. 2(b) shows the XPS spectra after etching HfO2 thin films with the CF4/Ar (¼20:80%) gas at an RF power of 600 W, a DC-bias voltage of 250 V, and a process pressure of 15 mTorr, and a substrate temperature of 40 C. We note that for the asdeposited film, the total Hf 4f peak can be decomposed into three
Fig. 4. The F 1s XPS narrow scan spectra of the etched HfO2 thin film surface. The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C.
Fig. 5. The O 1s XPS narrow scan spectra of the etched HfO2 thin film surface. The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C. (a) As-deposited, (b) CF4/Ar gas mixing ratio, (c) Pure CF4.
54
P.-S. Kang et al. / Vacuum 93 (2013) 50e55
removed by etching process as well. Therefore, the XPS data gave clear confirmation that, in the CF4-containing plasmas, the surface of the etched HfO2 thin film was covered by volatile reaction products such as HfeF. In this situation, the ion-stimulated desorption of the reaction products is assumed to be the limiting stage for the etch process in the CF4-rich plasmas, whereas the etch rate is expected to be rather sensitive to the factors influencing the intensity of ion bombardment. This last assumption is in good agreement with the data shown in Fig. 1. 3.3. FE-SEM analysis
Fig. 6. The cross-sectional SEM image of the HfO2 thin film etched in the CF4/Ar plasma. The RF power was maintained at 600 W, the DC-bias voltage was 250 V, the process pressure was 15 mTorr, and the substrate temperature was 40 C.
process. These results suggested that the vertical etching profile of the HfO2 thin film was obtained by ICP using F radicals and Ar ions, due to the low volatility of hafnium fluorides, as shown in Fig. 4. The binding energies of the peaks for the HfeO, HfeOx, and HfeHf bonds shifted for high energy with increasing amount of CF4 gas. Therefore, we can relate them to the signal from HfeFx. Fig. 3 shows the high resolution of C 1s XPS spectra obtained from the surface of the HfO2 thin film before etching and from the surface after etching as a function of the CF4/Ar (20:80%) gas mixing ratio at a fixed RF power of 600 W, a DC-bias voltage of 250 V, a process pressure of 2 Pa, and a substrate temperature of 40 C from top to bottom in the figure. After etching in CF4/Ar and CF4 only plasma, the peak intensity of C 1s decreased. This indicated that the polymer was accumulated on the HfO2 surface. Carbon and oxygen peaks were detected because the XPS samples had been exposed in air when they were transferred from etching system to XPS system. The peak at 285 eV was corresponded to CeC, CeO, CeOH or CeF bonds [25,26]. But, as the partial pressure of fluorine increased, the intensity of F 1s increased as shown in Fig. 4. Also, we could identify the F 1s peaks of 684.9 eV. This means that the Hf fluorides are removed during the etching process. But, after etching in CF4/Ar and CF4 only plasma, the peak intensity of F 1s increased and shifted to the higher energy. Therefore, we could consider that the chemical etching mechanism of the HfO2 thin film was changed by CF4 gas. Fig. 5 shows the high resolution O 1s XPS spectra obtained from the surface of HfO2 thin films which are as-deposited and etched at CF4/Ar and pure CF4 plasma. Fig. 5(a) shows the XPS narrow scan spectra obtained from the surface of the HfO2 thin film before etching. Fig. 5(b) shows the O 1s XPS spectra obtained from the HfO2 surface after etching as a function of CF4/Ar (20:80%) gas mixing ratio at an RF power of 600 W, a DC-bias voltage of 250 V, a process pressure of 15 mTorr, and a substrate temperature of 40 C. Fig. 5(c) shows the high resolution of O 1s XPS spectra obtained after etching the HfO2 surface by pure CF4 under the same conditions as those shown in Fig. 5(b). To investigate the chemical states of the surface, the O 1s peaks were deconvoluted. The O 1s peak was deconvoluted into four peaks which were deemed to correspond to the oxygen and/or hydroxyl group on the surface: SurfeO bond (SurfeO, 532.7 0.1 eV), OeO bond (531.7 0.1 eV), OxeHf bond (530.5 0.1 eV), and OeHf bond (529.7 0.1 eV) [27e 29]. For the etched films, the Hf 4f peaks at 17.7 eV and 17 eV dominated over their neighbors, while the intensity of the SurfeO peak at 532.7 eV showed the significant decrease for CF4/Ar (20:80%) etched one. The surface oxygen species were effectively
Fig. 6 shows the cross-sectional SEM images of the HfO2 thin film after etching in CF4/Ar (20:80%) plasma at a fixed RF power of 600 W, the DC-bias voltage of 250 V, the process pressure of 15 mTorr, and the substrate temperature of 40 C. As shown in Fig. 6, the cross-sections of the PR and HfO2 thin films were obtained with a PR mask. After etching in CF4/Ar for 60 s, the sidewall on the photoresist and HfO2 thin film was manufactured using a PR mask. However, a sloped angle of HfO2 was also obtained due to the low etch rate of HfO2 thin film. The experimental results showed that the etching profile of the plasma etched HfO2 thin film was associated with F radicals, due to the low etch rate of the HfO2 thin film. 4. Conclusions Etching characteristics of HfO2 thin films were investigated in terms of etch rate and selectivity using CF4/Ar plasma. Experiments were performed with variations of CF4/Ar gas mixing ratio, DC-bias voltage, and the chamber pressure. It was found that addition of Ar contents up to 80% leads the etch rate of HfO2 to increase in comparison with that of CF4 only. The maximum etch rate of HfO2 films was 54.48 nm/min under 20% CF4/(CF4 þ Ar) in 600 W, 250 V, 15 mTorr, and the substrate temperature of 40 C. The selectivity of HfO2 to SiO2 was 0.16. It showed that the increase of Ar addition enhanced ion bombardment and made the etch rate increase. The chemical states of etched HfO2 films were investigated using XPS and the etching mechanism of HfO2 thin films can be explained as follow. Hf interacted with the F radicals by adding CF4, but it remained at the surface due to non-volatility of HfFx. References [1] Rittersma ZM, Loo JJGP, Ponomarev YV, Verheijem MA, Kaiser M, Roozeboom F, et al. J Electrochem Soc 2004;151:G870. [2] Niwa M, Buturi Oyo. Jpn Soc Appl Phys 2003;72:1143. [3] Kang H, Roh Y, Bae G, Jun D, Yang CW. J Vac Sci Technol B 2002;20:1360. [4] Quevedo-Lopez MA, El-Bouanani M, Wallace RM, Gnade BE. J Vac Sci Technol 2002;A 60:1891. [5] Hamada D, Osari K, Nakamura K, Eriguchi K, Ono K, Oosawa M, et al, 6th Proc Int Symp Dry Process, (2006) p. 11. [6] Gevers PM, Beijerinck HCW, Van De Sanden MCM, Kessels WMM. J Appl Phys 2008;103:083304. [7] Ko JH, Kim DY, Park MS, Lee NE, Lee SS, Ahn J, et al. J Vac Sci Technol A 2007; 25:990. [8] Shin MH, Na SW, Lee NE, Oh TK, Kim JY, Lee TH, et al. Jpn J Appl Phys 2005;44: 5811. [9] Kim GH, Kim KT, Woo JC, Kim CI. Ferroelectrics 2007;357:41. [10] Kim DP, Kim GH, Woo JC, Kim HJ, Kim CI, Lee CI, et al. J Kor Phys Soc 2009; 54:934. [11] Kim DP, Kim GH, Woo JC, Yang X, Um DS, Kim CI. Ferroelectrics 2009;381:30. [12] Norasetthekul S, Park PY, Baik KH, Lee KP, Shin JH, Jeong BS, et al. Appl Surf Sci 2002;187:75. [13] Kim MK, Efremov AM, Lee HW, Park HH, Hong MP, Min NK, et al. Thin Solid Films 2011;519:6708. [14] Sungauer E, Mellhaoui X, Pargon E, Joubert O. Microelectro Eng 2009;86:965. [15] Nakamura K, Kitagawa T, Osari K, Takahashi K, Ono K. Vacuum 2006;80:761. [16] Vugts MJM, Hermans LJF, Beijerinck HCW. J Vac Sci Technol A 1996;14:2138. [17] Efremov AM, Kim DP, Kim CI. J Vac Sci Technol B 2004;75:133. [18] Choi CJ, Kown OS, Seol YS, Kim YW, Choi IH. J Vac Sci Technol B 2002;18:811.
P.-S. Kang et al. / Vacuum 93 (2013) 50e55 [19] Efremov AM, Koo SM, Kim DP, Kim KT, Kim CI. J Vac Sci Technol A 2004;22: 2101. [20] Efremov AM, Kim DP, Kim CI. IEEE Trans Plasma Sci 2004;32:1344. [21] Wuu DS, Horng RH, Huang SH, Chung CR. J Vac Sci Technol A 2002;20:766. [22] Jud E, Tang M, Chiang Y-M. J Appl Phys 2008;103:114108. [23] Jin H, Oh UK, Kang HJ, Lee SW, Lee YS, Lim KY. J Kor Phys Soc 2005;46:S52. [24] Tan R, Azuma Y, Kojima I. Appl Surf Sci 2005;241:135.
55
[25] Bertóti I. Surf Coat Technol 2002;151e152:194. [26] Kirchner CN, Hallmeier KH, Szargan R, Raschke T, Radehaus C, Wittstock G. Electroanalysis 2007;19:1023. [27] Feng L-P, Liu Z-T, Shen Y-M. Vacuum 2009;83:902. [28] Zhang L, Terauchi S-Y, Azuma Y, Fujimoto T. Surf Interface Anal 2008; 40:1701. [29] Cheng X, Wan L, Song Z, Yu Y, Shen D. Appl Phys Lett 2007;90:152910.