22nd International Symposium on Plasma Chemistry July 5-10, 2015; Antwerp, Belgium Low-k OSG damage and etching by F atoms at lowered temperatures S. Zyryanov1,2, K. Kurchikov1,2, D. Lopaev1, Yu. Mankelevich1, A. Palov1, T. Rakhimova1, E. Voronina1, N. Novikova3 and M. Baklanov4 1 Skobeltsyn Institute of Nuclear Physics, Lomonosov Moscow State University, RU-119991 Moscow, Russia 2 Faculty of Physics, Moscow State University, Moscow, Russia 3 Institute for Spectroscopy of Russian Academy of Sciences, Troitsk, Russia 4 IMEC vzw, BE-3001 Leuven-Heverlee, Belgium Abstract: Damage and etching by F atoms was studied in ICP downstream reactor at temperatures from +15 to -30°C for low-k OSG films. It is shown that OSG exposure to fluorine atoms leads to film fluorination, H abstraction reactions from –CH 3 groups with formation of –CH x F y species and film etching. Activation energy for H atom replacement by F atom in -CH 3 was estimated using the experimental data. Keywords: low-k SiOCH materials, low-k damage, etching, fluorine atoms 1. Introduction Ultra low-k dielectric materials are the key component of the microchips with structure size of 20 nm and below. Their decreased dielectric constant compared to SiO 2 allows decreasing the signal propagation delay as well as preventing the interconnect cross-talk and power dissipation. Most of the modern low-k materials are porous organosilicate glasses (OSG). They are deposited on the substrate by means of PECVD and spin-on technologies. Typical pore radius is 1-2.5 nm with porosity from 20 to 50% while the greater pore radius is typical for highporosity films. One of the main problems of low-k material processing in microchip manufacturing process is plasma damage [1]. During plasma treatment such as reactive ion etching the low-k film structure may be seriously modified and owing to that dielectric constant increases. The film becomes hydrophilic, which leads to the increase of k value due to water absorption as soon as the film is exposed to the atmosphere. Chemically active radicals [2-4], energetic positive ions [5,6] and VUV photons [7-10] are the reasons for low-k material plasma damage. Their interaction with low-k material can be both separate and synergetic. For example radicals can be adsorbed on the pore walls and then chemical reactions can be stimulated by VUV radiation and ions from plasma. To understand the damage mechanisms the impacts of radicals, ions and VUV photons have to be studied separately. This work is devoted to study low-k OSG damage and etching by F atoms [11,12] at the lowered temperatures. Temperature variation allows estimating the activation energy of chemical reactions that lead to low-k film damage during plasma treatment. 2. Experiment Low-k films of five different types were used in this study. Their parameters are represented in Table 1. CVD1 P-II-5-22 sample is a previous generation low-k film, now used in industry and so well studied. It is used in this study as a reference sample. ALK B is a PECVD ULK material. SOG1.8 – SOG2.2 are spin-on ULK films with different pore radius and porosity. This set of the films is quite adequate to understand the dependency of the damage and etching on pore geometry. Table 1.Low-k films. Name CVD1 ALK B SOG2.2 SOG2.0 SOG1.8 Type PECVD PECVD SOG SOG SOG Porosity (%) 24 40 39 44 51 Pore radius (nm) 0.8 1.5 2.1 2.6 2.7 Thckness (nm) 175 270 218 217 214 k value 2.5 2.1 2.2 2.0 1.8 SF6, 100 mTorr 13.56 MHz ICP plasma 200 W Grid to stop plasma Downstream zone F Screen to stop VUV emission Peltier cooled sample holder Pump 1 3. Low-k film damage The internal structure of the low-k film consists of the interconnected pores, formed by the Si-O-Si framework. Pore walls are covered by –CH 3 bonded to surface Si atoms. Therefore the damage during fluorine atom treatment is mainly connected with the partial extraction of the methyl groups from the pore volume as well as the replacement of hydrogen atoms by fluorine atoms. These effects can be seen in FTIR spectra of the film by measuring the decrease of the Si-CH 3 peak at 1275 cm-1. The dynamics of the –CH 3 density in low-k film during the treatment at various temperatures is shown in Fig 2. It can be clearly seen that the reaction rate is lowered with decreasing temperature. The higher porosity and pore radius promote the deeper penetration of fluorine atoms inside the film and therefore the remaining fraction of SiCH 3 chemical bonds in SOG2.0 with higher porosity is significantly lower than for CVD1 at the same F atom dose. According to XPS data shown in Fig. 3 methyl groups are not removed as a whole. Instead they are transformed from Si-CH 3 to Si-CH 2 F and then to Si-CHF 2 before being extracted. The activation energy for H abstraction reaction was estimated comparing the Si-CH 3 depletion rate for different temperatures. This comparison is shown in Fig. 4. The Arrhenius plot (dashed line) gives an estimation 2 1.2 Fraction of Si-CH3 CVD1 1.0 0.8 -30C -15C 0.6 0C +15C 0.4 0.2 0.0 0 1000 2000 3000 4000 5000 6000 F exposure time, s (F dose) 1.2 SOG2.0 Fraction of Si-CH3 All of the samples, listed above were treated by fluorine atoms in the downstream ICP chamber, shown in Fig. 1. The fluorine atom source was 13.56 MHz, 200 W ICP discharge in SF 6 at 100 mTorr. The plasma region was separated from the downstream region by the stainless steel grid with a small cell of 70x70 μm to prevent plasma penetration to treated samples. A metal screen was also used to avoid VUV light effect on the samples. As a result low-k films were treated only by chemically active fluorine atoms. Treatment time varied from 225 to 7200 s. Low-k samples were located on the Peltier-cooled holder that allowed temperature variation from +15 to 30°C. To prevent water absorption during cooling and heating of the samples these procedures were carried out in the chamber in SF 6 flow. After the exposure to fluorine atom the treated and pristine samples were analysed by using FTIR spectrometer, spectroscopic ellipsometer, EDX and XPS diagnostics. FTIR data was used to measure the remaining integral fraction of Si-CH 3 chemical bonds in the low-k film as the indicator of the film damage. Ellipsometry was used to measure the film thickness and refractive index to analyse both film damage and etching rate. EDX data was used to measure integral density of C, F, and O atoms in the film while XPS was used to analyse various chemical bonds appearing on the surface during phases of damage and etching. for activation energy in Si-CH 3 reaction with F atoms: ~1500 K. 1.0 0.8 -30C 0.6 -15C 0.4 0C 0.2 +15C 0.0 0 1000 2000 3000 4000 5000 6000 F exposure time, s (F dose) Fig. 2.Si-CH 3 depletion by F atoms 28000 XPS (7200 s), a.u. Fig. 1.ICP downstream reactor CF2 CF3 ALK B 24000 F(1s) F-C 20000 exper.data approximation F-Si 16000 12000 8000 4000 0 680 682 684 686 688 690 692 694 696 Binding energy, eV Fig. 3.XPS spectrum for F on ALK B surface after 7200s treatment by F atoms P-II-5-22 [Si-CH3]treated/[Si-CH3]pristine depletion rate Fig. 5.C, O and F atom density dependency on treatment time (with Si-CH 3 dynamics for comparison) The etching process has two distinct phases. In the first phase the film thickness is approximately constant while in the second phase it starts to decrease linearly. 1 0.1 Ea=1500+/-200 K CVD1 (24%) ALK B (40%) SOG2.2 (39%) SOG2.0 (44%) SOG1.8 (51%) 0.01 1.05 1.10 1.15 1.20 1.25 300/T, K-1 Fig. 4.Rates of Si-CH 3 depletion in low-k films for different temperatures 4. Low-k film etching However the reaction with –CH 3 groups is not the only effect of fluorine atoms on the low-k films. According to EDX data shown in Fig. 5 the initial step of treatment is the fluorination. It occurs even at small F doses when fluorine atom density in the film does not change significantly until the thickness of the film starts to decrease due to etching. Fig. 6. Evolution of the film thickness (reduced to 200 nm) of OSG films with F dose The dynamics of film thickness is shown in Fig 6. and can be explained in the following way. Fluorine atoms penetrate in the film bulk with fluorine atom density in the top layer of the film being higher due to loss in chemical reactions. Pore walls are etched by fluorine while the etching rate in the top layer is higher due to higher fluorine atom density. When pore walls are completely etched in the uppermost layer of the film, the thickness of the film starts to decrease providing some effective etch rate. This effective etch rate depends on the pore wall etch rate as well as the fluorine atom density profile in the film bulk. This mechanism is illustrated by Fig. 7. F dose = Fd F collisions with walls 1.2 SOG2.0 F etched pores 300 1.0 0.8 200 O 0.6 F C 0.4 0.2 0.0 F atoms (a.u.) Portion of Si-CH3; C and O F 100 SiOx pristine pores Si-CH3 0 1x1020 2x1020 3x1020 2 F atoms dose (at/cm ) P-II-5-22 0 Fig. 7. Porous low-k film etching by F atoms 3 It is worth noting that the etching rate similarly to the damage rate decreases with decreasing temperature. 5. Conclusion Fluorine atom impact on OSG low-k films leads to few effects: fluorination, damage and etching. Fluorination is fastest process leading to formation of surface fluorine complexes on pore walls. Damage is mostly caused by H abstraction reactions generating –CH x F y species (instead of –CH 2 and –CH 3 ) on the pore surface. Etching takes place not only at the surface of the film but also in the film bulk. Therefore the effective etch rate is higher for films with the deeper fluorine atom penetration. The higher porosity provides the higher area of open surface pores and thereby promotes the higher rates of damage and etching. Both damage and etching rates drop with decreasing temperature that indicates existence of activation energies. The obtained data allowed estimating activation energy in Si-CH 3 reaction with F atoms: ~1500 K. It shows that the low damage by fluorine atoms can be possible only at the temperature close to cryogenic. 6. Acknowledgements This research is supported by SRC program Contract 2012-KJ-2280, RFBR, research project No. 14-02-31599 mol-a and by Optec grant № 28/2014/71-Msk. 4 7. References [1] M.R. Baklanov, J.-F. de Marneffe, D. Shamiryan, et.al. J. Appl. Phys., 113, 041101 (2013) [2] N. Posseme, T. Chevolleau, T. David, M. Darnon, et.al. J. Vac. Sci. Technol. B, 25, 1928 (2007) [3] O.V. Braginsky, A.S. Kovalev, D.V. Lopaev, E.M. Malykhin, et.al. J. Appl. Phys., 108, 073303 (2010) [4] T.V. Rakhimova, O.V. Braginsky, A.S. Kovalev, D.V. Lopaev, et.al. IEEE Transact. Plasma Sci., 37, 1697 (2009) [5] K. Takeda, Y. Miyawaki, S. Takashima, M. Fukasawa, et.al. J. Appl. Phys., 109, 033303 (2011) [6] H. Shi, H. Huang, J. Bao, J. Liu, et.al. J. Vac. Sci. Technol. B, 30, 011206 (2012) [7] J. Lee and D.B. Graves J. Phys. D: Appl. Phys., 44, 325203 (2011) [8] T.V. Rakhimova, A.T. Rakhimov, Y.A. Mankelevich, D.V. Lopaev, et.el. Appl. Phys. Lett., 102, 111902 (2013) [9] T.V. Rakhimova, A.T. Rakhimov, Yu.A. Mankelevich, et.al. J. Phys. D: Appl. Phys., 47, 025102 (2014) [10] J. Shoeb, M.M. Wang and M.J. Kushner J. Vac. Sci. Technol. A, 30, 041303 (2012) [11] Y. Iba, S. Ozaki, M. Sasaki, Y. Kobayashi, et.al. Microelectronic Eng., 87, 451 (2010) [12] T.V. Rakhimova, D.V. Lopaev, Yu.A. Mankelevich, et.al. J.Phys.D: Appl.Phys., in press (2015) P-II-5-22
© Copyright 2026 Paperzz