This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright

Author's personal copy

Microelectronic Engineering 85 (2008) 2164–2168

Contents lists available at ScienceDirect

Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee

Oxygen chemiluminescence in He plasma as a method for plasma damage evaluation A.M. Urbanowicz a,b,*, D. Shamiryan a, M.R. Baklanov a, S. De Gendt a,b a b

IMEC, Kapeldreef 75, B-3001 Leuven, Belgium Department of Chemistry, Katholieke Univ Leuven, B-3001 Leuven, Belgium

a r t i c l e

i n f o

Article history: Received 12 March 2008 Received in revised form 16 March 2008 Accepted 17 March 2008 Available online 26 March 2008 Keywords: Low-k dielectrics Plasma damage Hydrophilisation Porous films Helium plasma

a b s t r a c t We propose a method for evaluating the hydrophilisation degree of low-k films upon plasma damage. The evaluation is based on optical emission spectroscopy analysis of O* emission during He plasma exposure of sample in question. The O* is presumably desorbed from damaged low-k film by vacuum–ultraviolet radiation from He plasma. The new method correlates well with other methods for plasma damage characterization such as Fourier Transform Infrared Spectroscopy and Water–Vapor Ellipsometric Porosimetry. The presented method gives a unique opportunity to assess the degree of hydrophilisation of low-k films immediately after processing. Ó 2008 Elsevier B.V. All rights reserved.

1. Introduction The plasma damage of low-k materials is one of the most important issues during copper/low-k integration into semiconductor processing [1]. The plasma is used for patterning, photo resist stripping and surface cleaning during the integration process flow. The advanced low-k dielectrics needed for high performance integrated circuits are porous in order to decrease the k-value. The open porosity of low-k materials significantly increases diffusivity of the plasma species. Due to this fact porous dielectrics are very sensitive to plasma-induced damage during processing in comparison to conventional non-porous films such as SiO2. In order to avoid degradation of the integrated k-value of low-k materials, both process optimization and development are needed. However, process optimization and development are very time consuming due to current ex situ plasma damage metrology. The plasma damage which occurs during processing such as etch and strip are evaluated using complicated analytical methods such as Fourier Transform Infra Red Spectroscopy (FTIR), Water Vapor Ellipsometric Porosimetry (WEP) and Thermo Desorption Spectroscopy (TDS), etc. [1,2]. Moreover, most of them are not suitable for patterned wafers comprising dense structures (to analyse damage on sidewalls of dense structures). The electrical characterization is used for final examination of the integrated low-k

dielectric quality [2]. However, this measures the sum of all lowk damaging occurred during all integration steps. Furthermore it is expensive and time consuming. Therefore, it is important to develop a simple and non-destructive method, which allows the evaluation of the plasma damage immediately after processing so that the process could be tuned further to avoid plasma damage and/or damaged samples can be eliminated from the process flow. One of the fundamental plasma damage mechanisms of low-k dielectrics is replacement of hydrophobic Si–CH3 groups into hydrophilic Si–OH groups [1]. Further moisture absorption into the porous structure increases the dielectric constant of the material. Therefore, the degree of damage is proportional to the degree of hydrophilisation. In this paper we propose a new method of plasma-damage evaluation based on Optical Emission Spectroscopy (OES) analysis during He plasma exposure of a studied sample [3]. We presume that He plasma exposure causes photolysis of water incorporated into the pores of the low-k material. Products of this reaction can be detected by OES. The analysis of 777 nm peak intensity related to O* radicals gives quantitative information about the degree of damage induced by a plasma process such as strip and etch. To verify the new method we compared it with already established method for plasma-damage evaluation (WEP and FTIR). 2. Experimental

* Corresponding author. Address: IMEC, Kapeldreef 75, B-3001 Leuven, Belgium. Tel.: +32 16 281469; fax: +32 16 281214. E-mail address: [email protected] (A.M. Urbanowicz). 0167-9317/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.mee.2008.03.009

All experiments are carried out in an industrial ICP plasma source (LAM Versys2300 STAR) on 300 mm silicon wafers.

Author's personal copy

A.M. Urbanowicz et al. / Microelectronic Engineering 85 (2008) 2164–2168

The wafer temperature is kept at 30 °C by He backside cooling (electrostatic chuck), while the anodized aluminum (Al2O3) walls of the chamber are kept at 50 °C. The plasma is excited by feeding an antenna (lying on a quartz window) with 13.56 MHz RF power, while the energy of the ions bombarding the wafer is independently controlled by a second 13.56 MHz RF power supply. During all experiments this power is set to 0 W. Therefore ion bombardment has negligible effect on the experiments. To obtain a high reproducibility of the results, the plasma chamber walls are cleaned before every experiment with standard WAC (wafer-less autoclean) recipe containing O2/SF6 plasma [4] and subsequent He plasma cleaning. The additional He plasma cleaning is introduced to remove oxygen from chamber walls [3]. The plasma chamber is equipped with an industrial OES-based endpoint detection system with the sensor located on the side of the reaction chamber. The light emitted by the plasma is directed on a dispersive element followed by a CCD array of 2048 pixels. The OES spectra are recorded in the range of 250–850 nm (with spectral resolution 2.5 nm). The following samples are used (all wafers are 300 mm in diameter): bare silicon wafers, wafers with 1.8 lm of photoresist and wafers with porous low-k dielectrics. The low-k materials used for the damage tests are porous carbon-doped silica films (SiOC:H) with thickness of 180 nm deposited by Plasma Enhanced Chemical Vapor Deposition (PECVD) on silicon substrates. The low-k films have open porosity close to 22% and pore radius 0.8–0.9 nm, as measured by Ellipsometric Porosimetry (EP) [5], dielectric constant in the range of 2.4–2.5 and refractive index of 1.35 at 633 nm. In order to introduce damage, the low-k films are treated with O2/ Cl2 19:1 photoresist strip plasma. The coil power is 1000 W and pressure 20 mTorr. We used varied exposure times of 5, 10, 20, and 30 s. The varied time of treatment is used to induce different level of hydrophilisation to low-k films. For the evaluation of hydrophobic properties of the studied materials we use OES recorded during 20 s of 30 mTorr 400 W (coil power) He plasma exposure for each wafer. The bonding structure of the low-k films before and after plasma treatments was analyzed using a Biorad QS2200 ME FTIR system. The bulk hydrophilicity (amount of adsorbed water) was

2165

measured by using WEP (ellipsometic porosimetry with water source) equipped with SENTECH 801 spectroscopic ellipsometer (k = 350–850 nm) [6].

3. Results We studied OES spectra during He plasma exposure of an empty plasma chamber. We found peaks at 656 nm and 777 nm which are not related to He as shown in Fig. 1. The first intensity might be related to H radicals with optical transitions: 2P ? 2D, 2P ? 2S and 2S ? 2P (7 lines). The second corresponds to 5P ? 5S (3 lines) transition of electronically excited oxygen radicals [7]. It is not obvious why the constant level of 656 nm and 777 nm intensities is observed during He plasma exposure of empty chamber. There might be two possible reasons. Possibly the He gas is contaminated and/or there is diffraction of He lines on the OES dispersive element. However, in our experiments the He peaks are constant while 777 nm peak is changing. Therefore relative change of intensity of O* peak is not related to diffraction of He lines. We presume that the O* and H* might be products of water photodissociation of water molecules adsorbed on chamber walls by vacuum–ultraviolet (VUV) radiation from He plasma. To verify this we studied the OES emission during He plasma exposure of hydrophilic and hydrophobic substrates. We chose the 777 nm peak as more intense that is not overlapped with other He lines. If the presence of O* in He plasma reflects desorption of O* from a substrate, then we should see more O* releasing from the hydrophilic substrate than the hydrophobic one. As a hydrophilic substrate we chose bare Si which is hydrophilic due to a native oxide presence on the surface. The hydrophobic substrate is represented by photoresist deposited on blanket Si which is hydrophobic due to the presence of carbon containing-polymers. We exposed those samples to He plasma for 20 s. The time traces of 777 nm peaks are shown in Fig. 2a. We can see that O intensity in the case of Si (hydrophilic substrate) is higher than the level of the empty chamber. This might be related to photo-dissociation of OH groups from top surface of Si. The intensity of O* for the photoresist (hydrophobic substrate)

Fig. 1. OES of He plasma recorded in empty chamber.

Author's personal copy

2166

A.M. Urbanowicz et al. / Microelectronic Engineering 85 (2008) 2164–2168

Fig. 2. Time traces and OES after 2 s of plasma process (zoomed area): (a) empty chamber, photoresist polymer on the top of Si wafer and bare Si, and (b) empty chamber, pristine low-k, and damaged low-k.

is below the empty chamber level. The results presented in Fig. 2a indicate that we are able to distinguish hydrophobic and hydrophilic substrates by analysis of 777 nm peak intensity during He plasma exposure. Then we extended our study to low-k materials. The damaged low-k samples contain H2O [1] so we should observe higher O* intensity from the damaged sample. We performed the tests on as deposited (pristine) and damaged low-k dielectrics as shown in Fig. 2b. The intensity for hydrophobic as deposited low-k is below the chamber level, as in the case of photoresist exposed to He plasma. In the case of the damaged low-k film (30 s O2/Cl2) we see clear increase of 777 nm peak intensity. The results presented in Fig. 2b indicate that we are able to distinguish between pristine and damaged low-k samples. If the intensity related to O* is proportional to the amount of adsorbed OH groups and water molecules (bonded by hydrogen bridges) then the low-k film more damaged should release more O* and the amount of O* should be proportional to the damage degree. To evaluate the damage degree of the low-k dielectrics, we decided to measure the amount of desorbing O* as a function of damage introduced by O2/Cl2 plasma. The amount of

released O* was estimated using integrated 777 nm intensity using Eq. (1)

Fig. 3. Time of O2/Cl2 plasma treatment of low-k dielectric versus integrated 777 nm intensity.

Author's personal copy

A.M. Urbanowicz et al. / Microelectronic Engineering 85 (2008) 2164–2168

Ii ¼

Z

t¼20 s

IðtÞdt

ð1Þ

t¼0 s

where Ii is the integrated 777 nm peak intensity, t is time of the He plasma exposure. The time traces were recorded as described in Fig. 2 (integral is an area under 777 nm curve). One can see that the amount of desorbed O* increases with increase of the O2/Cl2 treatment time as shown in Fig. 3. We can conclude that by monitoring of the O peak during He plasma treatment we are able to distinguish different degree of damage on the lowk film. In order to compare the OES technique to more conventional technique for estimation of plasma damage, we studied the plasma damaged low-k by FTIR and WEP. The FTIR is used to study changes in the bonding structure. The WEP measures the amount of water adsorbed in the internal pore network of the low-k film. The hydrophilisation is proportional to the amplitude of the Si– OH absorbance band. The amplitude of the Si–OH (3900– 3100 cm1) absorbance increases due to hydrophilic OH group incorporation [8] while the absorbance for C–H (2970 cm1) decreases as a result of carbon depletion as shown in Fig. 4a. The degree of hydrophilisation can be estimated by integrating the OH absorbance band area using Eq. (2) Z k¼3900 cm1 Ai ¼ AðkÞdk ð2Þ k¼3100 cm1

2167

where Ai is the integrated absorbance and k is the wavenumber. For a reference sample (as deposited low-k dielectric) the amplitude of OH groups is close to zero and intensity of C–H peak is maximum. The next curve shows the reference sample exposed to 20 s He plasma which we use for the damage evaluation test. This spectrum reflects the damage introduced by the method itself. The absorbance of OH groups is slightly increased. The intensity of the C–H absorbance band is reduced indicating some carbon depletion. The damage in the pure He plasma is very small in comparison with damage induced by O2-based plasma. Therefore, the damaging effect of the method itself is minimal. The next sample is the low-k exposed to 5 s of O2/Cl2 plasma. In this case, we observe much higher degree of damage than in the case of the sample treated with He plasma only. The highest degree of damage is found in the film which was exposed to 30 s of O2/Cl2 plasma. Another method to estimate the degree of hydrophilisation is WEP. It was found that the amount of water absorbed into dielectric pores at saturated vapor pressure increases with O2/Cl2 exposure time as shown in Fig. 4b. In the case of the pristine sample the amount of absorbed water was around 1.6%. The sample treated for 30 s in O2/Cl2 plasma adsorbs of 22 vol%. water at saturated vapor pressure. The amount of adsorbed water for the sample treated for 30 s is equal to the total porosity due to total hydrophilisation of pore sidewalls. As we can see, both FTIR and WEP techniques are sensitive to hydrophobic properties and therefore, to the damage degree of

Fig. 4. Absorbance of 3900–2900 cm1 as measured by FTIR (a). Percentage amount of water absorbed in low-k volume as measured by WEP (b).

Author's personal copy

2168

A.M. Urbanowicz et al. / Microelectronic Engineering 85 (2008) 2164–2168

Fig. 5. Time of O2/Cl2 plasma exposure of low-k dielectrics versus integrated absorbance (as measured by FTIR) and absorbed water in saturation pressure (as measured by WEP).

as a by-product of reaction of organic hydrophobic groups with O2based plasma, as it was proposed by Chang et al. [9]. The second way is damaged low-k exposure to air which leads to multi-molecular adsorption of water molecules in the internal pore network of the low-k film [2]. The moderately low processing temperature (20–60 °C) and low pressure (20–200 mTorr) during processing is not sufficient to desorb hydrogen bonded water. The part of water molecules remains adsorbed inside the pores of the low-k film. However the adsorbed moisture and OH groups can be photo-desorbed by UV radiation. The water shows an intense UV absorption between 140 nm and 190 nm due to X1A1-A1B1 transition [10]. As the extinction coefficient of SiOC:H low-k film rises sharply at the wavelength below 160 nm [11,12], the film is quite transparent to longer wavelength where water shows high adsorption. The appearance of O* radicals may be related to photolysis of adsorbed water by EUV and VUV photons from He plasma [13] from the bulk of the porous low-k film.

5. Conclusions A new method of evaluating plasma damage is proposed. The method is based on OES analysis during pure He plasma exposure of low-k films. We found peaks that are not related to optical transitions of He but might be related to emission of O* and H* radicals (777 nm and 656 nm, respectively). We presume that the O* is a product of He VUV radiation-induced photo-dissociation of water adsorbed on the reactor walls and on the studied samples. We use a time dependence of 777 nm peak intensity to evaluate the plasma damage of porous low-k materials since the degree of damage is proportional to the amount of water absorbed on the low-k film. It is found that the integrated intensity of O*-related peak gives quantitative information about the degree of damage of the studied low-k films. The obtained results correlate well with FTIR and WEP data. Fig. 6. Integrated time traces (as measured by OES) versus integrated absorbance in range 3900–3100 cm1 (as measured by FTIR) and absorbed water in saturation pressure (as measured by WEP).

the low-k film. They give similar results as shown in Fig. 5 where both integrated absorbance measured by FTIR and amount of adsorbed water measured by WEP are plotted versus O2/Cl2 exposure time. In order to compare the new method (OES) with the established ones (FTIR and WEP) we plotted FTIR and WEP data versus OES data as shown in Fig. 6. One can see linear correlation between OES on one hand and FTIR and WEP on the other hand. We can conclude that our method is suitable for quantitative evaluation of plasma damage. 4. Discussion The exposure of low-k films to O2-based plasma leads to loss of hydrophobic groups [1]. Removal of hydrophobic groups leads to the formation of hydrophilic silanol groups which favor the moisture adsorption. The adsorbed moisture drastically increases the dielectric constant because of the huge k-value of water molecules (k = 78.4 at 100 kHz at 25 °C). The adsorbed water might be formed

References [1] K. Maex, M.R. Baklanov, D. Shamiryan, F. Iacopi, S. Brongersma, Z.Sh. Yanovitskaya, J. Appl. Phys. 93 (11) (2003) 8793–8841. }kei, M. Baklanov, I. Ciofi, Y. Li, A. Urbanowicz, Semicond. Fabtech. 35 [2] Z. To (August) (2007) 110–115. [3] A.M. Urbanowicz, D. Shamiryan, D. Kim, M.R. Baklanov, in: PESM 1st International Workshop, Leuven, 10–11 September, 2007. [4] G. Cunge, B. Pelissier, O. Joubert, R. Ramos, C. Maurice, Plasma Sources Sci. Technol. 14 (2005) 599. [5] M.R. Baklanov, K.P. Mogilnikov, V.G. Polovinkin, F.N. Dultsev, J. Vac. Sci. Technol. B 18 (3) (2000) 1385–1931. [6] M.R. Baklanov, K.P. Mogilnikov, Q.T. Le, Microelectron. Eng. 83 (2006) 2287– 2291. [7] A.R. Striganov, N.S. Sventitskii, Tables of Spectral Lines of Neutral and Ionized Atoms, IFI/Plentun, New York, Washington, 1968, pp. 77–79 and 141–145. [8] A.M. Urbanowicz, A. Humbert, G. Mannaert, Z. Tokei, M.R. Baklanov, Solid State Phenom. 134 (2008) 317–320. [9] T.C. Chang, P.T. Liu, Y.S. Mor, T.M. Tsai, C.W. Chen, Y.J. Mei, F.M. Pan, W.F. Wu, S.M. Sze, J. Vac. Sci. Technol. B 20 (4) (2000) 1561–1566. [10] H. Okabe, Photochemistry of Small Molecules, Willey, New York, 1978. [11] S. Eslava, G. Eymery, P. Marsik, F. Iacopi, Ch.E.A. Kirschock, K. Maex, J.A. Martens, M. Baklanov, J. Electrochem. Soc. 155 (5) (2008) G115–G120. [12] A.M. Urbanowicz, B. Meshman, D. Schneider, M.R. Baklanov, Phys. Status Solidi A 205 (4) (2008) 829–832. [13] T. Tatsumi, S. Fakuda, Shingo Kadomura, Jpn. J. Appl. Phys. 33 (1994) 2175– 2178.