pubs.acs.org/Langmuir © 2010 American Chemical Society
Molecular-Scale Change in the Surface Properties of Silica and Copper in Pure Water with Time Cathy E. McNamee,*,† Masaki Nakayama,‡ and Ko Higashitani*,§ †
Shinshu University, Tokida 3-15-1, Ueda-shi, Nagano-ken 386-8567, Japan, ‡Mitsubishi Rayon Co., 10-1, Daikoku-cho, Tsurumi-ku, Yokohama 230-0053, Japan, and §Graduate School of Engineering, Kyoto University - Katsura, Nishikyo-ku, Kyoto, Kyoto 615-8530, Japan Received May 24, 2010. Revised Manuscript Received July 28, 2010
We aimed to understand one of the fundamentals of how silica and copper surfaces are flattened on the molecular scale by the chemical mechanical polishing (CMP) method, which is used in the fabrication of semiconductors. In particular, we examined whether even pure water affects the properties of silica and copper surfaces. This study used the atomic force microscope to detect how the morphologies, normal interaction, and friction forces of the silica and copper surfaces changed with their exposure time to water. We found that the surface properties of even the silica surfaces changed on the molecular scale when the surfaces were exposed to water for a sufficiently long time. In the case of copper, the surface properties were observed to undergo rapid changes. Gel-like layers were detected on the copper surface within a few minutes, even though copper surfaces oxidized by ambient air are considered to be rather stable to water.
1. Introduction One of the key technologies in the fabrication processes of semiconductors with high performance is the flattening of substrate surfaces on the molecular scale by chemical mechanical polishing (CMP).1 In the damascene process, the SiO2 substrate surface and the surface of the Cu interconnects are polished simultaneously by the CMP process, where the substrates are flattened by a surface abrasion aided by slurry particles. Microscopic SiO2 particles are usually used as the abrasive particles. Hence, the abrasion of the substrates proceeds through the contact, friction, and wear between the silica particles and silica substrate and also between the silica particles and copper substrate in slurry fluids containing various kinds of chemicals. However, it is still not clear why an angstrom order of surface flatness is achieved by such a simple abrasive process, i.e., CMP, in which both chemical and mechanical abrasions coexist. The operational conditions of CMP have been examined intensively, especially those used to flatten the soft surface of copper. The removal rate of copper can be controlled by adding oxidants and inhibitors to the slurry mixture.2 Oxidants increase the removal rate, where some oxidant examples include ammonia and hydroxylamine in basic solutions and various amino acids plus citric acid in acidic solutions.3,4 These materials act to complex the copper surface ions and change the interaction of the surface with water and the CMP removal rate.5 The inhibitor reduces the removal rate through the formation of a film on the *To whom correspondence should be addressed. E-mail: mcnamee@ shinshu-u.ac.jp (C.E.M.);
[email protected] (K.H.). (1) Kahng, A. B. IEEE Trans. Comput-Aided Des. Integr. Circuits Syst. 2008, 27 (1), 3–19. (2) Deshpande, S.; Kuiry, S. C.; Klimov, M.; Obeng, Y.; Seal, S. J. Electrochem. Soc. 2004, 151(11), G788–G794. (3) Paul, E.; Kaufman, F.; Brusic, V.; Zhang, J.; Sun, F.; Vacassy, R. J. Electrochem. Soc. 2005, 152(4), G322–G3238. (4) Du, T.; Tamboli, D.; Desai, V.; Seal, S. J. Electrochem. Soc. 2004, 151(4), G203–G235. (5) Maboudian, R.; Carraro, C. Annu. Rev. Phys. Chem. 2004, 55, 35–54. (6) El-Shafei, A. A.; Moussa, M. N. H.; El-Far, A. A. J. Appl. Electrochem. 1997, 27, 1075–1078.
15310 DOI: 10.1021/la1020934
copper surface, where glycine6,7 and benzotriazole8 have been reported as successful inhibitors. The material used here as the solute is mostly water containing salts, acids, or other organic materials. Therefore, it is also very important to identify how pure water itself affects the surfaces of copper and silica and to understand the corresponding mechanism on the molecular scale, which is a fundamental problem. We have studied the friction between silica surfaces in various electrolyte solutions by using an atomic force microscope (AFM).9-11 Various materials, such as silica and mica, have been used as the substrates in not only our measurements but also those from others who used the AFM, and their surface properties were mostly assumed to be stable, that is, independent of the experimental time. In the present study, we wanted to investigate whether the surface properties of silica and copper surfaces in pure water were influenced by their exposure to water during the experimental time, say 60 min. Hence, the molecular-scale changes in the surface morphology, the interaction surface forces, and the friction forces of silica and copper surfaces with time were investigated by using an AFM. The results of this study will contribute to the understanding of one of the fundamental features of CMP.
2. Experimental Section 2.1. Materials. The pure water used in this study was distilled
and deionized, giving a conductance of 18.2 MΩ cm-1 and a total organic content less than 5 ppm. The pH value of water was 5.8. The diameter of the silica particles used was 6.84 μm (Bangs Laboratory). Silica wafers (Silicon Quest Int.) and copper plates (plates made by Toplas Engineering, Japan, using the physical (7) Sz€ocs, E.; Vastag, G.; Shaban, A.; Konczos, G.; Kalman, E. J. Appl. Electrochem. 1999, 29, 1339–1345. (8) Yu, P.; Liao, D. M.; Luo, Y. B.; Chen, Z. G. Corrosion 2003, 59(4), 314–318. (9) Donose, B. C.; Vakarelski, I. U.; Higashitani, K. Langmuir 2005, 21, 1834– 1839. (10) Donose, B. C.; Vakarelski, I. U.; Taran, E.; Shinto, H.; Higashitani, K. Ind. Eng. Chem. Res. 2006, 45(21), 7035–7041. (11) Taran, E.; Kanda, Y.; Vakarelski, I. U.; Higashitani, K. J. Colloid Interface Sci. 2007, 307, 425–432.
Published on Web 09/03/2010
Langmuir 2010, 26(19), 15310–15315
McNamee et al.
Article
vapor deposition method, giving a Cu layer thickness of about 100 nm on a glass plate) were used as the substrates. The copper substrates were kept in a drybox without special care, so that the surface could be contaminated or oxidized by ambient air. 2.2. Methods. 2.2.1. Atomic Force Microscopy. An atomic force microscope (Digital Instruments NanoScope III Multimode) equipped with a fluid cell was used to image the surfaces of the copper and silica substrates in air and in water and also to measure the interaction and friction forces between a silica particle and the surfaces in water. The imaging was performed using regular V-shaped gold-plated Si3N4 cantilevers (NP-S, Veeco Nano Probe Tips, the normal spring constant k = 0.12 N m-1, the tip radius of curvature = 10-40 nm). Colloid probes were used for the normal force measurements. They were prepared by attaching a single dried silica particle to a Si3N4 cantilever (k = 0.12 N m-1 or 0.58 N m-1) by using an XYZ micromanipulator and an epoxy resin (Showa polymer, Japan). The friction force measurements were carried out using colloid probes attached to rectangular cantilevers (k=0.08 N m-1, CSC12/tipless/no Al, MikroMasch). 2.2.2. Experimental Procedure and Data Analysis. AFM images of the substrates were taken in air and in water using the normal contact-mode procedure. The scan size and rate were 1 1 um2 and 1.97 Hz, respectively, and both the resolution of data points per line and the number of lines were 512. All the images were modified by performing a first-order flattening along the scan lines without filtering. The experiments involving the AFM imaging in water were performed by injecting water into a contact-mode fluid cell with a syringe approximately every 5 min in order to replace the liquid with new water. This prevented the properties of the water being changed by the presence of any material that may have dissolved from either the substrate or probe. The damage of the cantilever tip during the experiment was checked using the following two methods. First, the consecutive imaging of the surfaces was performed by repeating two series of experiments on different surfaces using the same cantilever, and the results were compared. Second, the force curves between the tip and the surface before and after the force/imaging experiments were compared. No clear difference was found in both kinds of measurements. Hence, the tip was regarded as experienced a negligible wearing/modification during our experiments. The surface forces were measured using the force mode of the AFM between a substrate and a colloid probe using a contactmode fluid cell filled with water. In this case also, the water was flushed out of the cell and replaced with new water every 5 min. The force measurements were performed by bringing the colloid probe into contact with the substrate, during which time the change in the deflection of the cantilever was measured with a split photodiode as a function of the piezo displacement. The detector signal versus piezo position curves were then converted to force versus distance curves. This was done by subtracting a linear fit of the zero force line and determining the conversion factor CF between the detector signal in volts and the cantilever deflection from the slope of the linear compliance region. The force curves were then acquired by subtracting the cantilever deflection from the piezo position Δx and by using Hooke’s law F ¼ kΔx
ð1Þ
where F and k are the normal force and the nominal spring constant of the cantilever, respectively. In this study, we were interested in studying the time dependence of the interaction forces between a probe and a substrate surface in water. We therefore used the same colloid probe and substrate for each series. The forces were consequently plotted as a function of force (12) Ducker, W. A.; Senden, T. J.; Pashley, R. M. Langmuir 1992, 8, 1831–1836. (13) McNamee, C. E.; Matsumoto, M.; Hartley, P. G.; Mulvaney, P.; Tsujii, Y.; Nakahara, M. Langmuir 2001, 17, 6220–6227.
Langmuir 2010, 26(19), 15310–15315
versus separation distance between surfaces h and were not plotted as a function of force/probe radius versus h, as is often done.12,13 The friction force was measured using the AFM friction mode and a contact-mode fluid cell, where water flushing of the cell was done in the same way as described above. The friction was measured by applying a normal load to the substrate, while the substrate was slid horizontally underneath the cantilever at a given speed of 3.60 μm/s. The loading force L was calculated using L ¼ LV CF
ð2Þ
where LV is the applied load in volts. The lateral frictional force FL was calculated using FL ¼
V L SL K L 2H
ð3Þ
where VL, SL, KL, and H are the difference in the lateral force detector in one complete scan, the lateral detector sensitivity, the lateral spring constant, and the distance from the bottom of the sphere to the midpoint of the cantilever, respectively. SL was determined to be 3.1 10-4 rad/V from the method of Meurk and others,14 and the value of KL was determined from KL ¼
2kl 2 3ð1 þ υÞ
ð4Þ
where l is the length of the cantilever. The ν is the Poisson’s ratio given by the torsional ratio of the cantilever, where a typical value of 0.27 was used. The values of FL for each system and time were averaged from the values measured at different positions on the substrate in a 1 min time interval, in order to account for the possible differences on the substrates.
3. Results and Discussion Silica surfaces are often used as the substrates in AFM measurements in solutions, assuming that the surface properties are stable with time, at least during the experimental time. This is especially the case when pure water is used. A copper surface has also been considered to be stable to water,15-17 after the copper surface is covered by an oxidized layer of Cu2O or CuO after its exposure to air.18,19 However, these speculations were drawn from macroscopic measurements that lasted a few hundred hours. Here, we examined how SiO2 and Cu surfaces are affected on the molecular scale by their exposure time to water, texp, within the usual experimental time, 60 min, by investigating the changes in their surface morphology, normal interaction, and friction forces with time using the AFM. 3.1. Changes in the Properties of a Silica Surface in Water with Time. Here we examined the stability of a silica surface in water. The change was first characterized by imaging the silica wafer at the same position at different time intervals. The height images for the exposure time texp = 6, 19, 28, and 60 min are shown in parts a, b, c, and d of Figure 1, respectively. The height profiles corresponding to the lines in Figure 1 and their root-mean-square (rms) values are shown in Figure 2a and by the (14) Meurk, A.; Larson, I.; Bergstrom, L. Mater. Res. Soc. Symp. Proc. 1998, 552, 427. (15) Mattsson, E.; Fredriksson, A. M. Br. Corros. J. 1968, 3(5), 246–257. (16) Wranglen, G. An Introduction to Corrosion and Protection of Metals; Chapman & Hall: London, 1985. (17) Beverskog, B.; Puigdomenech, I. J. Electrochem. Soc. 1997, 144, 3476–3483. (18) Honkanen, M.; Vippola, M.; Lepisto, T. J. Mater. Res. 2008, 23(5), 1350– 1357. (19) Soon, A.; Todorova, M.; Delly, B.; Stampfl, C. Surf. Sci. 2007, 601, 5809– 5813.
DOI: 10.1021/la1020934
15311
Article
Figure 1. 1 1 μm2 height contact mode images of a silica plate in water and their time dependence.
Figure 2. Roughness of the surfaces: (a) height profiles along the lines shown in Figure 1; (b) the rms values (Data A is for the silica surface and corresponds to Figure 2a, and Data B is for the copper surface and corresponds to Figure 7).
“Data A” results in Figure 2b, respectively. The surfaces look very smooth with a height less than 0.2 nm for texp e 28 min. This implies that the silica surface is stable within this exposure time to water. However, the roughness of the surface apparently changed at texp = 60 min, as shown in Figure 2a, although the height difference was still within 0.4 nm. This was confirmed by the slight increase of the rms value at texp = 60 min, as shown by the Data A results in Figure 2b. This indicates that some morphological change occurred even in the case of the silica surface, when it was exposed to water for a sufficiently long time. Next, we examined the effect of water on a silica surface by measuring the interaction forces between the surfaces of a silica probe and a silica substrate in water. The approach and retract 15312 DOI: 10.1021/la1020934
McNamee et al.
Figure 3. Force curves measured between two silica surfaces in water as a function of contact time with water: (a) approach force curves; (b) retract force curves.
force curves at texp = 3, 12, 30, and 56 min are shown in Figure 3a,b. Long-range repulsive forces commencing at a separation distance of h ≈ 200 nm were observed in both the approach and retract force curves for all the time intervals studied. This repulsion is explained in terms of an electrostatic repulsion because both silica surfaces are negatively charged in water of pH 5.8.20 The magnitude of the repulsive forces slightly decreased with time, indicating a small change in the surface properties of the silica surfaces. It is especially interesting to note that in the case of the relatively long exposure time of texp = 56 min a steplike change appeared in the approach force curve at h
