ALD Resist Formed by Vapor-Deposited Self-Assembled Monolayers

A new process of applying molecular resists to block HfO2 and Pt atomic layer deposition has been investigated. Monolayer films are formed from ...
0 downloads 0 Views 213KB Size
Download PDF
1160

Langmuir 2007, 23, 1160-1165

ALD Resist Formed by Vapor-Deposited Self-Assembled Monolayers Junsic Hong,† David W. Porter,† Raghavasimhan Sreenivasan,‡ Paul C. McIntyre,‡ and Stacey F. Bent*,† Department of Chemical Engineering and Department of Material Science and Engineering, Stanford UniVersity, Stanford, California 94305 ReceiVed March 8, 2006. In Final Form: August 31, 2006 A new process of applying molecular resists to block HfO2 and Pt atomic layer deposition has been investigated. Monolayer films are formed from octadecyltrichlorosilane (ODTS) or tridecafluoro-1,1,2,2-tetrahydrooctyltrichlorosilane (FOTS) and water vapor on native silicon oxide surfaces and from 1-octadecene on hydrogen-passivated silicon surfaces through a low-pressure chemical vapor deposition process. X-ray photoelectron spectroscopy data indicates that surfaces blocked by these monolayer resists can prevent atomic layer deposition of both HfO2 and Pt successfully. Time-dependent studies show that the ODTS monolayers continue to improve in blocking ability for as long as 48 h of formation time, and infrared spectroscopy measurements confirm an evolution of packing order over these time scales.

Introduction With the increasing attention given to atomic layer deposition (ALD)1-6 as a method to deposit a wide range of materials, there is growing interest in developing techniques for carrying out area-selective ALD. Area-selective ALD is an additive method for depositing thin films directly onto a patterned template and hence eliminates the need for etching of the deposited material.7-15 Area-selective ALD utilizes an initial patterning step in which the substrate surface is modified chemically, followed by atomic layer deposition. Although a relatively new technique, it has been demonstrated using a variety of methods, including microcontact printing8-11 and patterned SiO2/Si.13-15 The ALD process is based on self-limiting chemisorption reactions in which two or more vapor-phase precursors are introduced to the substrate in an alternating fashion, separated by an inert gas purge to prevent gas-phase reaction. ALD has been examined as a way to deposit gate stack materials, such as hafnium oxide (HfO2) and zirconium oxide (ZrO2), for nextgeneration integrated circuits as well as for the deposition of dielectrics, semiconductors, and metals for many other applications.4,5 ALD is known for its ability to deposit films with excellent † ‡

Department of Chemical Engineering. Department of Material Science and Engineering.

(1) Ritala, M.; Kukli, K.; Rahtu, A.; Raisanen, P.; Leskela, M.; Sajavaara, T.; Keinonen, J. Science 2000, 288, 319-321. (2) Ritala, M.; Lekela, M. In Handbook of Thin Film Materials; Nalwa, H. S., Ed.; Academic Press: San Diego, CA, 2001; Vol. 1, Chapter 2. (3) Ritala, M.; Leskela, M. Nanotechnology 1999, 10, 19-24. (4) Kang, L.; Jeon, Y.; Onishi, K.; Lee, B. H.; Qi, W.-J.; Nieh, R.; Gopalan, S.; Lee, J. C. IEEE Symp. VLSI Technol. 2000, 44-45. (5) Sankur, H. O.; Gunning, W. Appl. Opt. 1989, 28, 2806-2808. (6) Ott, A. W.; Chang, R. P. H. Mater. Chem. Phys. 1999, 58, 132-138. (7) Lee, J. P.; Kim, H. K.; Park, C. R.; Park, G.; Kwak, H. T.; Koo, S. M.; Sung, M. M. J. Phys. Chem. B 2003, 107, 8997-9002. (8) Yan, M.; Koide, Y.; Babcock, J. R.; Markworth, P. R.; Belot, J. A.; Marks, T. J.; Chang, R. P. H. Appl. Phys. Lett. 2001, 79, 1709-1711. (9) Park, M. H.; Jang, Y. J.; Sung, H. M.; Sung, M. M. Langmuir 2004, 20, 2257-2260. (10) Park, K. J.; Doub, J. M.; Gougousi, T.; Parsons, G. N. Appl. Phys. Lett. 2005, 86, 51903-1. (11) Seo, E. K.; Lee, J. W.; Sung-Suh, H. M.; Sung, M. M. Chem. Mater. 2004, 16, 1878-1883. (12) Sinha, A.; Hess, D. W.; Henderson, C. L. J. Electrochem. Soc. 2006, 153, G465-469. (13) Chen, R.; Kim, H.; McIntyre, P. C.; Bent, S. F. Appl. Phys. Lett. 2004, 84, 4017-4019. (14) Chen, R.; Kim, H.; McIntyre, P. C.; Bent, S. F. Chem. Mater. 2005, 17, 536-544.

conformality and precise control of film thickness.2 In areaselective ALD, the goal is to extend the process to achieve spatial control in the lateral dimension in addition to vertical (film thickness) control. Area-selective ALD can be accomplished by spatially controlling the substrate properties to inhibit the precursor deposition chemistry over well-defined regions of the substrate. Most previous studies of area-selective atomic layer deposition have utilized self-assembled monolayers (SAMs) as ALD resists.7-11,13-15 The SAM is typically prepared at the liquid/ solid interface by simply immersing a substrate in a dilute solution of the precursor or by spreading the liquid precursor on a polydimethylsiloxane (PDMS) stamp for microcontact printing. Octadecyltrichlorosilane (ODTS) has been the most commonly used precursor for SAM formation on SiO2 substrates, and ODTS SAMs have been shown to result in successful blockage of ALD in materials such as HfO2 and ZrO2,13 Ru,10 TiO2,9,11,12 and ZnO.8 It has been shown that the requirements for the quality of the SAMs can be stringent. For example, only tightly packed, longchain alkane SAMs were shown to block the ALD of HfO2 completely.14 To our knowledge, however, none of these studies prepared the SAMs by vapor-phase processes. Recent data using dichlorodimethylsilane (DDMS) and 1-decene as precursors have demonstrated that vapor-phase processes provide significant advantages over traditional liquid-phase processes in SAM formation.16-18 For example, vapor-phase processes consume less precursor compared to liquid-phase processes. Furthermore, aggregation of the precursor molecules prior to deposition on the SiO2 surface, which can lead to a deterioration of the quality of the SAM in liquid-phase processes, was significantly reduced when vapor-phase processes were used. Aggregated precursors had lower vapor pressures than the single-molecule precursors and thus were rarely vaporized. In the case of area-selective ALD in particular, vapor-phase SAM formation provides another distinct advantage. Namely, (15) Chen, R.; Kim, H.; McIntyre, P. C.; Porter, D. W.; Bent, S. F. Appl. Phys. Lett. 2005, 86, 1-3. (16) Ashurst, W. R.; Yau, C.; Carraro, C.; Lee, C.; Kluth, G. J.; Howe, R. T.; Maboudian, R. Sens. Actuators, A 2001, 91, 239-248. (17) Ashurst, W. R.; Yau, C.; Carraro, C.; Maboudian, R.; Dugger, M. T. J. Microelectromech. Syst. 2001, 10, 41-49. (18) Ashurst, W. R.; Carraro, C.; Maboudian, R. IEEE Trans. DeVice Mater. Reliab. 2003, 3, 173-178.

10.1021/la0606401 CCC: $37.00 © 2007 American Chemical Society Published on Web 12/19/2006

ALD Resist Formed by Vapor-Deposited SAMs

Figure 1. Illustration of the SAM-formation chamber. The precursor and water pressures (Pp and Pw, respectively) and the substrate temperature Ts could be independently varied.

because it can be performed in a vacuum system, it allows for easier integration with an ALD reactor. In principle, SAM formation can be carried out in the ALD reactor itself by introducing the ODTS (or other) precursors before initiating the ALD cycles. Alternatively, a cluster-type tool could be used. With a patterned SiO2/Si substrate, selective attachment of SAMs on only one of the two materials (oxide or silicon) followed by area-selective ALD would lead to spatially patterned deposition in a single tool without the use of solvents.15 In this study, we investigated whether vapor-phase delivery of different self-assembled monolayers, including octadecyltrichlorosilane (ODTS) and tridecafluoro-1,1,2,2-tetrahydrooctyltrichlorosilane (FOTS), could form SAMs of sufficiently high quality on SiO2 surfaces to prevent the subsequent ALD of HfO2 and/or Pt. We also investigated a different type of monolayer formed through a hydrosilylation reaction on hydrogen-terminated silicon surfaces using 1-octadecene.19-24 The quality of SAM formation and subsequent ALD reaction was examined by X-ray photoelectron spectroscopy (XPS), infrared spectroscopy, contact angle, and ellipsometry. Experimental Methods p-type (boron-doped) silicon (100) wafers with a resistivity of 0.01 Ω cm were purchased from Si-Tech, Inc. The wafers were cut into 1 cm2 pieces, sonicated in chloroform for 5 min in order to remove any organic contaminants, and blown dry with compressed air. Following surface cleaning, the samples were treated in one of two ways. The first group was placed in an oxygen plasma cleaner for 3 min to create hydroxyl group (OH)-terminated oxidized silicon surfaces. The second group was etched by hydrofluoric acid after piranha treatment to generate hydrogen (H)-terminated silicon surfaces. The etching was carried out immediately before the samples were placed in a home-built chamber for self-assembled monolayer (SAM) formation. Figure 1 schematically illustrates the SAM chamber used in this study. It is a stainless steel chamber evacuated by a mechanical pump and served by two feed lines for introducing the SAM precursor and water. The addition of water vapor was necessary to form stable SAMs, as reported previously.25 The samples are set in a substrate holder at the bottom of the chamber, which acts as a hot wall reactor. The temperature of the chamber was raised by surrounding heating tapes and controlled by a Eurotherm temperature controller that maintained a stable temperature with less than (2 °C fluctuations throughout the experiments. The chamber base pressure was 20 mTorr. (19) Linford, M. R.; Fenter, P.; Eisenberger, P. M.; Chidsey, C. E. D. J. Am. Chem. Soc. 1995, 117, 3145-3155. (20) Buriak, J. M. Chem. ReV. 2002, 102, 1271-1308. (21) Lopinski, G. P.; Wayner, D. D. M.; Wolkow, R. A. Nature 2000, 406, 48-51. (22) Cicero, R. L.; Chidsey, C. E. D.; Lopinski, G. P.; Wayner, D. D. M.; Wolkow, R. A. Langmuir 2002, 18, 305-307. (23) Hofer, W. A.; Fisher, A. J.; Lopinski, G. P.; Wolkow, R. A. Chem. Phys. Lett. 2002, 365, 129-134. (24) Kosuri, M. R.; Gerung, H.; Li, Q. M.; Han, S. M.; Bunker, B. C.; Mayer, T. M. Langmuir 2003, 19, 9315-9320. (25) Mayer, T.; de Boer, M.; Shinn, N.; Clews, P.; Michalske, T. J. Vac. Sci. Technol., B 2000, 18, 2433-2440.

Langmuir, Vol. 23, No. 3, 2007 1161 Precursors (ODTS, FOTS, and 1-octadecene) and water were stored separately in glass vials that were connected through valves to the SAM chamber. Prior to precursor and water dosing into the SAM chamber, three freeze-pump-thaw cycles were used to remove any residual oxygen and other unwanted gases from the liquid precursors and water. Well-controlled amounts of water and precursors were then sequentially vaporized into the SAM chamber. For the ODTS SAM, 150 mTorr of ODTS and 450 mTorr of water were dosed at 170 °C; for FOTS, 250 mTorr of FOTS and 750 mTorr of water were dosed at 60 °C. For the 1-octadecene SAM, 250 mTorr of 1-octadecene was introduced into the SAM chamber. After dosing the precursors and water, the valve to the SAM chamber was turned off and remained closed throughout the selfassembled monolayer formation, which lasted up to 48 h. Following SAM formation, the treated samples were typically washed three or four times with anhydrous toluene and blown dry with air to remove any residual precursors that may be nonspecifically absorbed on the silicon surface. Some samples were not rinsed to investigate the effect of rinsing, and the results obtained were the same for both sets of samples. The thickness of the precursor film was measured and averaged over several spots on each sample (n ) 3 or 4 per sample) by ellipsometry, assuming a refractive index of n ) 1.46. The hydrophobicity of the precursor film was measured by average water contact angle. Some of the ODTS SAMs were characterized using multiple internal reflection Fourier transform infrared (MIRFTIR) spectroscopy. For these experiments, SAMs were deposited on silicon substrates cut into a trapezoidal geometry to allow for multiple internal reflection. Scans (4000) were collected over the mid-infrared region for each sample and referenced to a background spectrum of the clean substrate. The methodology used for ALD of HfO2 and Pt has been previously described.1,2,26-28 Here we use that method with only a slight modification. High-purity tetrakis-(dimethylamido)hafnium [Hf[N(CH3)2]4] and deionized water were used as ALD precursors for hafnium oxide thin films, and methyl(cyclopentadienyl)trimethylplatinum [CH3C5H4Pt(CH3)3] and dry air were used as ALD precursors for Pt thin film deposition. A custom-built reactor capable of handling 4 in. wafers was used for all the ALD experiments, with the exception of the experiment using the octadecane resist. For this experiment only, HfO2 ALD was carried out in a separate ALD reactor using HfCl4 and H2O as precursors. The reactor details and ALD conditions for this system have been described previously.13 The HfO2 deposition occurs according to the following half reactions1,2 A: Hf-OH* + Hf[N(CH3)2]4 f Hf-O-Hf[N(CH3)2]/3 + NH(CH3)2v B: Hf-N(CH3)/2 + H2O f Hf-OH* + NH(CH3)2v where the asterisk represents the surface species. In both half reactions, a gas-phase molecular precursor reacts with the surface functional species and saturates the surface in a self-limiting manner. During ALD, the samples were alternatively exposed to Hf[N(CH3)2]4 and water vapor. The Hf[N(CH3)2]4 precursor was maintained at about 65 °C while the H2O was at room temperature. After each exposure, the reaction chamber and gas manifold were purged with nitrogen to avoid possible chemical vapor deposition-type reactions and physisorption of unreacted species onto the surface. The exposure time for Hf[N(CH3)2]4 and water vapor was 0.1 and 0.2 s, respectively, followed by 3 min of nitrogen purging after each precursor pulse. The substrate temperature was 250 °C. This HfO2 ALD process was repeated for 50 cycles, resulting in a thin film 36-38 Å in thickness on reference native-oxide-passivated Si(100) substrates. (26) Suntola, T., Simpson, M., Eds. Atomic Layer Epitaxy; Chapman & Hall: New York, 1990. (27) Aaltonen, T.; Ritala, M.; Sajavaara, T.; Keinonen, J.; Leskela, M. Chem. Mat. 2003, 15, 1924-1928. (28) Puurunen, R. L. Chem. Vap. Dep. 2005, 11, 79-90.

1162 Langmuir, Vol. 23, No. 3, 2007

Hong et al.

For Pt ALD, the exposure time for both precursors (methylcyclopentadienyltrimethylplatinum and air) was 2 s, with a 90 s nitrogen purge time. The Pt ALD process also consists of two self-limiting chemical half-reactions27,28 A: Pt(s) + O2(g) f Pt-O/x B: Pt-O/x + CH3C5H4Pt(CH3)3 f Pt(s) + CO2(g)v + H2O(g)v + byproductv where the asterisk represents the surface species. Pt deposition was carried out at a substrate temperature of 300 °C. This Pt ALD process was repeated for 50 cycles, leading to the deposition of a 15-20 Å thin film of Pt. Compressed nitrogen gas that was free of any particulates, oxygen, water, and hydrocarbons (to better than sub-ppm levels) flowed at a rate of 200 sccm through the lines that connected precursors and water bubblers to the ALD reactor. The temperature of the ALD reactor body was maintained at approximately 150 °C, and the pressure was maintained at 500 mTorr by nitrogen carrier gas flow when there was no dosing of the precursor or water. The ALD cycles were continued until the desired film thickness was achieved. The results of the ALD process were examined by X-ray photoelectron spectroscopy (XPS) at a detection limit of 0.5 atom %.

Results and Discussion ALD Resists from ODTS and FOTS SAMs. Vaporized organosilane molecules react with the hydroxyl groups on an SiO2-covered Si substrate, resulting in the formation of a SAM. In the first step of monolayer formation, it is believed that the Si-Cl bonds of the precursor molecule are hydrolyzed by reacting either with vapor-phase water or with a water layer on the surface, resulting in the formation of Si-OH groups.29,30 All Si-Cl bonds are hydrolyzed, as indicated by the absence of chlorine in the X-ray photoelectron spectra. A condensation reaction results in the formation of covalent siloxane bonds between adjacent head groups of the molecules, leading to very stable bonding at the surface. Following SAM formation from both ODTS and FOTS, the samples were characterized by contact angle and ellipsometry. The film thickness of the ODTS- and FOTS-treated substrates were found to increase with SAM-formation time until a plateau was reached at values of 24 Å for ODTS and 12 Å for FOTS. These thicknesses for the deposited films are in good agreement with previous reports in the literature.13,14,25,31 The contact angle was found to exhibit similar behavior, reaching a maximum value of 110° for ODTS and 112° for FOTS. For ODTS, reactions times longer than 12 h were required before the maximum film thickness and contact angle were reached, whereas the substrate treated with FOTS took less time (6 h) to plateau. The difference in reaction time to form a well-packed monolayer can be partially ascribed to the different dosing conditions used for the two precursors. ODTS has a vapor pressure of 150 mTorr at 170 °C, whereas the vapor pressure of FOTS is around 960 mTorr at 60 °C. Higher precursor temperatures could not be used for ODTS because of the onset of thermal decomposition. Consequently, at the conditions used to form the densely packed self-assembled monolayers, FOTS has a significantly higher vapor pressure than ODTS during exposure. In addition to the different exposure conditions, FOTS and ODTS also vary in their inherent reactivity with silicon oxide, which will also impact the monolayer formation time. (29) Ulman, A. Chem. ReV. 1996, 96, 1533-1554. (30) Schreiber, F. Prog. Surf. Sci. 2000, 65, 151-256. (31) Hozumi, A.; Ushiyama, K.; Sugimura, H.; Takai, O. Langmuir 1999, 15, 7600-7604.

Figure 2. XPS spectra following HfO2 ALD on different substrates. (a) A native silicon oxide surface, (b) a native silicon oxide surface coated with ODTS, and (c) a native silicon oxide surface coated with FOTS.

Once the SAMs were formed, the samples were removed from the vapor SAM delivery chamber and introduced into the ALD reactor. Both HfO2 and Pt atomic layer deposition were used to test the characteristics of the SAM blocking layers and their suitability as an ALD resist. Following ALD, the samples were characterized with XPS. By measuring the amount of deposition by XPS, we found that even though the SAMs appeared to be well packed according to saturation values of the contact angle and film thickness not all of the SAMs could block ALD completely. We observed that long reaction times were crucial to making high-quality self-assembled monolayers that were effective as an atomic layer deposition resist. The time-dependent behavior will be described more fully below, and here we focus only on the blocking achievable by the fully packed, most resistant SAMs.

ALD Resist Formed by Vapor-Deposited SAMs

Figure 2 shows the results of ex-situ XPS experiments following HfO2 ALD on a thin (18 Å) native oxide (Figure 2a), SiO2 coated with ODTS (Figure 2b), and SiO2 coated with FOTS (Figure 2c). In Figure 2a, distinct hafnium peaks (i.e., Hf(4p) ) 437, 380 eV; Hf(4d) ) 222, 211 eV; and Hf(4f) ) 16, 14 eV) can clearly be discerned, indicating the successful deposition of HfO2 on the SiO2 sample. The thickness of the HfO2 film on a native silicon oxide-coated substrate is 36-38 Å (measured independently) after 50 cycles. The scan shows intense peaks due to the main film components (Hf and O) and the underlying substrate materials (Si and O). The carbon C 1s peak arises from hydrocarbon contamination, likely collected during the ex-situ analysis. The composition of Hf measured in the scan is 12%. This number is less than the 33.3% expected for a stoichiometric bulk HfO2 film, which is likely due to the contribution to the XPS spectrum from the substrate silicon and carbon and oxygen contamination. The ODTS sample in Figure 2b was formed by 2 days of reaction in the SAM-formation chamber. On the basis of the absence of Hf peaks in the XPS spectrum, we can conclude that no HfO2 was deposited on this well-formed SAM. Higherresolution XPS scans (Supporting Information) also show no presence of Hf. Furthermore, the intensity and shape of the XPS peaks after ALD were found to be unchanged from that of fully developed ODTS (not shown) prior to ALD. This result demonstrates that ODTS delivered by vapor-phase deposition is chemically stable under HfO2 ALD conditions and can be used as an effective molecular resist against HfO2 atomic layer deposition. We note that although satisfactory blocking was demonstrated for up to 50 cycles, this resist is unlikely to retain its blocking ability for large numbers of cycles. At 200 cycles on an ODTS resist, for example, Hf was evident in the XPS spectrum. Figure 2c shows the XPS spectrum collected after HfO2 ALD on an FOTS monolayer, which is fully developed after 6 h of reaction. It is clear that there are no hafnium peaks on the FOTScoated substrate also to within the detection limits of the spectrometer (