Mater. Res. Soc. Symp. Proc. Vol. 1155 © 2009 Materials Research Society

1155-C11-06

FTIR study of copper agglomeration during atomic layer deposition of copper Min Dai,1 Jinhee Kwon,2 Yves J. Chabal,2 Mathew D. Halls,3 and Roy G. Gordon4 Laboratory for Surface Modification, Rutgers University, 136 Frelinghuysen Road, Piscataway, NJ 08854, U. S. A. 2 Department of Materials Science and Engineering, The University of Texas at Dallas, Richardson, TX 75080, U.S.A. 3 Materials Science Division, Accelrys Inc., San Diego, CA 92121, U.S.A. 4 Department of Chemistry and Chemical Biology, Harvard University Cambridge, MA 02138, U.S.A.
1

ABSTRACT The growth of of metallic copper by atomic layer deposition (ALD) using copper(I) di-secbutylacetamidinate ([Cu(sBu-amd)]2) and molecular hydrogen (H2) on SiO2/Si surfaces has been studied. The mechanisms for the initial surface reaction and chemical bonding evolutions with each ALD cycle are inferred from in situ Fourier transform infrared spectroscopy (FTIR) data. Spectroscopic evidence for Cu agglomeration on SiO2 is presented involving the intensity variations of the SiO2 LO/TO phonon modes after chemical reaction with the Cu precursor and after the H2 precursor cycle. These intensity variations are observed over the first 20 ALD cycles at 185oC. INTRODUCTION Atomic layer deposition (ALD) has recently received great interest with the need for new materials and thin-film deposition techniques for device scaling in integrated circuits (IC). The self-limiting growth mechanism of ALD is ideal for producing very thin, conformal films even on surfaces with high aspect ratios with control of the thickness and composition at the atomic level. As the industry heads toward the 22 nm node, the interest in metal ALD is growing, especially for copper interconnects. Cu is replacing aluminum as an interconnect material in ICs due to its lower resistivity (1.72×10−6 Âcm vs. 2.82×10−6 Âcm) and higher melting point (1083 o C vs. 659oC).1 For this application, a highly conformal and continuous copper seed layer is required before subsequent electrochemical deposition of copper film with high growth rate.2 ALD of non-noble metals, however, has had limited success 3, 4. One of the reasons for these difficulties is related to a lack of suitable precursors which satisfy stringent requirements for ALD. 5 In addition, the lack of understanding of growth mechanisms of metal ALD hinders the development of suitable precursors. ALD of Cu has been considered previously. Several copper (I) and copper (II) precursors were commonly used in the past, such as CuCl4, 6, Cu(II)-2,2,6,6 -tetramethyl-3,5-heptandionate [Cu(thd)2]7, 8, Cu(II)-1,1,1,5,5,5-hexafluoro-2,4 -pentanedionate [Cu(hfac)2] 9, 10, and Cu(II) acetylacetonate [Cu(acac)2] 11, 12, but all suffered from undesirable properties as ALD

precursors.13 For example, CuCl has very low vapor pressure, and the Cu film deposited by Cu(hfac)2 contains fluorine which reduces the Cu adhesion on the substrate. They all have very low reactivity with low growth rate, so that either high temperature (> 200oC), which is not well suited for smooth Cu growth due to Cu agglomeration and diffusion 13-15, or plasma is needed to enhance the reactivity. Recently, Gordon et al.16 reported ALD growth of metallic copper with high conformality and high conductivity using an amidinate precursor (copper (I) di-sec-butylacetamidinate, [Cu(sBu-amd)]2) and molecular hydrogen at relatively low temperature (~185oC). One of the advantages of this precursor is its high reactivity with molecular hydrogen at reasonable temperatures. In order to understand the ALD Cu growth mechanisms with this amidinate precursor, we used in-situ Fourier transform infrared spectroscopy (FTIR) to investigate the chemical bonding evolution after each ALD cycle. EXPERIMENT Double side polished, float-zone grown, and lightly doped (~10 Âcm) Si(100) with thin thermal oxide (6-10 nm thick SiO2) is used. The sample is first rinsed by acetone, methanol, and deionized water (DI water, 18.2 M Âcm), then standard RCA 17, 18 cleaning is performed followed by thorough rinsing with deionized water and blow-drying with N2. Then the sample is immediately loaded in the ultra pure N2 (oxygen impurity < 10-3 ppm) purged ALD chamber. All experiments are done in a home-made ALD system integrated with a Thermo Nicolet 670 interferometer and a MCT/B detector for in-situ FTIR measurements.19 A single-pass transmission geometry is used with an incidence angle close to the Brewster angle to maximize transmission, minimize interference, and increase sensitivity to absorptions below ~1500 cm-1. Distinction of perpendicular from parallel modes to the surface is made by additional normal incidence measurements. Mid-infrared range of 400 - 4000 cm-1 is scanned with the resolution of 4 cm-1. [Cu(sBu-amd)]2 is kept at 95-100oC, and purified N2 is used as the carrier gas to deliver the copper precursor. Cu film is deposited by introducing [Cu(sBu-amd)]2 (exposure ~ 107 L) and H2 (exposure ~ 1011 L) (ultra high purity, 99.999%, purified by Aeronex Gate Keeper gas purifiers) alternatively into the ALD chamber. During deposition, the substrate temperature is kept at 185oC for optimum Cu deposition. After each precursor dosing, the ALD chamber is N2-purged thoroughly with pumping. RESULTS AND DISCUSSION Figure 1 shows differential IR absorbance spectra of the first cycle of Cu ALD on SiO2 at 185oC, where the maximum Cu growth rate is observed.13 Upon the first half cycle when the Cu precursor [Cu(sBu-amd)]2 (black) is introduced, the intensity of the transverse and longitudinal optical modes (TO/LO) of SiO2 at 1075/1247 cm-1 decreases. Their intensity loss indicates that the SiO2 matrix is chemically disrupted due to the reaction of the Cu precursor with surface OH groups. The absorption band at 1010 cm-1 corresponds to the formation of Si-O-Cu bonds due to the adsorbed Cu precursor on SiO2. The reaction scheme is shown as the inset in the right. Upon the H2 exposure (red), there is a partial recovery of SiO2 phonon modes (1075/1247 cm-1). The recovery of these modes suggests that after the H2 dosing, the Cu precursor on the surface is reduced to pure copper atoms which then migrate and self-agglomerate to form copper particles

13, 21, 22

. As they agglomerate, the chemical bonds established during the first half cycle between Cu and the underlying SiO2 matrix are broken and rehydroxylated, partially restoring the SiO2 phonon modes and hydroxyl sites on the surface.

2x10

-4

2x10
^ K^ dK >K

-4



1x10

-4

 K^

, 
, EE K,

750

1000

1250

1500

1750 2800

3000

3700 3800

t

Figure 1. Differential IR absorbance spectra of the first cycle of Cu ALD on SiO2 measured after the Cu precursor (black) and H2 (red) dosing. Each spectrum is referenced to the spectrum of the previous treatment and the bottom is referenced to initial oxide surface. The incident angle of IR beam is 74o. The inset in the right shows a schematic of surface reactions corresponding to observations in each spectrum. In addition, the removal of the ligands by H2 reduction opens more reaction sites which were initially blocked by the intact ligands. In addition to the repetitive change of SiO2 phonon modes, the surface Si-O-Cu bonds at 1010 cm-1 follows the similar pattern of gain and recovery after Cu precursor and H2, respectively. This process is again schematically illustrated in the inset of Figure 1. The repetitive loss and gain of SiO2 modes are observed for the 20th cycle, implying that the surface is not saturated and Cu atoms still agglomerate after 20 ALD cycles. This observation is consistent with the high percolation thickness for Cu on glass reported previously.13 In Figure 1, the mode at 1510 cm-1 is assigned to (N-C-N) of the intact ligand attached to Cu atoms on the surface. Along with CHx stretching modes at 2800-3000 cm-1 (not shown), their intensity increases after each copper precursor and decreases after each H2, confirming the ALD process through ligand exchange. Cu agglomeration (i.e. formation of Cu structures with substantially thicker diameter than the average Cu thickness) is also observed in the shape of Rutherford backscattering spectroscopy (RBS) of Cu in Figure 2. The areal density of Cu atoms is measured by ex situ RBS with 2 MeV He+ ions. The detector is placed at 160o backscattering angle.

200 150 Counts 100 50 0 240

Cu

260  Channels

280

Figure 2. RBS spectrum (red dots) after 10 ALD cycles and simulations assuming uniform Cu coverage (gray) and non-uniform Cu coverage (blue). The detector is placed at a grazing angle of 160o backscattering angle. The FWHM of the simulation assuming uniform coverage of Cu atoms on the surface (gray) is narrower than that of the experimental RBS data (red dots). The blue line shows the simulated result after taking into account agglomerated Cu atoms of an island structure on the surface. The simulation of non-uniform Cu coverage is reasonably well fitted to the RBS data, confirming the agglomerated Cu on SiO2. CONCLUSIONS The surface reaction and the growth mechanism of Cu ALD with a liquid amidinate Cu precursor ([Cu(sBu-amd)]2) and molecular hydrogen on oxidized Si surfaces (6-10 nm thick SiO2) have been investigated by using in situ FTIR spectroscopy and ex situ RBS. The agglomeration of Cu atoms on SiO2 was shown to have a significant effect on the evolution of SiO2 phonon modes. The hydrogen reduction of the surface ligand is observed through the modes attributed to CHx and -NCN- bonds, confirming ALD process through ligand exchange. The Cu agglomeration on the oxide is also confirmed by a broader RBS spectrum of Cu. ACKNOWLEDGMENTS This work was supported by the National Science Foundation (CHE-0415652). REFERENCES 1. The International Technology Roadmap for Semiconductors, Semiconductor Industry Association. http://pulic.itrs.net. 2005. 2. Jae Jeong, K.; Soo-Kil, K.; Chang Hwa, L.; Yong Shik, K., Investigation of various copper seed layers for copper electrodeposition applicable to ultralarge-scale integration interconnection. Journal of Vacuum Science & Technology B 2003, 21, (1), 33-38.

3. Lim, B. S.; Rahtu, A.; Gordon, R. G., Atomic layer deposition of transition metals. Nat Mater 2003, 2, (11), 749-754. 4. Marika, J.; Mikko, R.; Markku, L., Deposition of copper films by an alternate supply of CuCl and Zn. Journal of Vacuum Science & Technology A 1997, 15, (4), 2330-2333. 5. Musgrave, C.; Gordon, R. G., Precursors for Atomic Layer Deposition of High-K Dielectrics. Future Fab International 2005, 18, 126-128. 6. Martensson, P.; Carlsson, J.-O., Atomic Layer Epitaxy of Copper on Tantalum. Chemical Vapor Deposition 1997, 3, (1), 45-50. 7. Per, M.; Jan-Otto, C., Atomic Layer Epitaxy of Copper. Journal of the Electrochemical Society 1998, 145, (8), 2926-2931. 8. Christopher, J.; Lanford, W. A.; Christopher, J. W.; Singh, J. P.; Pei, I. W.; Jay, J. S.; Toh-Ming, L., Inductively Coupled Hydrogen Plasma-Assisted Cu ALD on Metallic and Dielectric Surfaces. Journal of the Electrochemical Society 2005, 152, (2), C60-C64. 9. Raj, S.; Balu, P., Atomic Layer Deposition of Copper Seed Layers. Electrochemical and Solid-State Letters 2000, 3, (10), 479-480. 10. Huo, J.; Solanki, R.; McAndrew, J., Characteristics of copper films produced via atomic layer deposition. Journal of Materials Research 2002, 17, 2394-2398. 11. Utriainen, M.; Kröger-Laukkanen, M.; Johansson, L.-S.; Niinist, L., Studies of metallic thin film growth in an atomic layer epitaxy reactor using M(acac)2 (M=Ni, Cu, Pt) precursors. Applied Surface Science 2000, 157, (3), 151-158. 12. Antti, N.; Antti, R.; Timo, S.; Kai, A.; Mikko, R.; Markku, L., Radical-Enhanced Atomic Layer Deposition of Metallic Copper Thin Films. Journal of the Electrochemical Society 2005, 152, (1), G25-G28. 13. Li, Z.; Rahtu, A.; Gordon, R. G., Atomic Layer Deposition of Ultrathin Copper Metal Films from a Liquid Copper(I) Amidinate Precursor. Journal of Electrochemical Society 2006, 153, C787-C794. 14. Benouattas, N.; Mosser, A.; Raiser, D.; Faerber, J.; Bouabellou, A., Behaviour of copper atoms in annealed Cu/SiOx/Si systems. Applied Surface Science 2000, 153, (2-3), 79-84. 15. McBrayer, J. D.; Swanson, R. M.; Sigmon, T. W., Diffusion of Metals in Silicon Dioxide. Journal of the Electrochemical Society 1986, 133, (6), 1242-1246. 16. Li, Z.; Barry, S. T.; Gordon, R. G., Synthesis and Characterization of Cu(I) Amidinates as Precursors for Atomic Layer Deposition (ALD) of Copper Metal. Inorganic Chemistry 2005, 44, 1728-1735. 17. Higashi, G. S.; Chabal, Y. J., Silicon surface chemical composition and morphology, Chapter in Handbook of Silicon Wafer Cleaning Technology: Science, Technology, and Applications Werner Kern ed, Noyes Pub., 1993. 18. Weldon, M. K.; Marsico, V. E.; Chabal, Y. J.; Hamann, D. R.; Christman, S. B.; E. E. Chaban, S. S., Infrared Spectroscopy as a Probe of Fundamental Processes in Microelectronics: Silicon Wafer Cleaning and Bonding, . Surface Science 1996, 368, 163. 19. Kwon, J.; Dai, M.; Langereis, E.; Halls, M. D.; Chabal, Y. J.; Gordon, R. G., In-situ Infrared Characterization during Atomic Layer Deposition of Lanthanum Oxide. Journal of Physical Chemistry C 2009, 113, (2), 654-660. 20. Conley, R. T., Infrared Spectroscopy. Allyn and Bacon, Inc. 1972. 21. Yang, C.-Y.; Jeng, J. S.; Chen, J. S., Grain growth, agglomeration and interfacial reaction of copper interconnects. Thin Solid Films 2002, 420-421, 398-402.

22. Ching-Yu, Y.; Chen, J. S., Investigation of Copper Agglomeration at Elevated Temperatures. Journal of the Electrochemical Society 2003, 150, (12), G826-G830.