studied by low energy electron diffraction and Auger ...

Oxygen adsorption on Cu–9 at.%Al(111) studied by low energy electron diffraction and Auger electron spectroscopy Michiko Yoshitake, Santanu Bera, Yasuhiro Yamauchi, and Weijie Song Citation: Journal of Vacuum Science & Technology A 21, 1290 (2003); doi: 10.1116/1.1560719 View online: http://dx.doi.org/10.1116/1.1560719 View Table of Contents: http://scitation.aip.org/content/avs/journal/jvsta/21/4?ver=pdfcov Published by the AVS: Science & Technology of Materials, Interfaces, and Processing Articles you may be interested in Position of segregated Al atoms and the work function: Experimental low energy electron diffraction intensity analysis and first-principles calculation of the ( √ 3 × √ 3 ) R 30 ° superlattice phase on the (111) surface of a Cu – 9 at. % Al alloy J. Vac. Sci. Technol. A 28, 152 (2010); 10.1116/1.3273533 Adsorption of oxygen on ultrathin Cu/Pt(111) films J. Vac. Sci. Technol. A 19, 2217 (2001); 10.1116/1.1379801 Auger electron spectroscopy low energy electron diffraction study of the growth mode of Ag on Au(111), (311), and (554) single-crystal surfaces J. Vac. Sci. Technol. A 17, 1647 (1999); 10.1116/1.581866 An x-ray spectromicroscopic study of electromigration in patterned Al(Cu) lines Appl. Phys. Lett. 74, 22 (1999); 10.1063/1.123120 Analysis of degradation in AlCu-metallization by low frequency noise-spectroscopy AIP Conf. Proc. 418, 68 (1998); 10.1063/1.54675

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 155.33.16.124 On: Sun, 23 Nov 2014 06:03:13

Oxygen adsorption on Cu–9 at. %Al„111… studied by low energy electron diffraction and Auger electron spectroscopy Michiko Yoshitake,a) Santanu Bera, Yasuhiro Yamauchi, and Weijie Song Nanomaterials Laboratory, National Institute for Materials Science, 3-13 Sakura, Tsukuba 305-0003, Japan

共Received 4 October 2002; accepted 18 November 2002; published 1 July 2003兲 Cu-based alloys have been used for electric cables for long time. In the field of microelectronics, Al had been used for electrical wiring. However, it became clear that electromigration occurs in Al that causes breaking of wires in minute wirings. Due to this problem, Cu wiring is used in most advanced microprocessors. Cu metal is more corrosive than Al and Cu-based alloys with a small amount of Al is expected to solve problems both on electromigration and corrosion. The initial stage of corrosion is oxygen adsorption. We studied surface segregation of Al on Cu–9% Al共111兲 and oxygen adsorption on the surface with/without Al segregation in ultrahigh vacuum by low energy electron diffraction 共LEED兲 and Auger electron spectroscopy. It was found that Al segregates on the surface to form 共)⫻)兲R30° structure and the structure vanishes above 595 K to give 共1⫻1兲 structure while Al still segregates. The specimen was exposed to oxygen at different temperatures. The amount of oxygen uptake was not structure dependent but temperature dependent. Below 595 K, only a small amount of oxygen adsorbed. Between 595 and 870 K, oxygen adsorbed surface showed amorphous LEED pattern. The specimen was annealed at 1070 K after oxygen exposure. When the specimen was exposed oxygen below 870 K, the oxygen Auger intensity decreased significantly by annealing and the annealed surface showed 共)⫻)兲R30° structure at room temperature. When the specimen was exposed to oxygen at 870 K, diffused spots developed newly in LEED pattern but the pattern disappeared after 1070 K annealing while oxygen Auger intensity remained almost constant. Exposing the specimen to oxygen at 995 K resulted in clear spots in the LEED pattern, which were attributed to the 共7/)⫻7)兲R30° structure. © 2003 American Vacuum Society. 关DOI: 10.1116/1.1560719兴

I. INTRODUCTION Copper–aluminum alloy has been used for electric cables for long time because of its high electric conductivity. In the field of large scale integration 共LSI兲, Al instead of Cu has long been used as an electric wiring material for the reasons of the LSI fabrication processes. However, because electromigration phenomenon in Al wire causes severe problems with decreasing width of electrodes, Cu electrodes are now replacing Al electrodes in the LSI technology. One of the problems with Cu electrodes and wirings is corrosion because on Cu no passive oxide layer is formed in contrast on Al. Therefore, oxidation behavior of Cu alloys has become increasingly important as Cu has become promising wiring material for electronic devices. Oxygen uptake on Cu–1.5, 5, and 12 at. %Al共100兲 at room temperature showed saturation at around 50 L.1 Although there is no aluminum enrichment on the clean alloy 共100兲 surface after annealing,1 heating the oxygen-exposed alloy at 600 K resulted in aluminum segregation and formation of aluminum–oxygen bonding on Cu–12 at. %Al共100兲.1 Also on Cu–17 at. %Al共100兲, no Al enrichment was observed on the clean surface, but oxygen exposure at room temperature resulted in Al enrichment at the surface and formation of aluminum oxide.2 a兲

Electronic mail: [email protected]

1290

J. Vac. Sci. Technol. A 21„4…, JulÕAug 2003

The studies on the 共111兲 surface of Cu–Al alloys3– 6 revealed that the alloys with Al more than 9 at. % in bulk composition exhibit Al segregation on the 共111兲 surface. There has been no study on oxygen adsorption on the 共111兲 surface of Cu–Al alloys. We are interested in the initial oxidation on an Al-enriched surface. The Cu–9 at. %Al共111兲 was chosen because: 共1兲 Al segregates on its clean surface; 共2兲 the 共111兲 plane is expected to fit an epitaxial alumina of either ␣ -Al2 O3 (0001) or ␥ -Al2 O3 (111), which is mostly reported epitaxial ultrathin alumina grown on metals and intermetallic compounds;7–10 and 共3兲 it is commercially available. II. EXPERIMENTS Cu–9 at. %Al共111兲 crystal was purchased from the Surface Preparation Laboratory Inc. which was cut within 1° and mechanically polished. The crystal was cleaned by repeating Ar ion sputtering and annealing at 770 K in a low energy electron diffraction/Auger electron spectroscopy 共LEED/AES兲 system. The cleanliness and structure were confirmed by LEED and AES. Oxygen was introduced at different temperatures under 1⫻10⫺7 Torr followed by annealing at 1070 K in ultrahigh vacuum 共UHV兲 for 1 h. After each process, the structure and composition were monitored by LEED and AES. The primary electron energy for LEED was 120 eV, and that for AES was 3 keV. Auger peak intensity ratio, O KLL (503 eV兲/Cu LM M (920 eV) was used for

0734-2101Õ2003Õ21„4…Õ1290Õ4Õ$19.00

©2003 American Vacuum Society

1290


1291

Yoshitake et al.: Oxygen adsorption on Cu-9% At„111…

1291

III. RESULTS AND DISCUSSIONS A. Clean surface

FIG. 1. LEED patterns of three different clean surfaces.

quantification. In situ oxidation was also carried out in a VG AES instrument to obtain additional information on chemical composition. The primary energy of 5 keV was used so that spectra of Al KLL (1390 eV) could be obtained. Temperature was monitored by a pyrometer.

Cu–9 at. % Al共111兲 showed surface segregation of Al at an elevated temperature as reported.3 In Fig. 1, LEED patterns of clean surface with different conditions are shown. At the beginning, the specimen introduced from the atmosphere was sputtered, heated at 725 K, and cooled down in UHV. The LEED pattern showed 共)⫻)兲R30° structure 共b兲 as reported in the Ref. 3. When the crystal with 共)⫻)兲R30° structure was sputtered and heated below 570 K or it was slightly sputtered below 570 K, the surface with 共1⫻1兲 structure was obtained 共a兲. By heating the sputtered surface at or higher than 600 K and cooling down, it showed 共)⫻)兲R30° structure 共b兲 again. If the specimen was kept at a temperature higher than 600 K, 770 K in this case, where the surface with 共1⫻1兲 structure 共c兲 was obtained again. The Auger spectra of Al KLL and Cu LM M corresponding to these three types of surfaces are shown in Fig. 2. Although the background intensity is different with different conditions, the ordinate scale for three spectra is the same. Therefore, it is clear that the Al KLL intensity is stronger on 共b兲 and 共c兲 than on 共a兲. In contrast, the Cu LM M intensity on 共a兲 is stronger than on 共b兲 and 共c兲. It was reported that the Cu–Al alloy with Al content between 9 and 16 at. % showed 共)⫻)兲R30° superstructure that was caused by the ordering of segregated Al.5 We found that Al was preferentially sputtered. Heating below 570 K may not be enough for Al diffusion to segregate. Therefore, Al concentration is depleted in 共a兲 and is stronger in 共b兲. At 770 K, the LEED pattern shows 共1⫻1兲 which is the same as 共a兲, however, Al Auger intensity is the same as 共b兲. It is explained in the following way: surface segregation of Al still occurs at elevated temperature but the segregation sites of Al becomes random due to entropy effect and the 共)⫻)兲

FIG. 2. Auger electron spectra of Al KLL and Cu LM M for the three different surfaces in Fig. 1.

JVST A - Vacuum, Surfaces, and Films


1292


1292

FIG. 4. Auger intensity ratios of O KLL to Cu LM M on oxidized Cu–9 at. %Al共111兲 as a function of annealing temperature after 1200 L oxygen dose at room temperature. FIG. 3. Auger intensity ratios of O KLL to Cu LM M as a function of oxygen dosage.

annealing, the LEED pattern showed 共)⫻)兲R30° and the Auger intensity ratio was reduced to 0.41. The phenomena observed on the clean 共)⫻)兲R30° surface was almost similar to those on the clean 共1⫻1兲 surface except for the LEED patterns on the clean surface. When oxygen was introduced at 595 K, an amorphous LEED pattern appeared with the Auger intensity ratio of 1.6. The 共)⫻)兲R30° LEED pattern appeared after annealing where the Auger intensity ratio decreased to 0.77. At 670 K, observation of the LEED pattern and Auger intensity gave almost the same results as those at 595 K. When oxygen was introduced at 870 K, the Auger intensity ratio was higher, 2.5, and some spots appeared in the LEED pattern in addition to the amorphous pattern. After annealing, however, the 共)⫻)兲R30° LEED pattern appeared with the Auger intensity ratio of 2.04, which was much higher than other cases. At present, the origin of the 共)⫻)兲R30° pattern is not clear because the amount of oxygen varied for the same pattern. The thermal stability of adsorbed oxygen was studied by adsorbing oxygen at room temperature and heating its surface in a vacuum. Figure 4 shows the variation of the Auger intensity ratio of O KLL to Cu LM M with temperature rise. It can be seen that the ratio decreased with temperature gradually and dropped steeply above 970 K. It suggests that at this critical temperature absorbed oxygen was given enough thermal energy to move. We introduced 1200 L oxygen at around this critical temperature and obtained the wellordered alumina layer.

R30° superstructure vanishes. We observed the reversible phase transition between 共)⫻)兲R30° and 共1⫻1兲 at 580 K, which well coincides with the reported transition at 575 K in Cu–12.5 at. %Al alloy.4 B. Initial oxidation

Oxygen was introduced on different types of surfaces and at different temperatures. The amount of oxygen uptake was measured as a function of oxygen dosage and the results are shown in Fig. 3. It turned out that the absolute amount of oxygen uptake was largely dependent on the surface morphology, but we confirmed that the amount of oxygen uptake was not structure dependent but temperature dependent. Small uptake was observed at 495 K, either on 共)⫻)兲 R30° or 共1⫻1兲 structures. More oxygen adsorbed on the surface at 595 K or at higher temperature. At 1105 K, the speed of oxygen uptake was high. Table I shows the variation of the surface structure and the Auger intensity ratio at different temperatures under different treatments. When oxygen was introduced on the 共1⫻1兲 surface at 495 K, the LEED pattern shows diffused 共1⫻1兲 and the Auger intensity ratio of O KLL to Cu LM M was 0.6. By annealing this surface at 1070 K, the Auger intensity ratio decreased to 0.45 and the LEED pattern showed 共)⫻)兲 R30°. On the clean 共)⫻)兲 R30° surface at 495 K, the oxygen dose resulted in a 共1⫻1兲 LEED pattern with an Auger intensity ratio of 0.6. After

TABLE I. Summary of LEED patterns and AES intensity ratios at different surface conditions.

Clean Oxygen exposure

Annealed at 1070 K 共observed at RT兲

495 K

495 K

595 K

670 K

870 K

LEED LEED

1⫻1 diffused 1⫻1

()⫻))R30° 1⫻1

1⫻1 amorphous

1⫻1 amorphous

AES O/Cu LEED AES O/Cu

0.6

0.6

1.6

1.6

1⫻1 amorphous ⫹spots 2.5

共)⫻)兲R30° 0.45

共)⫻)兲R30° 0.41

共)⫻)兲R30° 0.77

共)⫻)兲R30° 0.7

共)⫻)兲R30° 2.04

J. Vac. Sci. Technol. A, Vol. 21, No. 4, JulÕAug 2003


1293


1293

FIG. 5. LEED pattern of the well-ordered alumina on Cu–9 at. %Al共111兲共a兲 and its schematic representation 共b兲. Large closed circles are diffractions from the substrate and the small open circles are those from well-ordered alumina.

C. Well-ordered layer

The LEED pattern of the well-ordered alumina and its schematic representation are shown in Fig. 5. The diffraction spots marked with big closed circles come from the substrate and the small open circles show the diffraction from the well-ordered alumina. The structure of well-ordered alumina is attributed to (7/冑3⫻7/冑3)R30°. From the Auger spectra analysis, it has confirmed that only Al is oxidized11 and we can attribute the new LEED spots to alumina. Details of the structure assignment are discussed in Ref. 11. In this article, the condition for the well-ordered growth is discussed by comparing well-ordered alumina growth in other systems. Ultrathin well-ordered alumina growth has been reported on NiAl共110兲9 and Ni3 Al. 10 From the results on NiAl共110兲 in Ref. 12, temperature around 1070 K seems necessary for atoms in amorphous alumina to crystallize. We also obtained the same result on crystallization and concluded that the two-step process, oxygen uptake for stoichiometric alumina formation and crystallization, is important for well-ordered alumina growth on NiAl共110兲.13 On Cu–9 at. %Al共111兲, the strength of interaction between substrate atom and alumina is weaker and amorphous alumina seems to evaporate at 970 K 共Fig. 4兲, which is well below the crystallization temperature. Therefore, the two-step process, oxygen uptake for stoichiometric alumina formation and crystallization, is not applicable. Oxygen was introduced at higher temperature so that oxygen uptake and crystallization occur at the same time. This procedure seemed to work and we were successful in obtaining the well-ordered alumina layer. Figure 6 shows the thermal stability of the well-ordered alumina. The alumina was stable up to 1100 K but started decomposing at 1130 K 共for reference: melting point of the alloy is 1330 K兲. By comparing Fig. 6 with Fig. 4, one can see that the well-ordered alumina is more stable than amorphous alumina, which is grown at room temperature. This stability difference caused by the crystal structure is similar on NiAl共110兲, where amorphous alumina decomposed at 1070 K14 but well-ordered alumina was stable up to 1300 K.12 However, the well-ordered alumina on Cu–9 at. %Al共111兲 is less stable than that on NiAl共110兲. This is consistent with the above assumption that the strength of interaction between Cu–9 at. %Al共111兲 substrate and alumina is weaker than that between NiAl共110兲 and alumina.

FIG. 6. Thermal stability of well-ordered alumina with annealing in UHV, monitored by the variation of Auger intensity ratios of O KLL to Cu LM M .

IV. CONCLUSION Oxygen adsorption on the Cu–9 at. %Al共111兲 surface with different structures and at different temperatures has been studied by AES and LEED. Cu–9 at. %Al共111兲 showed surface segregation of Al when its sputtered surface was heated above 570 K and the 共)⫻)兲R30° superstructure caused by the ordering of segregated Al was observed below 580 K. The amount of saturated oxygen uptake was only dependent on the specimen temperature but not on the surface structure or surface concentration of Al. The oxygen uptake behavior can be divided into three temperature ranges: 共1兲 low uptake with slow uptake speed below 570 K, 共2兲 high uptake with intermediate uptake speed between 595 and 1040 K, and 共3兲 high uptake with high uptake speed above 1100 K. Only when oxygen was introduced around 995 K did we obtain well-ordered alumina which has (7/冑3⫻7/冑3)R30° structure.

A. Hoffman, T. Maniv, and M. Folman, Surf. Sci. 193, 57 共1988兲. W. Shen, G. L. Nyberg, and A. Hoffman, Surf. Sci. 334, 209 共1995兲. 3 J. Ferrante, Acta Metall. 19, 743 共1971兲. 4 R. J. Baird and W. Eberhardt, J. Vac. Sci. Technol. 18, 538 共1981兲. 5 R. J. Baird and T. J. Potter, J. Vac. Sci. Technol. A 3, 1371 共1985兲. 6 R. J. Baird, D. F. Ogletree, M. A. van Hove, and G. A. Somorjai, Surf. Sci. 165, 345 共1986兲. 7 P. J. Chen and D. W. Goodman, Surf. Sci. Lett. 312, L767 共1994兲. 8 Y. Wu, E. Garfunckel, and T. E. Madey, Surf. Sci. 365, 337 共1996兲. 9 R. M. Jaeger, H. Kuhlenbeck, H.-J. Freund, M. Wuttig, W. Hoffmann, R. Franchy, and H. Ibach, Surf. Sci. 259, 235 共1991兲. 10 U. Bardi, A. Atrei, and G. Rovida, Surf. Sci. 268, 87 共1992兲. 11 M. Yoshitake, S. Bera, Y. Yamauchi, and W. Song, Surf. Sci. 共submitted兲; Presented at ECOSS-21, June 24 –28, 2002. 12 K. F. McCarty, Surf. Sci. Lett. 474, L165 共2001兲. 13 T. Lay, M. Yoshitake, and B. Mebarki, J. Vac. Sci. Technol. A 20, 2027 共2002兲; Presented at American Vacuum Society 48th Conference, October 29–November 2, 2001. 14 M. Yoshitake and B. Mebarki 共unpublished data兲. 1 2

JVST A - Vacuum, Surfaces, and Films
