- Email: [email protected]
Ion beam synthesized cobalt germanide alloy by metal vapor vacuum arc implantation
Nuclear Instruments and Methods in Physics Research B 148 (1999) 604±609
Ion beam synthesized cobalt germanide alloy by metal vapor vacuum arc implantation C.S. Lee, I.H. Wilson *, W.Y. Cheung, Y.J. Chen, J.B. Xu, S.P. Wong Department of Electronic Engineering and Materials Science and Technology Research Center, The Chinese University of Hong Kong, Shatin, N. T., Hong Kong, People's Republic of China
Abstract Germanide formation and evolution in Co-implanted germanium have been studied by h±2h X-ray-diraction (XRD), atomic force microscopy (AFM) and the Rutherford backscattering spectrometry (RBS). A metal vapor vacuum arc (MEVVA) ion source implanter operated at 60 kV was used to implant Co into Ge(1 0 0) wafers with doses ranging from 6 ´ 1014 to 6 ´ 1018 cmÿ2 , and mean current density being either 15 or 210 lA cmÿ2 . The sequence of phase formation was obtained by XRD. Dramatic cellular and columnar nanostructures were observed by AFM in the low and high mean current density samples, respectively. Information on the roughness was obtained. The degree of crystallinity and the thickness of the damaged layer were determined by RBS and ion channeling. The phase formation and evolution are discussed with respect to beam heating eect and the heat of formation. Ó 1999 Elsevier Science B.V. All rights reserved. PACS: 61.72.T; 61.10.N; 61.16.C; 61.85 Keywords: Ion beam synthesis; Cobalt germanide; X-ray diraction; Atomic force microscopy; Rutherford backscattering spectroscopy
1. Introduction Because of the increasing packing density and speed in the VLSI technology, reducing the interconnection resistance has long been an important issue. Metal silicides such as titanium disilicide (TiSi2 ) and cobalt disilicide (CoSi2 ) have been studied for this purpose due to their low resistivity and good thermal stability [1±3].
* Corresponding author. Tel.: +852 26098276; fax: +852 26035558; e-mail: [email protected]
In the last few years, Si1ÿx Gex alloy system has proved to be useful in optoelectronics, high-speed switching applications and other band gap engineered devices [4,5]. Many of these device structures require a metal±Si1ÿx Gex Ohmic contact with low resistivity. Titanium germanide (TiGe2 ) and cobalt germanide (CoGe2 ) have been shown to have a resistivity comparable to those of the corresponding silicides [6,7]. With the development of the metal vapor vacuum arc (MEVVA) ion source, invented by Brown et al. [8], metal ions with high current density can readily be extracted. In this article, we report the
0168-583X/98/$ ± see front matter Ó 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 5 8 3 X ( 9 8 ) 0 0 8 4 2 - 8
C.S. Lee et al. / Nucl. Instr. and Meth. in Phys. Res. B 148 (1999) 604±609
formation of cobalt germanide by direct Co-ion implantation into single crystal Ge(1 0 0). 2. Experiments The wafers used were chemomechanically polished n-type Ge(1 0 0), <0.4 X cm. The implantation was carried out in a JYZ-8010W MEVVA ion source implanter in normal incidence with an extraction voltage of 60 kV, ion doses ranging from 6 ´ 1014 to 6´1018 cmÿ2 . The substrate temperature is monitored by a thermocouple. The implanter was operated in pulse mode with mean current density being either 15 or 210 lA cmÿ2 . Because of the multiple charges nature of MEVVA, the particle ¯ux corresponded to 34% Co , 59% Co2 and 7% Co3 [9]. The maximum temperature and the implantation time are shown in Table 1. XRD measurements were conducted by a SIEMENS D5005 X-Ray Diractometer with a Cu Ka source in a h±2h con®guration. AFM (1.540560 A) observations were made in air at room temperature using Digital Instruments Nanoscope III, operating in the tapping mode. Oine image processing of ®rst order ¯attening and third order plane®t were performed throughout. RBS random spectra with samples tilted at 7° and ion channel-
605
ing spectra were obtained by HV Tandetron Accelerator at a 170° scattering angle with 2.0 MeV 4 He2 particles. 3. Results and discussion 3.1. XRD In the low mean current density (15 lA cmÿ2 ) samples, no germanide can be observed, until the dose comes up to 6 ´ 1017 cmÿ2 where Co3 Ge2 [10] was formed. The XRD spectra of three samples implanted by a high mean current density (210 lA cmÿ2 ) are shown in Fig. 1. Starting from a dose of 6 ´ 1015 and up to 6 ´ 1016 cmÿ2 (Fig. 1(a)) both Co3 Ge2 and Co5 Ge7 were formed. By increasing the dose, CoGe2 started to form. As can be seen in Fig. 1(b), at a dose of 6 ´ 1017 cmÿ2 , no Co5 Ge7 is observed, Co3 Ge2 still exists. More diraction peaks of CoGe2 are seen in the higher dose than the lower one signaling a better crystallinity. When the dose was up to 1 ´ 1018 cmÿ2 , Co5 Ge7 reformed and more diraction peaks of Co5 Ge7 are observed at a dose of 6 ´ 1018 cmÿ2 (Fig. 1(c)). A summary of the XRD results are given in Table 1. Therefore the sequence of phase formation of cobalt germanides are:
Table 1 Summary of temperature, implantation time and XRD results of the Co as-implanted Ge samples Dose/cmÿ2
15 lA cmÿ2
ÿ2
210 lA cm
Temp/°C
Implant. Time/min
XRD results Co3 Ge2
6 ´ 1014 1 ´ 1015 6 ´ 1015 1 ´ 1016 6 ´ 1016 1 ´ 1017 6 ´ 1017
36 53 53 60 98 105 110
<1 <1 1 2 13 38 289
+
15
93 107 150 258 360 430 509
<1 <1 3 18 30 46 343
+ + + + + + +
6 ´ 10 1 ´ 1016 6 ´ 1016 1 ´ 1017 6 ´ 1017 1 ´ 1018 6 ´ 1018
Co5 Ge7
+ + + + +
CoGe2
+ + + +
606
C.S. Lee et al. / Nucl. Instr. and Meth. in Phys. Res. B 148 (1999) 604±609
Comparing sequences (1) and (2), the ®rst phase formed and the order of formation of the other two phases match with that predicted by the thermodynamics. It is suggested that more than one phase forms at one time because forming a mixture of these phases can lower the energy of such a non-equilibrium system. Co5 Ge7 reformed at higher temperatures because the thermodynamic factors no longer dominate the whole reaction process. Kinetic constraints have to be taken into consideration in the later stage. Fig. 1. XRD spectra of Co implanted Ge at a condition of: (a) 6 ´ 1016 cmÿ2 , 210 lA cmÿ2 ; (b) 6 ´ 1017 cmÿ2 , 210 lA cmÿ2 ; (c) 6 ´ 1018 cmÿ2 , 210 lA cmÿ2 .
Co3 Ge2 ! Co3 Ge2 ; Co5 Ge7 ! Co3 Ge2 ; CoGe2 ! Co3 Ge2 ; CoGe2 ; Co5 Ge7 : 1 The results are readily expected by considering the beam heating eect and the implantation time (Table 1) of each sample. Co3 Ge2 was formed in samples of 6 ´ 1017 cmÿ2 , 15 lA cmÿ2 and 6 ´ 1015 cmÿ2 , 210 lA cmÿ2 . The dramatic dierences in dose and implantation time indicate the temperature of the reaction to be the dominating factor for phase formation in the early stage. Co5 Ge7 formed at 6 ´ 1015 cmÿ2 , 210 lA cmÿ2 and then transformed completely into CoGe2 at 1 ´ 1017 cmÿ2 with temperature around 258°C. Co5 Ge7 reformed at higher dose around 430°C. The temperatures of phase formation are comparable to that reported before (CoGe2 : 300°C, Co5 Ge7 : 425°C) [7]. The sequence of phase formation due to implantation has always been under debate [11]. From a solely thermodynamic point of view, the phase with the most negative heat of formation has the largest driving force and is expected to form ®rst. As the values of the heat of formation of Co±Ge compounds are not available, we infer their relative values from the Co±Ge binary phase diagram according to Lui [12]. Referring to the phase diagram given by Hansen et al. [13], the heat of formation of Co±Ge compounds in ascending order is determined below: Co3 Ge2 ! Co5 Ge7 ! CoGe2 :
2
3.2. AFM As dimensions are being scaled down, it is of interest to gain some idea of the roughness introduced by implantation. Fig. 2 shows the tapping mode AFM micrographs of the low and high mean current density samples. For the low mean current density samples, cellular nanostructures are observed in the doses ranging from 6 ´ 1015 to 6 ´ 1016 cmÿ2 . At a dose of 1 ´ 1017 cmÿ2 (Fig. 2(b)), a complex porous structure is seen. The Digital software obtains the root mean square roughness (RMS roughness) of each sample and is plotted in Fig. 3. The RMS roughness follows quite well with an exponential function. For the high mean current density samples, completely dierent morphologies are seen (Figs. 2(c) and (d)). Protrusions are formed on top of some secondary features. The RMS roughness goes down ®rst with increasing dose, then goes up again, following quite well with a third order polynomial. At 6´1017 cmÿ2 , the protrusions agglomerated and formed islands. 3.3. RBS and ion channeling Typical RBS and ion channeling spectra for the low and high mean current density samples are shown in Figs. 4(a) and (b), respectively. As revealed in Fig. 4(a), in the low mean current density samples, for a dose up to 6 ´ 1015 cmÿ2 , the random spectra do not show much dierence from a virgin crystal. Starting from 6 ´ 1016 cmÿ2 , a large yield de®cit occurs in the near surface region of the random spectra when compared with the virgin one. The yield de®cit may be due to: (i) the stopping cross-section of the scattering particles with
C.S. Lee et al. / Nucl. Instr. and Meth. in Phys. Res. B 148 (1999) 604±609
607
Fig. 2. Tapping mode AFM micrographs of Co implanted Ge at a condition of: (a) 1 ´ 1016 cmÿ2 , 15 lA cmÿ2 , 800 nm ´ 800 nm, 70 nm in height scale; (b) 1 ´ 1017 cmÿ2 , 15 lA cmÿ2 , 5 lm ´ 5 lm, 500 nm; (c) 1 ´ 1016 cmÿ2 , 210 lA cmÿ2 , 1 lm ´ 1 lm, 34 nm; (d) 1 ´ 1017 cmÿ2 , 210 lA cmÿ2 , 1 lm ´ 1 lm, 240 nm.
608
C.S. Lee et al. / Nucl. Instr. and Meth. in Phys. Res. B 148 (1999) 604±609
Fig. 3. RMS roughness vs dose for the low and high mean current density samples.
the presence of the implanted Co atoms does not follow strictly the Bragg rule, which is the assumption taken by the simulation software. However, this may not be the full explanation as a similar de®cit is seen in Ge self-ion implantation at 60 kV, 6 ´ 1016 cmÿ2 , 15 lA cmÿ2 (spectrum not shown in here); (ii) contamination by impurities such as C and O due to the drastically increased surface area as suggested by Holland et al. [14]. We carried out two non-Rutherford scattering experiments, including 3.04 MeV a±16 O and 1.76 MeV p±12 C. The determined concentrations of C and O in the near surface region are only comparable to the virgin one, so this reason may not apply in our data; (iii) decrease of density of Ge in the near surface region due to the dramatic cellular and porous structures as shown in the previous AFM micrographs. The eect of severe roughness and porosity is the rounding of the leading edge of the random spectrum as the yields that originally count in one channel are now spread over the neighbors. The channeling spectra show that the thickness of the damaged layer increases with increasing dose. The thickness is counted from the surface up to the end of the back edge of the channeling yield. For the high mean current density samples, though the random spectra are similar, their channeling spectra show a much smaller vmin , indicating a higher degree of crystallinity, and
Fig. 4. RBS and ion channeling spectra of Co implanted Ge at a condition of; (a) low mean current density, 6 ´ 1015 , 6 ´ 1017 cmÿ2 ; (b) high mean current density, 6 ´ 1015 , 6 ´ 1017 cmÿ2 .
no speci®c damaged layer can be seen. This is attributed to the higher self-annealing temperature. 4. Conclusions Cobalt germanides can be formed directly by high mean current density Co-ion implantation without postannealing. Because the high mean current density can readily produce the high temperatures required for the formation of various phases. From XRD, Co3 Ge2 , Co5 Ge7 and CoGe2 were formed in the as-implanted samples. Dramatic cellular and columnar nanostructures were observed in the low and high mean current density samples respectively. RBS reveals a yield de®cit in the near surface region which may be attributed to the stopping cross-section not following strictly the Bragg rule, or the decrease of Ge density due to the cellular and porous structures.
C.S. Lee et al. / Nucl. Instr. and Meth. in Phys. Res. B 148 (1999) 604±609
Acknowledgements Special thanks to Dr. Sundaravel for his help in doing the non-Rutherford scattering experiments and precious discussion, and Dr. E.Z. Luo for his help in maintaining the RBS system. References [1] L. Van den Hove, R. Wolters, K. Maex, R.F. Dekeersmaecker, G.J. Declerck, IEEE Trans. Electron Devices, ED-34 (1987) 554. [2] M.E. Alperin, T.C. Holloway, R.A. Haken, C.D. Gosmeyer, R.V. Karnaugh, W.D. Parmanite, IEEE Trans. Electron Devices, ED-32 (1985) 141. [3] L. Van den Hove, R. Wolters, K. Maex, R. Dekeersmaecker, G. Declerck, J. Vac. Sci. Technol. B4 (1986) 1358.
609
[4] G.L. Patton, S.S. Lyer, S.L. Delage, S. Tiwari, J.M.C. Stork, IEEE Electron Device Lett. 9 (1988) 165. [5] Y. Rajekarunanayake, J.C. McGill, Appl. Phys. Lett. 55 (1989) 1537. urk, J.J. Wortman, G. Harris, J. [6] S.P. Ashburn, M.C. Ozt Honeycutt, D.M. Maher, J. Electron Mater. 21 (1992) 81. urk, G. Harris, D.M. Maher, J. [7] S.P. Ashburn, M.C. Ozt Appl. Phys. 74 (7) (1993) 4455. [8] I.G. Brown, J.E. Gavin, R.A. MacGrill, Appl. Phys. Lett. 47 (1985) 358. [9] Andre Anders, Physical Review E 55(1) (1997) 969. [10] The interplanar spacings are matched with those listed in JCPDS, card no. 30-435, 29-474, 4-545, 7-162. [11] W. Xia, C.A. Hewett, M. Fernandes, S.S. Lau, D.B. Poker, J. Appl. Phys. 65 (6) (1989) 2300. [12] Rui-Qing Li, Can. Metall. Q. 29(4) (1990) 279. [13] M. Hansen, K. Anderko, Constitution of Binary Alloys, McGraw Hill, New York, 1958, p. 476. [14] O.W. Holland, B.R. Appleton, J. Narayan, J. Appl. Phys. 54 (5) (1983) 2295.
Ion beam synthesized cobalt germanide alloy by metal vapor vacuum arc implantation C.S. Lee, I.H. Wilson *, W.Y. Cheung, Y.J. Chen, J.B. Xu, S.P. Wong Department of Electronic Engineering and Materials Science and Technology Research Center, The Chinese University of Hong Kong, Shatin, N. T., Hong Kong, People's Republic of China
Abstract Germanide formation and evolution in Co-implanted germanium have been studied by h±2h X-ray-diraction (XRD), atomic force microscopy (AFM) and the Rutherford backscattering spectrometry (RBS). A metal vapor vacuum arc (MEVVA) ion source implanter operated at 60 kV was used to implant Co into Ge(1 0 0) wafers with doses ranging from 6 ´ 1014 to 6 ´ 1018 cmÿ2 , and mean current density being either 15 or 210 lA cmÿ2 . The sequence of phase formation was obtained by XRD. Dramatic cellular and columnar nanostructures were observed by AFM in the low and high mean current density samples, respectively. Information on the roughness was obtained. The degree of crystallinity and the thickness of the damaged layer were determined by RBS and ion channeling. The phase formation and evolution are discussed with respect to beam heating eect and the heat of formation. Ó 1999 Elsevier Science B.V. All rights reserved. PACS: 61.72.T; 61.10.N; 61.16.C; 61.85 Keywords: Ion beam synthesis; Cobalt germanide; X-ray diraction; Atomic force microscopy; Rutherford backscattering spectroscopy
1. Introduction Because of the increasing packing density and speed in the VLSI technology, reducing the interconnection resistance has long been an important issue. Metal silicides such as titanium disilicide (TiSi2 ) and cobalt disilicide (CoSi2 ) have been studied for this purpose due to their low resistivity and good thermal stability [1±3].
* Corresponding author. Tel.: +852 26098276; fax: +852 26035558; e-mail: [email protected]
In the last few years, Si1ÿx Gex alloy system has proved to be useful in optoelectronics, high-speed switching applications and other band gap engineered devices [4,5]. Many of these device structures require a metal±Si1ÿx Gex Ohmic contact with low resistivity. Titanium germanide (TiGe2 ) and cobalt germanide (CoGe2 ) have been shown to have a resistivity comparable to those of the corresponding silicides [6,7]. With the development of the metal vapor vacuum arc (MEVVA) ion source, invented by Brown et al. [8], metal ions with high current density can readily be extracted. In this article, we report the
0168-583X/98/$ ± see front matter Ó 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 5 8 3 X ( 9 8 ) 0 0 8 4 2 - 8
C.S. Lee et al. / Nucl. Instr. and Meth. in Phys. Res. B 148 (1999) 604±609
formation of cobalt germanide by direct Co-ion implantation into single crystal Ge(1 0 0). 2. Experiments The wafers used were chemomechanically polished n-type Ge(1 0 0), <0.4 X cm. The implantation was carried out in a JYZ-8010W MEVVA ion source implanter in normal incidence with an extraction voltage of 60 kV, ion doses ranging from 6 ´ 1014 to 6´1018 cmÿ2 . The substrate temperature is monitored by a thermocouple. The implanter was operated in pulse mode with mean current density being either 15 or 210 lA cmÿ2 . Because of the multiple charges nature of MEVVA, the particle ¯ux corresponded to 34% Co , 59% Co2 and 7% Co3 [9]. The maximum temperature and the implantation time are shown in Table 1. XRD measurements were conducted by a SIEMENS D5005 X-Ray Diractometer with a Cu Ka source in a h±2h con®guration. AFM (1.540560 A) observations were made in air at room temperature using Digital Instruments Nanoscope III, operating in the tapping mode. Oine image processing of ®rst order ¯attening and third order plane®t were performed throughout. RBS random spectra with samples tilted at 7° and ion channel-
605
ing spectra were obtained by HV Tandetron Accelerator at a 170° scattering angle with 2.0 MeV 4 He2 particles. 3. Results and discussion 3.1. XRD In the low mean current density (15 lA cmÿ2 ) samples, no germanide can be observed, until the dose comes up to 6 ´ 1017 cmÿ2 where Co3 Ge2 [10] was formed. The XRD spectra of three samples implanted by a high mean current density (210 lA cmÿ2 ) are shown in Fig. 1. Starting from a dose of 6 ´ 1015 and up to 6 ´ 1016 cmÿ2 (Fig. 1(a)) both Co3 Ge2 and Co5 Ge7 were formed. By increasing the dose, CoGe2 started to form. As can be seen in Fig. 1(b), at a dose of 6 ´ 1017 cmÿ2 , no Co5 Ge7 is observed, Co3 Ge2 still exists. More diraction peaks of CoGe2 are seen in the higher dose than the lower one signaling a better crystallinity. When the dose was up to 1 ´ 1018 cmÿ2 , Co5 Ge7 reformed and more diraction peaks of Co5 Ge7 are observed at a dose of 6 ´ 1018 cmÿ2 (Fig. 1(c)). A summary of the XRD results are given in Table 1. Therefore the sequence of phase formation of cobalt germanides are:
Table 1 Summary of temperature, implantation time and XRD results of the Co as-implanted Ge samples Dose/cmÿ2
15 lA cmÿ2
ÿ2
210 lA cm
Temp/°C
Implant. Time/min
XRD results Co3 Ge2
6 ´ 1014 1 ´ 1015 6 ´ 1015 1 ´ 1016 6 ´ 1016 1 ´ 1017 6 ´ 1017
36 53 53 60 98 105 110
<1 <1 1 2 13 38 289
+
15
93 107 150 258 360 430 509
<1 <1 3 18 30 46 343
+ + + + + + +
6 ´ 10 1 ´ 1016 6 ´ 1016 1 ´ 1017 6 ´ 1017 1 ´ 1018 6 ´ 1018
Co5 Ge7
+ + + + +
CoGe2
+ + + +
606
C.S. Lee et al. / Nucl. Instr. and Meth. in Phys. Res. B 148 (1999) 604±609
Comparing sequences (1) and (2), the ®rst phase formed and the order of formation of the other two phases match with that predicted by the thermodynamics. It is suggested that more than one phase forms at one time because forming a mixture of these phases can lower the energy of such a non-equilibrium system. Co5 Ge7 reformed at higher temperatures because the thermodynamic factors no longer dominate the whole reaction process. Kinetic constraints have to be taken into consideration in the later stage. Fig. 1. XRD spectra of Co implanted Ge at a condition of: (a) 6 ´ 1016 cmÿ2 , 210 lA cmÿ2 ; (b) 6 ´ 1017 cmÿ2 , 210 lA cmÿ2 ; (c) 6 ´ 1018 cmÿ2 , 210 lA cmÿ2 .
Co3 Ge2 ! Co3 Ge2 ; Co5 Ge7 ! Co3 Ge2 ; CoGe2 ! Co3 Ge2 ; CoGe2 ; Co5 Ge7 : 1 The results are readily expected by considering the beam heating eect and the implantation time (Table 1) of each sample. Co3 Ge2 was formed in samples of 6 ´ 1017 cmÿ2 , 15 lA cmÿ2 and 6 ´ 1015 cmÿ2 , 210 lA cmÿ2 . The dramatic dierences in dose and implantation time indicate the temperature of the reaction to be the dominating factor for phase formation in the early stage. Co5 Ge7 formed at 6 ´ 1015 cmÿ2 , 210 lA cmÿ2 and then transformed completely into CoGe2 at 1 ´ 1017 cmÿ2 with temperature around 258°C. Co5 Ge7 reformed at higher dose around 430°C. The temperatures of phase formation are comparable to that reported before (CoGe2 : 300°C, Co5 Ge7 : 425°C) [7]. The sequence of phase formation due to implantation has always been under debate [11]. From a solely thermodynamic point of view, the phase with the most negative heat of formation has the largest driving force and is expected to form ®rst. As the values of the heat of formation of Co±Ge compounds are not available, we infer their relative values from the Co±Ge binary phase diagram according to Lui [12]. Referring to the phase diagram given by Hansen et al. [13], the heat of formation of Co±Ge compounds in ascending order is determined below: Co3 Ge2 ! Co5 Ge7 ! CoGe2 :
2
3.2. AFM As dimensions are being scaled down, it is of interest to gain some idea of the roughness introduced by implantation. Fig. 2 shows the tapping mode AFM micrographs of the low and high mean current density samples. For the low mean current density samples, cellular nanostructures are observed in the doses ranging from 6 ´ 1015 to 6 ´ 1016 cmÿ2 . At a dose of 1 ´ 1017 cmÿ2 (Fig. 2(b)), a complex porous structure is seen. The Digital software obtains the root mean square roughness (RMS roughness) of each sample and is plotted in Fig. 3. The RMS roughness follows quite well with an exponential function. For the high mean current density samples, completely dierent morphologies are seen (Figs. 2(c) and (d)). Protrusions are formed on top of some secondary features. The RMS roughness goes down ®rst with increasing dose, then goes up again, following quite well with a third order polynomial. At 6´1017 cmÿ2 , the protrusions agglomerated and formed islands. 3.3. RBS and ion channeling Typical RBS and ion channeling spectra for the low and high mean current density samples are shown in Figs. 4(a) and (b), respectively. As revealed in Fig. 4(a), in the low mean current density samples, for a dose up to 6 ´ 1015 cmÿ2 , the random spectra do not show much dierence from a virgin crystal. Starting from 6 ´ 1016 cmÿ2 , a large yield de®cit occurs in the near surface region of the random spectra when compared with the virgin one. The yield de®cit may be due to: (i) the stopping cross-section of the scattering particles with
C.S. Lee et al. / Nucl. Instr. and Meth. in Phys. Res. B 148 (1999) 604±609
607
Fig. 2. Tapping mode AFM micrographs of Co implanted Ge at a condition of: (a) 1 ´ 1016 cmÿ2 , 15 lA cmÿ2 , 800 nm ´ 800 nm, 70 nm in height scale; (b) 1 ´ 1017 cmÿ2 , 15 lA cmÿ2 , 5 lm ´ 5 lm, 500 nm; (c) 1 ´ 1016 cmÿ2 , 210 lA cmÿ2 , 1 lm ´ 1 lm, 34 nm; (d) 1 ´ 1017 cmÿ2 , 210 lA cmÿ2 , 1 lm ´ 1 lm, 240 nm.
608
C.S. Lee et al. / Nucl. Instr. and Meth. in Phys. Res. B 148 (1999) 604±609
Fig. 3. RMS roughness vs dose for the low and high mean current density samples.
the presence of the implanted Co atoms does not follow strictly the Bragg rule, which is the assumption taken by the simulation software. However, this may not be the full explanation as a similar de®cit is seen in Ge self-ion implantation at 60 kV, 6 ´ 1016 cmÿ2 , 15 lA cmÿ2 (spectrum not shown in here); (ii) contamination by impurities such as C and O due to the drastically increased surface area as suggested by Holland et al. [14]. We carried out two non-Rutherford scattering experiments, including 3.04 MeV a±16 O and 1.76 MeV p±12 C. The determined concentrations of C and O in the near surface region are only comparable to the virgin one, so this reason may not apply in our data; (iii) decrease of density of Ge in the near surface region due to the dramatic cellular and porous structures as shown in the previous AFM micrographs. The eect of severe roughness and porosity is the rounding of the leading edge of the random spectrum as the yields that originally count in one channel are now spread over the neighbors. The channeling spectra show that the thickness of the damaged layer increases with increasing dose. The thickness is counted from the surface up to the end of the back edge of the channeling yield. For the high mean current density samples, though the random spectra are similar, their channeling spectra show a much smaller vmin , indicating a higher degree of crystallinity, and
Fig. 4. RBS and ion channeling spectra of Co implanted Ge at a condition of; (a) low mean current density, 6 ´ 1015 , 6 ´ 1017 cmÿ2 ; (b) high mean current density, 6 ´ 1015 , 6 ´ 1017 cmÿ2 .
no speci®c damaged layer can be seen. This is attributed to the higher self-annealing temperature. 4. Conclusions Cobalt germanides can be formed directly by high mean current density Co-ion implantation without postannealing. Because the high mean current density can readily produce the high temperatures required for the formation of various phases. From XRD, Co3 Ge2 , Co5 Ge7 and CoGe2 were formed in the as-implanted samples. Dramatic cellular and columnar nanostructures were observed in the low and high mean current density samples respectively. RBS reveals a yield de®cit in the near surface region which may be attributed to the stopping cross-section not following strictly the Bragg rule, or the decrease of Ge density due to the cellular and porous structures.
C.S. Lee et al. / Nucl. Instr. and Meth. in Phys. Res. B 148 (1999) 604±609
Acknowledgements Special thanks to Dr. Sundaravel for his help in doing the non-Rutherford scattering experiments and precious discussion, and Dr. E.Z. Luo for his help in maintaining the RBS system. References [1] L. Van den Hove, R. Wolters, K. Maex, R.F. Dekeersmaecker, G.J. Declerck, IEEE Trans. Electron Devices, ED-34 (1987) 554. [2] M.E. Alperin, T.C. Holloway, R.A. Haken, C.D. Gosmeyer, R.V. Karnaugh, W.D. Parmanite, IEEE Trans. Electron Devices, ED-32 (1985) 141. [3] L. Van den Hove, R. Wolters, K. Maex, R. Dekeersmaecker, G. Declerck, J. Vac. Sci. Technol. B4 (1986) 1358.
609
[4] G.L. Patton, S.S. Lyer, S.L. Delage, S. Tiwari, J.M.C. Stork, IEEE Electron Device Lett. 9 (1988) 165. [5] Y. Rajekarunanayake, J.C. McGill, Appl. Phys. Lett. 55 (1989) 1537. urk, J.J. Wortman, G. Harris, J. [6] S.P. Ashburn, M.C. Ozt Honeycutt, D.M. Maher, J. Electron Mater. 21 (1992) 81. urk, G. Harris, D.M. Maher, J. [7] S.P. Ashburn, M.C. Ozt Appl. Phys. 74 (7) (1993) 4455. [8] I.G. Brown, J.E. Gavin, R.A. MacGrill, Appl. Phys. Lett. 47 (1985) 358. [9] Andre Anders, Physical Review E 55(1) (1997) 969. [10] The interplanar spacings are matched with those listed in JCPDS, card no. 30-435, 29-474, 4-545, 7-162. [11] W. Xia, C.A. Hewett, M. Fernandes, S.S. Lau, D.B. Poker, J. Appl. Phys. 65 (6) (1989) 2300. [12] Rui-Qing Li, Can. Metall. Q. 29(4) (1990) 279. [13] M. Hansen, K. Anderko, Constitution of Binary Alloys, McGraw Hill, New York, 1958, p. 476. [14] O.W. Holland, B.R. Appleton, J. Narayan, J. Appl. Phys. 54 (5) (1983) 2295.