- Email: [email protected]
Cathodoluminescence and ion beam analysis of ion-implanted combinatorial materials libraries on thermally grown SiO2
Nuclear Instruments and Methods in Physics Research B 159 (1999) 81±88
www.elsevier.nl/locate/nimb
Cathodoluminescence and ion beam analysis of ion-implanted combinatorial materials libraries on thermally grown SiO2 Chang-Ming Chen *, H.C. Pan, D.Z. Zhu, J. Hu, M.Q. Li
1
Shanghai Institute of Nuclear Research, Chinese Academy of Sciences, P.O. Box 800-204, Shanghai 201800, People's Republic of China Received 1 December 1998; received in revised form 20 April 1999
Abstract A method combining ion implantation and physical masking technique has been used to generate material libraries of various ion-implanted samples. Ion species of C, Ga, N, Pb, Sn, Y have been sequentially implanted to an SiO2 ®lm grown on a silicon wafer through combinatorial masks and consequently a library of 64 (26 ) samples is generated by 6 masking combinations. This approach oers rapid synthesis of samples with potential new compounds formed in the matrix, which may have speci®c luminescent properties. The depth-resolved cathodoluminescence (CL) measurements revealed some speci®c optical property in the samples correlated with implanted ion distributions. A marker-based technique is developed for the convenient location of sample site in the analysis of Rutherford backscattering spectrometry (RBS) and proton elastic scattering (PES), intended to characterize rapidly the ion implanted ®lm libraries. These measurements demonstrate the power of nondestructively and rapidly characterizing composition and the inhomogeneity of the combinatorial ®lm libraries, which may determine their physical properties. Ó 1999 Elsevier Science B.V. All rights reserved.
1. Introduction The combinatorial approach, in which the synthesis and screening of libraries with a large member of dierent organic molecules are implemented, has been widely used for the drug discovery [1]. Recently, Xiang et al. [2] pioneered the application of this approach to the discovery of
* Corresponding author. Tel.: +86-21-5955-2394; fax: +8621-5955-3021; e-mail: [email protected] 1 Tel.: +86-21-5955-2394; fax: +86-21-5955-3021; e-mail: [email protected].
new solid-state inorganic materials with desired properties. Combining combinatorial physical/lithography masks with thin ®lm synthesis technique [2], integrated material chips with a large number of spatially de®ned dierent compounds are generated. Contact and non-contact detection techniques [3] were developed to screen the libraries for desirable physical properties, such as superconductivity, giant magnetoresistance, ecient luminescence and low loss dielectrics [4±8]. The combinatorial technique can be widely used in material science to increase the rate of new material discovery and optimization by several orders of magnitude.
0168-583X/99/$ - see front matter Ó 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 5 8 3 X ( 9 9 ) 0 0 4 4 9 - 8
82
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
During the last few years special interest has been devoted to create new nanocrystals of metals or compounds in SiO2 , which may exhibit speci®c optical emission properties and be interesting for optoelectronic application [9,10]. The ion implantation is attractive for this purpose, because it is compatible with planar microelectronic technology and generates nanocrystals with well-de®ned concentration at pre-calculated depths below the host material surface. Implantation of various individual elements such as Si [11], Ge [12] and C [13] has been reported, which yields dierent luminescent properties. Recently, a sequential ionimplantation of Ga and N has been used to create GaN nanocrystals in sapphire (Al2 O3 ) with speci®c optical property [14]. Our ®rst application of combinatorial approach in ion implantation is intended to increase greatly the eciency of sample synthesis and the possibility of novel nanocrystal formation, which may yield higher energy emissions meeting the needs of high density optical storage. Although the application of combinatorial approach could revolutionize material discovery ± much as it has in pharmaceutical applications [2], the in-depth study of the materials libraries is in some respects restricted by the diculty in phase determination and composition inhomogeneity examination of the ®lm libraries, which may determine their properties. In this paper, we report the application of cathodoluminescence (CL) technique to identify local bonding environments which contribute to the luminescent properties, and the ®rst application of Rutherford backscattering spectrometry (RBS) and proton elastic scattering (PES) method to determine the depthresolved ion distribution of the combinatorial ion implanted samples. 2. Experimental Boron (P-type) doped 0.5 mm thick single crystal silicon wafers (á1 0 0ñ orientation and resistivity of 5.0 X cm) were used for the implantation. SiO2 thin ®lm of 400 nm was formed by thermal oxidation of wafer in H2 + O2 atmosphere. Several elements, C, Ga, N, Pb, Sn and Y, were
selected for implantation considering the potential in forming nanocrystals of GaN, C, Sn, SiC or other unexpected recombination centers in the host-matrices. The energies for the implantation were 40±80 keV and each ion doses are 1±5 ´ 1016 /cm2 , depending on ion species. Implantation of various elements was accomplished at room temperature using a mass-separated beam through an electromagnetic isotope separator (EMIS-SINR-02). The ion beam was electrostatically scanned in a 50 cm2 area to homogenize the surface ion dose distribution. The ion beam was incident on the substrate through a set of physical masks, of which the combination of ion species is arranged in a similar way as described in Ref [2]. The generated material chip (library) was subsequently thermally treated at 800°C for 30 min in ¯owing Ar atmosphere and was further characterized by CL, RBS and PES analysis. Comparing with photoluminescence measurement, cathodoluminescence measurement can oer depth-resolved optical properties due to various penetration of electron beam with dierent accelerating voltage. This would also provide additional local information on the defects and impurities in the matrix [15,16]. The cathodoluminescence measurement system capable of producing beam current of 1±3 lA was employed to this study. The dispersed emission was detected with the HTV-R-446 photomultiplier and WDG05-II monochromator. RBS analysis was performed on the 4 MeV pelletron by using 2 MeV 4 He beams at a scattering angle of 170°. The energy resolution of the detector is about 16 keV and total dose of 4 He was 3 lC. The material chip holder for RBS and PES measurement is schemed in Fig. 1. A rectangular stainless steel substrate is axed with four quartz strips along each side, with each strip length equal to that of the substrate side. A rectangular frame of stainless steel is overlaid on the quartz and stuck by silver paste. On each side of the frame, eight 2.5 mm±spaced (depends on the sample spacing in the chip) holes of 1 mm diameter is produced to expose the quartz surface. The material chip is placed on the top of the stainless steel substrate with sample sites aligned to the holes located in the stainless steel frame. In the
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
83
spots is under ion bombardment to emit reddish± white light, which can be observed either by naked eye or through a telescope. Afterwards, ion beam analysis for the selected sample can be accomplished by moving sample location relative to the reference point. 3. Cathodoluminescences of materials libraries
Fig. 1. Schematic of material chip holder for RBS and PES analysis. The double arrow line represents that the holder is controllably moved along X and Y directions to let the selected sample site aligned with the analyzing beam.
RBS and PES measurement, a primary reference point is ®rst located by moving the chip holder along X and Y directions using a feedthrough manipulator, so that one of the exposed quartz
Fig. 2 shows a material chip (library) of 64 dierent samples generated by ion beam implantation and 6 masking combinations, imaged under natural lighting. The dierent colors on various sample sites may originate from speci®c light re¯ections and interference from dierent interfaces. We also conducted CL measurements of the material chip with various electron-impinging voltages. As the material chip is exposed to an electron beam irradiation of 5 kV, 1 lA, we can easily distinguish the implanted samples from the pristine substrate. Most samples are of blue
Fig. 2. An image of the ion implanted silicon-based material chip with a library of 64 samples prior to annealing under white lighting. The silicon wafer is 25 mm ´ 25 mm and each sample site is of 2 mm diameter. Indicated elements and braced numbers corresponds the implanted ion species and their implantation sequences, respectively. Dierent implantation conditions for various elements are described as: (a) C, 50 keV, 5 ´ 1016 /cm2 ; (b) Ga, 80 keV, 1 ´ 1016 /cm2 ; (c) N, 50 keV, 1.5 ´ 1016 /cm2 ; (d) Pb, 80 keV, 1.5 ´ 1016 /cm2 ; (e) Sn, 70 keV, 1 ´ 1016 /cm2 ; (f) Y, 65 keV, 1 ´ 1016 /cm2 .
84
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
emission, and dierences in color and intensity of emitting light from the samples can be observed by naked eye. Yellow±green color from C containing implanted samples and violet±blue from Sn containing implanted samples are observed under 2 keV electron bombardment, with some major CL peaks revealed in Table 1. As evident in Table 1, the blue (450 nm) luminescent band is observed in all samples, and this is commonly observed in pristine SiO2 and amorphous SiO2 or silica, which is thought to be ascribed to the oxygen vacancy or E0 center and the two-fold-correlated silicon center [17,18]. The 450 nm CL band is caused by electron bombardment and its intensity is increased by increasing electron accelerating voltage to 10 keV. Subsequent implantation of Ga and N into SiO2 is expected to form embedded GaN nanocrystals, which is a great important compound for optoelectronic devices. However, CL measurements for the sample S(Ga, N) did not show any evidences for the existence of GaN crystals in SiO2 and this may be due to N diusion through the ®lm to escape into the air or form Si3 Ox Ny , which can also be proved by their nearly no N existence determined by PES measurements. All samples implanted with C reveals some yellow±green color. The S(C) exhibits a weak additional emission peaked 555 nm. The C presence in the host is hopefully to induce the formation of C-cluster, silicon carbide or silicon oxycarbide nanoparticles, which is of great interest because of the shift of the luminescent emission towards
higher energies compared to Si nanocrystals in SiO2 with 680 nm peak emission, when emissions in the violet range are expected. Indeed, the C can react with O in SiO2 to form CO and/or CO2 , and it is assumed that the reduced Si can react with C to form silicon carbide or silicon oxycarbide when access C is remained in SiO2 . This similar emission band is also observed in silicon±carbon based ®lms prepared by plasma enhanced chemical deposition [19] and is attributed to amorphous Six C1ÿx formation. Co-implantation of C and Ga in SiO2 induce a strong emission peaked at 570 nm, even followed by Pb and Sn implantation. The Ga can react with SiO2 in a form as follows: 4Ga 3SiO2 3Si 2Ga2 O3 :
The Ga2 O3 is a phosphor usually crystallized in highly anisotropy crystal structure [20], forming interconnected tetrahedra or octahedra which surround a large ``tunnel''. It is believed this unique one-directional ``tunnels'' play a major role in contributing to the superior capacity of these phosphors to generate hot electrons to incite the luminescent centers. From the evident diusion of Ga to the C-peaked nearby region after annealing, one may deduce that C implantation would have induced the formation Gax Siy O3 Cx , which is attributed to bonding impetus between Ga and C in SiO2 rather than to the strain induced by implantation damage, as similar phenomena is not observed in N-peaked nearby region. The complex consisting of a gallium vacancy and a carbon atom
Table 1 Characteristic CL spectra for the selected samples under 2 keV electron bombardment Sample S(C) S(N) S(Ga, C) S(Ga, N) S(Ga, Pb, C) S(Ga, Pb, Sn, C) S(Pb, Sn, N) S(Pb, N) a
Dominant peak position (nm)a
415 (0.75) 405 (1) 435 (0.6)
1
450 450 450 445 450 450 450 450
The data in the bracket indicates normalized emission intensity relative to 450 nm.
555 (0.10) 570 (0.56) 567 (0.60) 565 (0.60)
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
may also play a role in the emission of 570 nm in samples co-implanted with C and Ga, as like that happens in C implanted GaN [21]. Table 1 also shows that Sn and Pb co-implantation causes emission shift to higher energy. Fig. 3 shows a comparison of CL spectra for S(Pb, Sn, N) and the prsistine substrate with various electron energies. The peak near 450 nm was observed in the spectra for both implanted sample and pristine SiO2 substrate under the irradiation of electron beam energy of 10 keV, with four times luminescence intensity for implanted sample to that of pristine SiO2 . The luminescence intensity for S(Pb, Sn, N) is enhanced by a factor of 3.3 and 8 with respect to that for the pristine substrate, corresponding impinging energies of 5 and 2 keV, respectively. Also shown in Fig. 3 are the depth-resolved CL spectra for S(Pb, Sn, N) measured with electron beam energies of 2, 5, and 10 keV, respectively. The band peak shifts to lower wavelength region as the beam energy decreases, with peaks at 450, 422 and 405 nm corresponding to beam energies of 10, 5 and 2 keV, respectively. The shifted curves shown in Fig. 3 display the density of electron± hole pairs which indicate speci®c luminescence correlated with an inhomogeneous depth±impurity distribution in the sample.
Fig. 3. Cathodoluminescence spectra of S(Pb, Sn, N) and unimplanted SiO2 at various electron energies. The spectra were measured at the room temperature.
85
We have calculated the energy loss during electron penetrating into SiO2 using a density of 2.25 g/cm3 for SiO2 , as shown in Fig. 4. The concentration of excess electron±hole pairs created in the material is directly proportional to the energy loss as electrons penetrated the solid. Comparing Fig. 3 with Fig. 4, it evidently results that implantation of Pb and Sn in SiO2 causes an increase of pair formation and higher energy CL. The CL peak position of S(Pb, Sn, N) at 10 keV is almost the same with that of SiO2 . This is because the SiO2 area near the Si/SiO2 interface is largely excited in the condition and the correspondingly produced spectrum band may dominate in this case. As the electron energy decreases, the region approaching the surface is largely excited and peak band shifts to higher energies, that is obviously owing to the eects of Pb- and Sn- related luminescent centers in the material, which can match the ion distributions measured by RBS Fig. 6(b). The CL peak shift is often thought to relate to new phase formation or defect-impurity complex [16,22]. Considering the two-fold-coordinated silicon O±Si±O or O±Ge±O as the origin of the blue luminescence in Si1ÿx Gex O2 glass [23], one may expect that O±Pb±O and O±Sn±O may contribute to the peak shifts. On the other hand, one may also think that nanocrystal formation may take an effect, as Sn nanocrystals were observed in Sn implanted SiO2 followed by thermal treatment [24].
Fig. 4. Calculated electron energy loss vs depth during impinging into the SiO2 .
86
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
4. Ion depth-resolved pro®ling by RBS and PES Ion analysis is also used to quantitatively determination of constitution and help elucidating the interaction among implanted ions which may yield correlated speci®c luminescence. The spectra obtained by RBS measurements for the 800°C annealing material chip are partially shown in Fig. 5. Fig. 5(a) shows the spectra of S(Ga, N) and S(Ga, C). The Ga spectrum of S(Ga, N) is peaked at 36 nm from the surface, with no observable peak shift with respect to that of the sample prior to annealing (the spectrum is not shown here). Concerning the N depth range which is estimated to be peaked at about 130 nm from the surface by TRIM code calculation, we can conclude that no GaN formed in the region of N peak, although GaN nanocrystals formation is reported in Ga and N implanted Al2 O3 after annealing [14]. Two peaks of Ga spectrum for S(Ga, C) are revealed in Fig. 5(a), which indicates that the C existence induces the redistribution of Ga in SiO2 after annealing. The pro®le for C distribution is peaked at about 180 nm from the surface by TRIM code calculation, which matches the location of the lower energy peak of Ga spectrum (172 nm from the surface). It is expected that C induced Ga-related compounds may be formed in the SiO2 and that is worth further study. Subsequent implantation of Pb or both Pb and Sn into the samples with previous C implantation did not alter the Ga spectrum peak location, as partially shown in Fig. 5(b). We therefore conclude that the Ga redistribution is aected only by C implantation. From Fig. 5(c), it results that the interaction between Pb and Sn is little comparing the spectra of S(Pb, N) and S(Pb, Sn, N). The previous discussions are also con®rmed by the ion concentration depth-pro®les calculated with RUMP program, which are shown in Fig. 6, corresponding to Fig. 5. The Ga pro®le of S(Ga, C) is peaked at 36 and 172 nm from the surface, corresponding atom fractions of 1.5% and 0.56%, respectively (Fig. 6(a)). The pro®ling curves of S(Ga, Pb, C) and S(Ga, Pb, Sn, C) also each shows two peaks of Ga concentration, located at nearly the same position with respect to that of S(Ga, C). From Fig. 6(b), it reveals that Sn diusion
Fig. 5. 2.0 MeV 4 He RBS spectra from material chip after annealing at 800°C for 30 min in Ar gas. (a) S(Ga, C) and S(Ga, N); (b) S(Ga, Pb, C) and S(Ga, Pb, Sn, C); (c) S(Pb, Sn, N) and S(Pb, N).
approaching the surface for S(Pb, Sn, N) is evident, which is a cause of higher energy luminescent
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
87
Fig. 7. Proton elastic scattering spectra using 3.0 MeV H on SiO2 , S(C) and S(N).
5. Summary
Fig. 6. Ion concentration depth-pro®les determined from the RBS measurements for the material chip. (a) Ga pro®les for S(Ga, C), S(Ga, Pb, C), S(Ga, Pb, Sn, C) and S(Ga, N); (b) Pro®les of Pb and Sn for S(Pb, Sn, N) in comparison with Pb pro®les for S(Pb, N).
emission when bombarding it with lower energy electrons. Fig. 7 shows PES spectra of S(C), S(N) and pristine SiO2 by using 3 MeV H beams at a scattering angle of 170°. The peak concentration of C for S(C) is 4.12 at%, calculated using the proton elastic scattering cross section enhancement factors given in Ref. [25]. As shown in Fig. 7, the spectrum of S(N) did not reveal any trace of N after sample annealing, and that can be owing to the N diusion and N escape-out in SiO2 during annealing. Our previous work shows that for proton elastic scattering with energy >2.5 MeV, the cross sections of light elements C, N and O are enhanced and therefore, the simultaneously pro®ling of N, C and O over a sucient depth range in a heavy material can be readily accomplished [25].
We have successfully developed a combinatorial ion synthesis and ion beam analysis technique to study ion-implanted material libraries, by which the eciency of sample synthesis and analysis is greatly increased. Panchromatic imaging of optical properties combining with element pro®ling for material chip (library) can be used for screening optical property. Samples co-implanted with C and Ga yield yellow±green CL emission owing to the interaction between Ga and C, which is corroborated by RBS measurements. Sn and Pb co-implantation induces emission shift to higher energy bands, with possible Sn nanocrystal formation. No evidence of GaN nanocrystal formation is observed by CL and PES measurments. Most important of all, our approach can be used directly to nondestructively and rapidly characterizing composition and the inhomogeneity of the as grown combinatorial ®lm libraries. Acknowledgements The authors thank Mr. J.M. Wang, Prof. M. Chen for ion implantation and Mr. J.Q. Cao for assistance in ion beam analysis. This work is supported by the Chinese National Science
88
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
Foundation under Contract No. 19875075 and K.C. Wong Education Foundation. References [1] B.A. Bunin, M.J. Plunkett, J.A. Ellman, Proc. Natl. Acad. Sci. USA 91 (1994) 4708. [2] X.-D. Xiang, X.D. Sun, G. Briceno, Y. Lou, K.-A. Wang, H. Chang, W.G. Wallace-Freedman, S.W. Chen, P.G. Schultz, Science 268 (1995) 1739. [3] T. Wei, X.D. Xiang, W.G. Wallace-Freedman, P.G. Schultz, Appl. Phys. Lett. 68 (1996) 3506. [4] G. Briceno, H. Chang, X. Sun, P.G. Schultz, X.D. Xiang, Science 268 (1995) 1739. [5] X.D. Sun, C. Gao, J. Wang, X.D. Xiang, Appl. Phys. Lett. 70 (1997) 3353. [6] J. Wang, Y. Yoo, C. Gao, I. Takeuchi, X. Sun, X.D. Xiang, P.G. Schultz, Science 279 (1998) 1712. [7] X.D. Sun, X.D. Xiang, Appl. Phys. Lett. 72 (1998) 525. [8] H. Chang, C. Gao, I. Takeuchi, Y. Yoo, J. Wang, P.G. Schultz, X.D. Xiang, R.P. Sharma, M. Downes, T. Venkatesan, Appl. Phys. Lett. 72 (1998) 2185. [9] A.P. Alivisatos, Science 271 (1996) 933. [10] F. Hache, D. Ricard, C. Flytzanis, U. Kreibig, Appl. Phys. A: Solids Surf. 47 (1988) 347.
[11] H.Z. Song, X.M. Bao, N.S. Li, J.Y. Zhang, J. Appl. Phys. 82 (1997) 4028. [12] M. Zacharias, J. Christen, J. B asing, D. Bimberg, J. NonCryst. Solids 198±200 (1996) 115. [13] L.S. Liao, Z.H. Xiong, X. Zhou, X.B. Liu, X.Y. Hou, Appl. Phys. Lett. 71 (1997) 2193. [14] J.A. Wolk, K.M. Yu, E.D. Bourret-Couchesne, E. Johnson, Appl. Phys. Lett. 70 (1997) 2268. [15] B.G. Yacobi, D.B. Holt, Cathodoluminescence Microscopy of Inorganic Solids, Plenum Press, New York, 1990. [16] B.J. Pierce, R.L. Hengehold, J. Appl. Phys. 47 (1976) 644. [17] M. Goldbergy, H.J. Fitting, A. Trukhin, J. Non-Cryst. Solids 220 (1997) 69. [18] Y. Morimoto, R.A. Weeks, A.V. Barnes, N.H. Tolk, R.A. Zuhr, J. Non-Cryst. Solids 203 (1996) 62. [19] D. R uter, S. Rolf, Bauhofer. Appl. Phys. Lett. 67 (1995) 149. [20] T. Xiao, A.H. Kitai, G. Liu, A. Nakua, J. Barbier, Appl. Phys. Lett. 22 (1998) 3356. [21] T. Ogino, M. Aoki, Jpn. J. Appl. Phys. 19 (1980) 2395. [22] H. Koyama, J. Appl. Phys. 51 (1980) 2228. [23] L.P. Ginzburg, A.A. Gordeev, A.P. Gorchakov, A.P. Jilinsky, J. Non-Cryst. Solids 183 (1995) 234. [24] A. Nakajima, T. Futatsugi, N. Horiguchi, N. Yokoyama, Appl. Phys. Lett. 71 (1997) 3652. [25] G. Yang, D. Zhu, H. Xu, H. Pan, Nucl. Instr. and Meth. B 61 (1991) 175.
www.elsevier.nl/locate/nimb
Cathodoluminescence and ion beam analysis of ion-implanted combinatorial materials libraries on thermally grown SiO2 Chang-Ming Chen *, H.C. Pan, D.Z. Zhu, J. Hu, M.Q. Li
1
Shanghai Institute of Nuclear Research, Chinese Academy of Sciences, P.O. Box 800-204, Shanghai 201800, People's Republic of China Received 1 December 1998; received in revised form 20 April 1999
Abstract A method combining ion implantation and physical masking technique has been used to generate material libraries of various ion-implanted samples. Ion species of C, Ga, N, Pb, Sn, Y have been sequentially implanted to an SiO2 ®lm grown on a silicon wafer through combinatorial masks and consequently a library of 64 (26 ) samples is generated by 6 masking combinations. This approach oers rapid synthesis of samples with potential new compounds formed in the matrix, which may have speci®c luminescent properties. The depth-resolved cathodoluminescence (CL) measurements revealed some speci®c optical property in the samples correlated with implanted ion distributions. A marker-based technique is developed for the convenient location of sample site in the analysis of Rutherford backscattering spectrometry (RBS) and proton elastic scattering (PES), intended to characterize rapidly the ion implanted ®lm libraries. These measurements demonstrate the power of nondestructively and rapidly characterizing composition and the inhomogeneity of the combinatorial ®lm libraries, which may determine their physical properties. Ó 1999 Elsevier Science B.V. All rights reserved.
1. Introduction The combinatorial approach, in which the synthesis and screening of libraries with a large member of dierent organic molecules are implemented, has been widely used for the drug discovery [1]. Recently, Xiang et al. [2] pioneered the application of this approach to the discovery of
* Corresponding author. Tel.: +86-21-5955-2394; fax: +8621-5955-3021; e-mail: [email protected] 1 Tel.: +86-21-5955-2394; fax: +86-21-5955-3021; e-mail: [email protected].
new solid-state inorganic materials with desired properties. Combining combinatorial physical/lithography masks with thin ®lm synthesis technique [2], integrated material chips with a large number of spatially de®ned dierent compounds are generated. Contact and non-contact detection techniques [3] were developed to screen the libraries for desirable physical properties, such as superconductivity, giant magnetoresistance, ecient luminescence and low loss dielectrics [4±8]. The combinatorial technique can be widely used in material science to increase the rate of new material discovery and optimization by several orders of magnitude.
0168-583X/99/$ - see front matter Ó 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 5 8 3 X ( 9 9 ) 0 0 4 4 9 - 8
82
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
During the last few years special interest has been devoted to create new nanocrystals of metals or compounds in SiO2 , which may exhibit speci®c optical emission properties and be interesting for optoelectronic application [9,10]. The ion implantation is attractive for this purpose, because it is compatible with planar microelectronic technology and generates nanocrystals with well-de®ned concentration at pre-calculated depths below the host material surface. Implantation of various individual elements such as Si [11], Ge [12] and C [13] has been reported, which yields dierent luminescent properties. Recently, a sequential ionimplantation of Ga and N has been used to create GaN nanocrystals in sapphire (Al2 O3 ) with speci®c optical property [14]. Our ®rst application of combinatorial approach in ion implantation is intended to increase greatly the eciency of sample synthesis and the possibility of novel nanocrystal formation, which may yield higher energy emissions meeting the needs of high density optical storage. Although the application of combinatorial approach could revolutionize material discovery ± much as it has in pharmaceutical applications [2], the in-depth study of the materials libraries is in some respects restricted by the diculty in phase determination and composition inhomogeneity examination of the ®lm libraries, which may determine their properties. In this paper, we report the application of cathodoluminescence (CL) technique to identify local bonding environments which contribute to the luminescent properties, and the ®rst application of Rutherford backscattering spectrometry (RBS) and proton elastic scattering (PES) method to determine the depthresolved ion distribution of the combinatorial ion implanted samples. 2. Experimental Boron (P-type) doped 0.5 mm thick single crystal silicon wafers (á1 0 0ñ orientation and resistivity of 5.0 X cm) were used for the implantation. SiO2 thin ®lm of 400 nm was formed by thermal oxidation of wafer in H2 + O2 atmosphere. Several elements, C, Ga, N, Pb, Sn and Y, were
selected for implantation considering the potential in forming nanocrystals of GaN, C, Sn, SiC or other unexpected recombination centers in the host-matrices. The energies for the implantation were 40±80 keV and each ion doses are 1±5 ´ 1016 /cm2 , depending on ion species. Implantation of various elements was accomplished at room temperature using a mass-separated beam through an electromagnetic isotope separator (EMIS-SINR-02). The ion beam was electrostatically scanned in a 50 cm2 area to homogenize the surface ion dose distribution. The ion beam was incident on the substrate through a set of physical masks, of which the combination of ion species is arranged in a similar way as described in Ref [2]. The generated material chip (library) was subsequently thermally treated at 800°C for 30 min in ¯owing Ar atmosphere and was further characterized by CL, RBS and PES analysis. Comparing with photoluminescence measurement, cathodoluminescence measurement can oer depth-resolved optical properties due to various penetration of electron beam with dierent accelerating voltage. This would also provide additional local information on the defects and impurities in the matrix [15,16]. The cathodoluminescence measurement system capable of producing beam current of 1±3 lA was employed to this study. The dispersed emission was detected with the HTV-R-446 photomultiplier and WDG05-II monochromator. RBS analysis was performed on the 4 MeV pelletron by using 2 MeV 4 He beams at a scattering angle of 170°. The energy resolution of the detector is about 16 keV and total dose of 4 He was 3 lC. The material chip holder for RBS and PES measurement is schemed in Fig. 1. A rectangular stainless steel substrate is axed with four quartz strips along each side, with each strip length equal to that of the substrate side. A rectangular frame of stainless steel is overlaid on the quartz and stuck by silver paste. On each side of the frame, eight 2.5 mm±spaced (depends on the sample spacing in the chip) holes of 1 mm diameter is produced to expose the quartz surface. The material chip is placed on the top of the stainless steel substrate with sample sites aligned to the holes located in the stainless steel frame. In the
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
83
spots is under ion bombardment to emit reddish± white light, which can be observed either by naked eye or through a telescope. Afterwards, ion beam analysis for the selected sample can be accomplished by moving sample location relative to the reference point. 3. Cathodoluminescences of materials libraries
Fig. 1. Schematic of material chip holder for RBS and PES analysis. The double arrow line represents that the holder is controllably moved along X and Y directions to let the selected sample site aligned with the analyzing beam.
RBS and PES measurement, a primary reference point is ®rst located by moving the chip holder along X and Y directions using a feedthrough manipulator, so that one of the exposed quartz
Fig. 2 shows a material chip (library) of 64 dierent samples generated by ion beam implantation and 6 masking combinations, imaged under natural lighting. The dierent colors on various sample sites may originate from speci®c light re¯ections and interference from dierent interfaces. We also conducted CL measurements of the material chip with various electron-impinging voltages. As the material chip is exposed to an electron beam irradiation of 5 kV, 1 lA, we can easily distinguish the implanted samples from the pristine substrate. Most samples are of blue
Fig. 2. An image of the ion implanted silicon-based material chip with a library of 64 samples prior to annealing under white lighting. The silicon wafer is 25 mm ´ 25 mm and each sample site is of 2 mm diameter. Indicated elements and braced numbers corresponds the implanted ion species and their implantation sequences, respectively. Dierent implantation conditions for various elements are described as: (a) C, 50 keV, 5 ´ 1016 /cm2 ; (b) Ga, 80 keV, 1 ´ 1016 /cm2 ; (c) N, 50 keV, 1.5 ´ 1016 /cm2 ; (d) Pb, 80 keV, 1.5 ´ 1016 /cm2 ; (e) Sn, 70 keV, 1 ´ 1016 /cm2 ; (f) Y, 65 keV, 1 ´ 1016 /cm2 .
84
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
emission, and dierences in color and intensity of emitting light from the samples can be observed by naked eye. Yellow±green color from C containing implanted samples and violet±blue from Sn containing implanted samples are observed under 2 keV electron bombardment, with some major CL peaks revealed in Table 1. As evident in Table 1, the blue (450 nm) luminescent band is observed in all samples, and this is commonly observed in pristine SiO2 and amorphous SiO2 or silica, which is thought to be ascribed to the oxygen vacancy or E0 center and the two-fold-correlated silicon center [17,18]. The 450 nm CL band is caused by electron bombardment and its intensity is increased by increasing electron accelerating voltage to 10 keV. Subsequent implantation of Ga and N into SiO2 is expected to form embedded GaN nanocrystals, which is a great important compound for optoelectronic devices. However, CL measurements for the sample S(Ga, N) did not show any evidences for the existence of GaN crystals in SiO2 and this may be due to N diusion through the ®lm to escape into the air or form Si3 Ox Ny , which can also be proved by their nearly no N existence determined by PES measurements. All samples implanted with C reveals some yellow±green color. The S(C) exhibits a weak additional emission peaked 555 nm. The C presence in the host is hopefully to induce the formation of C-cluster, silicon carbide or silicon oxycarbide nanoparticles, which is of great interest because of the shift of the luminescent emission towards
higher energies compared to Si nanocrystals in SiO2 with 680 nm peak emission, when emissions in the violet range are expected. Indeed, the C can react with O in SiO2 to form CO and/or CO2 , and it is assumed that the reduced Si can react with C to form silicon carbide or silicon oxycarbide when access C is remained in SiO2 . This similar emission band is also observed in silicon±carbon based ®lms prepared by plasma enhanced chemical deposition [19] and is attributed to amorphous Six C1ÿx formation. Co-implantation of C and Ga in SiO2 induce a strong emission peaked at 570 nm, even followed by Pb and Sn implantation. The Ga can react with SiO2 in a form as follows: 4Ga 3SiO2 3Si 2Ga2 O3 :
The Ga2 O3 is a phosphor usually crystallized in highly anisotropy crystal structure [20], forming interconnected tetrahedra or octahedra which surround a large ``tunnel''. It is believed this unique one-directional ``tunnels'' play a major role in contributing to the superior capacity of these phosphors to generate hot electrons to incite the luminescent centers. From the evident diusion of Ga to the C-peaked nearby region after annealing, one may deduce that C implantation would have induced the formation Gax Siy O3 Cx , which is attributed to bonding impetus between Ga and C in SiO2 rather than to the strain induced by implantation damage, as similar phenomena is not observed in N-peaked nearby region. The complex consisting of a gallium vacancy and a carbon atom
Table 1 Characteristic CL spectra for the selected samples under 2 keV electron bombardment Sample S(C) S(N) S(Ga, C) S(Ga, N) S(Ga, Pb, C) S(Ga, Pb, Sn, C) S(Pb, Sn, N) S(Pb, N) a
Dominant peak position (nm)a
415 (0.75) 405 (1) 435 (0.6)
1
450 450 450 445 450 450 450 450
The data in the bracket indicates normalized emission intensity relative to 450 nm.
555 (0.10) 570 (0.56) 567 (0.60) 565 (0.60)
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
may also play a role in the emission of 570 nm in samples co-implanted with C and Ga, as like that happens in C implanted GaN [21]. Table 1 also shows that Sn and Pb co-implantation causes emission shift to higher energy. Fig. 3 shows a comparison of CL spectra for S(Pb, Sn, N) and the prsistine substrate with various electron energies. The peak near 450 nm was observed in the spectra for both implanted sample and pristine SiO2 substrate under the irradiation of electron beam energy of 10 keV, with four times luminescence intensity for implanted sample to that of pristine SiO2 . The luminescence intensity for S(Pb, Sn, N) is enhanced by a factor of 3.3 and 8 with respect to that for the pristine substrate, corresponding impinging energies of 5 and 2 keV, respectively. Also shown in Fig. 3 are the depth-resolved CL spectra for S(Pb, Sn, N) measured with electron beam energies of 2, 5, and 10 keV, respectively. The band peak shifts to lower wavelength region as the beam energy decreases, with peaks at 450, 422 and 405 nm corresponding to beam energies of 10, 5 and 2 keV, respectively. The shifted curves shown in Fig. 3 display the density of electron± hole pairs which indicate speci®c luminescence correlated with an inhomogeneous depth±impurity distribution in the sample.
Fig. 3. Cathodoluminescence spectra of S(Pb, Sn, N) and unimplanted SiO2 at various electron energies. The spectra were measured at the room temperature.
85
We have calculated the energy loss during electron penetrating into SiO2 using a density of 2.25 g/cm3 for SiO2 , as shown in Fig. 4. The concentration of excess electron±hole pairs created in the material is directly proportional to the energy loss as electrons penetrated the solid. Comparing Fig. 3 with Fig. 4, it evidently results that implantation of Pb and Sn in SiO2 causes an increase of pair formation and higher energy CL. The CL peak position of S(Pb, Sn, N) at 10 keV is almost the same with that of SiO2 . This is because the SiO2 area near the Si/SiO2 interface is largely excited in the condition and the correspondingly produced spectrum band may dominate in this case. As the electron energy decreases, the region approaching the surface is largely excited and peak band shifts to higher energies, that is obviously owing to the eects of Pb- and Sn- related luminescent centers in the material, which can match the ion distributions measured by RBS Fig. 6(b). The CL peak shift is often thought to relate to new phase formation or defect-impurity complex [16,22]. Considering the two-fold-coordinated silicon O±Si±O or O±Ge±O as the origin of the blue luminescence in Si1ÿx Gex O2 glass [23], one may expect that O±Pb±O and O±Sn±O may contribute to the peak shifts. On the other hand, one may also think that nanocrystal formation may take an effect, as Sn nanocrystals were observed in Sn implanted SiO2 followed by thermal treatment [24].
Fig. 4. Calculated electron energy loss vs depth during impinging into the SiO2 .
86
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
4. Ion depth-resolved pro®ling by RBS and PES Ion analysis is also used to quantitatively determination of constitution and help elucidating the interaction among implanted ions which may yield correlated speci®c luminescence. The spectra obtained by RBS measurements for the 800°C annealing material chip are partially shown in Fig. 5. Fig. 5(a) shows the spectra of S(Ga, N) and S(Ga, C). The Ga spectrum of S(Ga, N) is peaked at 36 nm from the surface, with no observable peak shift with respect to that of the sample prior to annealing (the spectrum is not shown here). Concerning the N depth range which is estimated to be peaked at about 130 nm from the surface by TRIM code calculation, we can conclude that no GaN formed in the region of N peak, although GaN nanocrystals formation is reported in Ga and N implanted Al2 O3 after annealing [14]. Two peaks of Ga spectrum for S(Ga, C) are revealed in Fig. 5(a), which indicates that the C existence induces the redistribution of Ga in SiO2 after annealing. The pro®le for C distribution is peaked at about 180 nm from the surface by TRIM code calculation, which matches the location of the lower energy peak of Ga spectrum (172 nm from the surface). It is expected that C induced Ga-related compounds may be formed in the SiO2 and that is worth further study. Subsequent implantation of Pb or both Pb and Sn into the samples with previous C implantation did not alter the Ga spectrum peak location, as partially shown in Fig. 5(b). We therefore conclude that the Ga redistribution is aected only by C implantation. From Fig. 5(c), it results that the interaction between Pb and Sn is little comparing the spectra of S(Pb, N) and S(Pb, Sn, N). The previous discussions are also con®rmed by the ion concentration depth-pro®les calculated with RUMP program, which are shown in Fig. 6, corresponding to Fig. 5. The Ga pro®le of S(Ga, C) is peaked at 36 and 172 nm from the surface, corresponding atom fractions of 1.5% and 0.56%, respectively (Fig. 6(a)). The pro®ling curves of S(Ga, Pb, C) and S(Ga, Pb, Sn, C) also each shows two peaks of Ga concentration, located at nearly the same position with respect to that of S(Ga, C). From Fig. 6(b), it reveals that Sn diusion
Fig. 5. 2.0 MeV 4 He RBS spectra from material chip after annealing at 800°C for 30 min in Ar gas. (a) S(Ga, C) and S(Ga, N); (b) S(Ga, Pb, C) and S(Ga, Pb, Sn, C); (c) S(Pb, Sn, N) and S(Pb, N).
approaching the surface for S(Pb, Sn, N) is evident, which is a cause of higher energy luminescent
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
87
Fig. 7. Proton elastic scattering spectra using 3.0 MeV H on SiO2 , S(C) and S(N).
5. Summary
Fig. 6. Ion concentration depth-pro®les determined from the RBS measurements for the material chip. (a) Ga pro®les for S(Ga, C), S(Ga, Pb, C), S(Ga, Pb, Sn, C) and S(Ga, N); (b) Pro®les of Pb and Sn for S(Pb, Sn, N) in comparison with Pb pro®les for S(Pb, N).
emission when bombarding it with lower energy electrons. Fig. 7 shows PES spectra of S(C), S(N) and pristine SiO2 by using 3 MeV H beams at a scattering angle of 170°. The peak concentration of C for S(C) is 4.12 at%, calculated using the proton elastic scattering cross section enhancement factors given in Ref. [25]. As shown in Fig. 7, the spectrum of S(N) did not reveal any trace of N after sample annealing, and that can be owing to the N diusion and N escape-out in SiO2 during annealing. Our previous work shows that for proton elastic scattering with energy >2.5 MeV, the cross sections of light elements C, N and O are enhanced and therefore, the simultaneously pro®ling of N, C and O over a sucient depth range in a heavy material can be readily accomplished [25].
We have successfully developed a combinatorial ion synthesis and ion beam analysis technique to study ion-implanted material libraries, by which the eciency of sample synthesis and analysis is greatly increased. Panchromatic imaging of optical properties combining with element pro®ling for material chip (library) can be used for screening optical property. Samples co-implanted with C and Ga yield yellow±green CL emission owing to the interaction between Ga and C, which is corroborated by RBS measurements. Sn and Pb co-implantation induces emission shift to higher energy bands, with possible Sn nanocrystal formation. No evidence of GaN nanocrystal formation is observed by CL and PES measurments. Most important of all, our approach can be used directly to nondestructively and rapidly characterizing composition and the inhomogeneity of the as grown combinatorial ®lm libraries. Acknowledgements The authors thank Mr. J.M. Wang, Prof. M. Chen for ion implantation and Mr. J.Q. Cao for assistance in ion beam analysis. This work is supported by the Chinese National Science
88
C.-M. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 159 (1999) 81±88
Foundation under Contract No. 19875075 and K.C. Wong Education Foundation. References [1] B.A. Bunin, M.J. Plunkett, J.A. Ellman, Proc. Natl. Acad. Sci. USA 91 (1994) 4708. [2] X.-D. Xiang, X.D. Sun, G. Briceno, Y. Lou, K.-A. Wang, H. Chang, W.G. Wallace-Freedman, S.W. Chen, P.G. Schultz, Science 268 (1995) 1739. [3] T. Wei, X.D. Xiang, W.G. Wallace-Freedman, P.G. Schultz, Appl. Phys. Lett. 68 (1996) 3506. [4] G. Briceno, H. Chang, X. Sun, P.G. Schultz, X.D. Xiang, Science 268 (1995) 1739. [5] X.D. Sun, C. Gao, J. Wang, X.D. Xiang, Appl. Phys. Lett. 70 (1997) 3353. [6] J. Wang, Y. Yoo, C. Gao, I. Takeuchi, X. Sun, X.D. Xiang, P.G. Schultz, Science 279 (1998) 1712. [7] X.D. Sun, X.D. Xiang, Appl. Phys. Lett. 72 (1998) 525. [8] H. Chang, C. Gao, I. Takeuchi, Y. Yoo, J. Wang, P.G. Schultz, X.D. Xiang, R.P. Sharma, M. Downes, T. Venkatesan, Appl. Phys. Lett. 72 (1998) 2185. [9] A.P. Alivisatos, Science 271 (1996) 933. [10] F. Hache, D. Ricard, C. Flytzanis, U. Kreibig, Appl. Phys. A: Solids Surf. 47 (1988) 347.
[11] H.Z. Song, X.M. Bao, N.S. Li, J.Y. Zhang, J. Appl. Phys. 82 (1997) 4028. [12] M. Zacharias, J. Christen, J. B asing, D. Bimberg, J. NonCryst. Solids 198±200 (1996) 115. [13] L.S. Liao, Z.H. Xiong, X. Zhou, X.B. Liu, X.Y. Hou, Appl. Phys. Lett. 71 (1997) 2193. [14] J.A. Wolk, K.M. Yu, E.D. Bourret-Couchesne, E. Johnson, Appl. Phys. Lett. 70 (1997) 2268. [15] B.G. Yacobi, D.B. Holt, Cathodoluminescence Microscopy of Inorganic Solids, Plenum Press, New York, 1990. [16] B.J. Pierce, R.L. Hengehold, J. Appl. Phys. 47 (1976) 644. [17] M. Goldbergy, H.J. Fitting, A. Trukhin, J. Non-Cryst. Solids 220 (1997) 69. [18] Y. Morimoto, R.A. Weeks, A.V. Barnes, N.H. Tolk, R.A. Zuhr, J. Non-Cryst. Solids 203 (1996) 62. [19] D. R uter, S. Rolf, Bauhofer. Appl. Phys. Lett. 67 (1995) 149. [20] T. Xiao, A.H. Kitai, G. Liu, A. Nakua, J. Barbier, Appl. Phys. Lett. 22 (1998) 3356. [21] T. Ogino, M. Aoki, Jpn. J. Appl. Phys. 19 (1980) 2395. [22] H. Koyama, J. Appl. Phys. 51 (1980) 2228. [23] L.P. Ginzburg, A.A. Gordeev, A.P. Gorchakov, A.P. Jilinsky, J. Non-Cryst. Solids 183 (1995) 234. [24] A. Nakajima, T. Futatsugi, N. Horiguchi, N. Yokoyama, Appl. Phys. Lett. 71 (1997) 3652. [25] G. Yang, D. Zhu, H. Xu, H. Pan, Nucl. Instr. and Meth. B 61 (1991) 175.