dose-rate dependence of negative copper ion implantation into silica glasses and effects on colloid...

5
Dose-rate dependence of negative copper ion implantation into silica glasses and eects on colloid formation N. Kishimoto a, * , V.T. Gritsyna b , Y. Takeda a , C.G. Lee a , T. Saito a a National Research Institute for Metals, 1-2-1 Sengen, Tsukuba, lbaraki 305-0047, Japan b Kharkov State University, Kharkov 310077, Ukraine Abstract Although high-current implantation of negative ions, which alleviates charging, is potentially capable of ecient sur- face modification of insulators, the beam–solid interactions and material kinetics have not been well understood. We have studied the implantation of 60 keV negative Cu ions into silica glasses, at various current densities up to 260 lA/cm 2 , at a fixed fluence of 3.0 · 10 16 ions/cm 2 . A masking method was developed to dissipate the beam loading, and multi- or single-hole masks were alternatively applied to change the boundary condition. Optical properties were determined in the range of 1.4–6.2 eV and the microstructures were examined by TEM. A significant dose-rate eect occurred for optical absorption and reflection. There was an optimum dose-rate to obtain a certain absorbance. Chan- ges in the mask shape resulted in dierent dose-rate dependence, which indicated that beam–solid interactions near the surface aected the implantation. At high dose rates, spherical Cu colloids with a pronounced bimodal distribution were produced. The surface morphology also changed and suggested significant atomic migration. Ó 1998 Elsevier Science B.V. All rights reserved. PACS: 42.70.C; 61.82.Ms Keywords: High-current implantation; Negative ion; Colloid; Dose-rate dependence; Silica glass 1. Introduction Negative-ion implantation is a promising tool for surface modification of insulators, because the incident negative charges alleviate positive sur- face charging via secondary electron emission. If one used positive ions, the specimen would charge up to the acceleration voltage or the breakdown voltage [1]. The charge buildup may cause inaccu- rate implantation or even damage on the surface. In particular, low-energy implantation around 1– 10 2 keV should be significantly aected by the sur- face charging. Recently, formation of metal col- loids in insulators has attracted our attention, since the metal colloids showed optical nonlinear- ity due to the surface plasmon resonance [2] and the fast response [3]. The metal precipitation re- quires the larger dose level and high-current im- plantation becomes favorable. Up to now, we have applied a relatively low energy (60 keV) to Nuclear Instruments and Methods in Physics Research B 141 (1998) 299–303 * Corresponding author. Tel.: +81-298-59-5059; fax: +81- 298-59-5010; e-mail: [email protected]. 0168-583X/98/$19.00 Ó 1998 Elsevier Science B.V. All rights reserved. PII S0168-583X(98)00180-3

Upload: n-kishimoto

Post on 16-Sep-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Dose-rate dependence of negative copper ion implantation intosilica glasses and e�ects on colloid formation

N. Kishimoto a,*, V.T. Gritsyna b, Y. Takeda a, C.G. Lee a, T. Saito a

a National Research Institute for Metals, 1-2-1 Sengen, Tsukuba, lbaraki 305-0047, Japanb Kharkov State University, Kharkov 310077, Ukraine

Abstract

Although high-current implantation of negative ions, which alleviates charging, is potentially capable of e�cient sur-

face modi®cation of insulators, the beam±solid interactions and material kinetics have not been well understood. We

have studied the implantation of 60 keV negative Cu ions into silica glasses, at various current densities up to 260

lA/cm2, at a ®xed ¯uence of 3.0 ´ 1016 ions/cm2. A masking method was developed to dissipate the beam loading,

and multi- or single-hole masks were alternatively applied to change the boundary condition. Optical properties were

determined in the range of 1.4±6.2 eV and the microstructures were examined by TEM. A signi®cant dose-rate e�ect

occurred for optical absorption and re¯ection. There was an optimum dose-rate to obtain a certain absorbance. Chan-

ges in the mask shape resulted in di�erent dose-rate dependence, which indicated that beam±solid interactions near the

surface a�ected the implantation. At high dose rates, spherical Cu colloids with a pronounced bimodal distribution

were produced. The surface morphology also changed and suggested signi®cant atomic migration. Ó 1998 Elsevier

Science B.V. All rights reserved.

PACS: 42.70.C; 61.82.Ms

Keywords: High-current implantation; Negative ion; Colloid; Dose-rate dependence; Silica glass

1. Introduction

Negative-ion implantation is a promising toolfor surface modi®cation of insulators, becausethe incident negative charges alleviate positive sur-face charging via secondary electron emission. Ifone used positive ions, the specimen would chargeup to the acceleration voltage or the breakdown

voltage [1]. The charge buildup may cause inaccu-rate implantation or even damage on the surface.In particular, low-energy implantation around 1±102 keV should be signi®cantly a�ected by the sur-face charging. Recently, formation of metal col-loids in insulators has attracted our attention,since the metal colloids showed optical nonlinear-ity due to the surface plasmon resonance [2] andthe fast response [3]. The metal precipitation re-quires the larger dose level and high-current im-plantation becomes favorable. Up to now, wehave applied a relatively low energy (60 keV) to

Nuclear Instruments and Methods in Physics Research B 141 (1998) 299±303

* Corresponding author. Tel.: +81-298-59-5059; fax: +81-

298-59-5010; e-mail: [email protected].

0168-583X/98/$19.00 Ó 1998 Elsevier Science B.V. All rights reserved.

PII S 0 1 6 8 - 5 8 3 X ( 9 8 ) 0 0 1 8 0 - 3

fabricate metal colloids embedded within a shal-low layer. Consequently, high-current implanta-tion using negative ions becomes of necessity.

However, the high-current technology has notyet been established and the material kinetics havenot been well understood, especially for negativeions. As a typical system of metal colloids in insu-lators, we have focused on Cu-implantation intosilica glasses. Although colloid formation process-es, at less than a few lA/cm2, have been successful-ly explained by quasi-equilibrium thermodynamics[4], the material kinetics at higher dose rates arestill open to question. Previously, we reported [5]that dose-rate dependent variations take place incolloid formation and resultant optical properties.The present purpose is to further investigate theprocesses focusing on the beam±solid interactions.

2. Experimental procedures

Negative Cu ions (60 keV) were implanted intosilica-glass substrates. The Cu current density ran-ged up to 260 lA/cm2, with a ®xed ¯uence of3.0 ´ 1016 ion/cm2. Details of negative-ion tech-niques were described elsewhere [6]. The depthpro®le and sputtering yield of 60 keV Cu were es-timated with the TRIM code [7]. The projectilerange and straggling width are 45 and �15 nm, res-pectively.

Substrates used were silica glasses of 15 mmdiam. ´ 0.5 mm thick and contained 820 ppmOHÿ. Metal (Cu) masks were mounted on thespecimen surface to dissipate the beam load. Eithera 2-mm diam. (multi-holes; 2.2-mm diam. ´37) ora 12-mm diam. mask covered the surface and wasattached to the specimen stage (earth potential).The specimen temperature was monitored by athermocouple and an infrared pyrometer.

Optical absorption was measured in a photonrange from 1.4 to 6.2 eV using a dual beam spec-trometer. Re¯ectivity was also measured calibrat-ing to an A1 mirror. Cross-sectional electronmicroscopy (TEM) was conducted to study micro-structures of the Cu implanted region [6]. Surfacemorphology of the implanted substrates was ob-served by atomic force microscopy (AFM) in thetapping mode.

3. Experimental results and discussion

High-current implantation has been performedwithout any macroscopic damage, by virtue ofnegative ions and the masks. The surface tempera-ture was maintained below 500 K. Fig. 1 shows aTEM image of specimen, Cu-implanted at a doserate 260 lA/cm2 (tilted from the edge-on). Almostperfect spheres of Cu colloids formed within a nar-row band near the surface. The diameter of thespheres ranges from 15 to 20 mm and they aremostly single crystals. Some of the spheres includeslightly tilted twins. Crystallographic orientationof the colloids is isotropic. On the other hand, ®neblack dots spread in the deeper region. Conse-quently, the whole size-distribution becomes bi-modal [8]. The total implanted zone at 260 lA/cm2 ranges over 130 nm from the surface and is1.5 times as much as that predicted by the TRIMcode. The wider distribution indicates di�usion en-hancement during the implantation process. Thebimodal distribution and enhanced di�usion be-come less dominant, with decreasing dose rate.

First, we applied the 2 mm diam. mask and ob-served an unusual dependence of optical absorp-tion on dose rate [5]. This dependence wasattributed to a variation in the number of retainedCu atoms. To change the boundary conditions, wenext applied the mask with a larger hole (12 mmdiam.) where the charge and heat dissipation be-come less e�ective and the electrostatic potential

Fig. 1. TEM image of a specimen, Cu-implanted at 260 lA/cm2

(2-mm diam.-mask), tilted from the edge-on direction. A Cr-

marker ®lm covers the beam-incident surface.

300 N. Kishimoto et al. / Nucl. Instr. and Meth. in Phys. Res. B 141 (1998) 299±303

may increase above the surface. Fig. 2 shows theoptical absorbance of Cu-implanted specimens(12-mm diam. mask), with various dose rates, atthe total ¯uence of 3.0 ´ 1016 ion/cm2. The spectraconsist of a plasmon peak at 2.2 eV [9] and a broadband in the UV region. The former is superim-posed with a wide band (2.7 eV) which was identi-®ed as indicating the presence of nanocrystals ofCu [10]. The latter UV band may well include pri-marily E0 centers (5.8 eV), nonbridging oxygenholes (4.8 eV) [11], oxygen-de®cient centers (4.95,5.15 eV) and/or peroxy radicals (5 eV) [12]. Here,we simply regard the spectrum as the plasmon at2.2 eV and the E0-band in the UV region.

Increasing the dose-rate shifts the absorbancecurve. Fig. 3 shows the dose-rate dependence ofabsorbance at the plasmon energy. The dose-rate

dependence, for the 2 mm diam. mask, gave atwo-humped variation, both for absorbance andre¯ectivity [5]. In contrast to the 2 mm diam.mask, the 12 mm diam. mask gives an earlier dropin the absorbance, as if the second maximum van-ished. The parallel shift of the absorption spectraagain suggests a change in retained Cu concentra-tion. Although the wider area of implantation istechnically preferred, the allowable dose rate islower. It is clearly shown by this experiment thatinteractions near the surface a�ects the inwardmass transport. However, we cannot yet identifywhich is more important, surface sputtering whichincreases the beam load or electrostatic blockingwhich decreases the charge dissipation, since thewider mask changes both of them. We point outat least, the importance of surface sputtering oratomic rearrangement, as is hereafter discussed.

The dose-rate dependence of the plasmon peakand the E0-band given in Fig. 4. Each peak inten-sity was obtained by subtracting the linear back-ground from the spectrum. The background wastaken at a position between the plasmon peakand the E0-band. As shown in Fig. 4, the variationof E0 centers is monotonous as compared to that ofplasmons. Although the plasmon intensity de-creases at 30 lA/cm2, the E0 intensity keeps a cer-tain level. This deviation indicates that the ionbombardment retains the E0 defects, even if a largepart of the Cu atoms was lost. It is suggested fromthis result that the Cu atoms escape from the sur-face after being implanted, not being repelled fromthe surface.

Fig. 2. Absorbance of Cu-implanted specimens, in the case of

12-mm diam. mask, with various dose rates at the total dose

of 3.0 ´ 1016 ion/cm2.

Fig. 3. Dose-rate dependence of absorbance at the plasmon en-

ergy, for the 2 mm diam. and the 12 mm diam. mask.

Fig. 4. Correlation between the plasmon peak and the E0-band,

as a function of the dose rate. Each peak intensity is obtained,

subtracting the linear B.G. from the spectrum.

N. Kishimoto et al. / Nucl. Instr. and Meth. in Phys. Res. B 141 (1998) 299±303 301

The surface morphology of the Cu implantedspecimens was observed by AFM. A surface imageof a specimen implanted at 260 lA/cm2 is given inFig. 5. After the high-current implantation, thesurface morphology changes to a rough textureand the blister-like humps have a diameter of100±150 nm. The surface roughness tends to de-crease with decreasing the dose rate, mainly dueto a decrease in the diameter. The change in mor-phology implies that signi®cant atomic migrationoccurs over the specimen surface. At the moment,either sputtering or surface di�usion/agglomera-tion is possible for the candidate mechanism. Thesputtering yield for 60 keV Cu bombarding SiO2

can be estimated with the TRIM code. The sput-tered atoms, estimated for the present dose, corre-sponds to a thickness of about 10 nm [6]. Thecharacteristic size of the blistered structures ismuch larger than the value predicted for the ordi-nary sputtering process. It is therefore concludedthat enhanced atomic migration signi®cantly con-tributes to the surface reconstruction, as well as

the sputtering process. Changing to the widermask tends to promote the surface roughness,which is consistent with the more severe condition.

As for the mechanism to form perfect spheres athigh dose rates, the experimental results indicatethe major roles of the interfacial surface tensionand the enhanced di�usion. In the high current im-plantation, the enhanced di�usion and surfaceatomic motion become dominant not only in thecolloid formation but also in the surface recon-struction. Even if the surface charge problems weresolved by negative ions, understandings of thosekinetics are requisite to utilize the technology.

4. Conclusions

By the aid of metal masks, high-current nega-tive Cu implantation was demonstrated to occurwithout detectable macroscopic damage. Thelow-energy implantation at 60 keV producedspherical colloids in a narrow layer near the sur-

Fig. 5. Surface morphology of a specimen, Cu-implanted at 260 lA/cm2 to the total dose of 3.0 ´ 1016 ion/cm2.

302 N. Kishimoto et al. / Nucl. Instr. and Meth. in Phys. Res. B 141 (1998) 299±303

face. At high dose rates, colloid formation depend-ed on the dose rate, accompanied by coarseningand bimodal distribution, even at the low energy.Absorbance and re¯ectivity also depended on thedose rate, resulting mainly from a variation ofCu atoms retained. Change to the wider mask re-sulted in an earlier drop of absorbance, in thedose-rate dependence. Surface morphology chan-ged to a blister-like texture and increased theroughness, at high dose rates. Thus, beam±surfaceinteractions played an important role in the masstransport. Enhanced di�usion and atomic migra-tion become dominant both inside the solid andon the surface.

Acknowledgements

The authors are grateful to Prof. Y. Mori atINS, Tokyo University for originally developingthe ion source and for valuable advice.

References

[1] J. Ishikawa, H. Tsuji, Y. Toyota, Y. Gotoh, K. Matsuda,

T. Tanjo, S. Sasaki, J. Nucl. Instr. and Meth. 96 (1995) 7.

[2] F. Hache, D. Richard, C. Flyzanis, U. Kreibig, Appl.

Phys. A 47 (1988) 347.

[3] T. Tokizaki, A. Nakamura, S. Kaneko, K. Uchida, S.

Omi, H. Tanji, Y. Asahara, Appl. Phys. Lett. 65 (1994)

941.

[4] H. Hosono, H. Fukushima, Y. Abe, R.A. Weeks, R.A.

Zuhr, J. Non-Cryst. Solids 143 (1992) 57.

[5] N. Kishimoto, V.T. Gritsyna, K. Kono, H. Amekura, T.

Saito, Nucl. Instr. and Meth. B 127/128 (1997) 579.

[6] N. Kishimoto, V.T. Gritsyna, K. Kono, H. Amekura, T.

Saito, Mater. Res. Soc. Sump. Proc. 438 (1997) 435.

[7] J.F. Ziegler, J.P. Biersack, U. Littmark, The Stopping and

Range of Ions in Solids, ch. 8 Pergamon Press, New York,

1985.

[8] N.D. Skelland, P.D. Townsend, J. Non-Cryst. Solids 188

(1995) 243.

[9] R. Ruppin, J. Appl. Phys. 59 (1986) 1355.

[10] H. Nishikawa, J. Appl. Phys. 80 (1996) 3513.

[11] R. Boscaino, Nucl. Instr. and Meth. B 116 (1996) 373.

[12] W. Marimoto, J. Non-Cryst. Solids 196 (1996) 106.

N. Kishimoto et al. / Nucl. Instr. and Meth. in Phys. Res. B 141 (1998) 299±303 303