Nuclear Instruments and Methods in Physics Research B 237 (2005) 155–159

www.elsevier.com/locate/nimb

The effects of energy non-monochromaticityof 11B ion beams on 11B diffusion

John Chen a,*, Lin Shao b, Tony Lin a, Jiarui Liu c, Wei-Kan Chu c

a Advanced Ion Beam Technology, Inc., 81 Daggett Drive, San Jose, CA 94086, USAb Los Alamos National Laboratory, Los Alamos, NM 87545, USA

c Department of Physics, Texas Center for Superconductivity and Advanced Materials, University of Houston, Houston, TX 77204, USA

Available online 5 July 2005

Abstract

We have shown that energy contamination introduced by ion beam deceleration technology that is used to increasethe beam currents available for low energy boron implants, can affect fabricated junctions adversely. A 4 keV 11B beamis extracted and retarded by a potential of �3.5 keV for 0.5 keV 11B implantation, or by a potential of �3.8 keV for0.2 keV 11B implantation. Intentional beam contamination was introduced by turning off the retarding potential toallow the 4 keV 11B ions to irradiate Si wafers directly. The percentage of contamination, at levels of 0.1%, 0.2%and 0.3% was introduced. Rapid thermal annealing of all the implanted samples was performed under N2 ambientat 1050 �C for 1 s. The dopant tail profiles themselves are not significant if the contamination levels are low. However,the much higher damage level coming from high energy contamination increases the transient enhanced diffusion of 11Bmore than proportionately, resulting in considerable boron diffusion. Energy contamination at a level of 0.1% canextend the profile of 0.5 keV 11B implants 10 nm deeper after a 1050 �C spike annealing. The study shows a highlymonoenergetic beam with energy contamination less than 0.1% is required for sub-micron devices.� 2005 Elsevier B.V. All rights reserved.

PACS: 66.30.Jt; 61.72.Ss; 61.72.Ji

Keywords: Shallow junction; Boron diffusion; Energy contamination

0168-583X/$ - see front matter � 2005 Elsevier B.V. All rights reservdoi:10.1016/j.nimb.2005.04.091

* Corresponding author. Tel.: +1 408 240 3266; fax: +1 408434 0490.

E-mail address: jchen@aibt-inc.com (J. Chen).

1. Introduction

The fabrication of complimentary metal oxidesemiconductor (CMOS) devices beyond the90 nm node demands a continued reduction of dif-fusion lengths of dopants in Si [1]. One significant

ed.

156 J. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 237 (2005) 155–159

technological barrier has to be overcome is the lowbeam current in ultra-low energy implantation dueto space charge effects. One method of achievingadequate boron beam current is the use of post-mass analysis ion beam deceleration technology,in which a reverse bias is applied to deceleratethe ions down to the final desired energy prior toimplanting targets or wafers. However, charge ex-changes and neutral transport before and duringdeceleration can give rise to energy-contaminatedbeams. Control of energy contamination differsamong various implanters. Even with ultra-low en-ergy contamination level while as-implanted dop-ant profiles almost have no visible differences, theextra radiation damage coming from high-energycontamination increases the transient enhanceddiffusion of B11 more than proportionately, result-ing in considerable junction spreading afterannealing by interstitial diffusion mechanism [2,3].

In the present study, the effects of energy non-monochromaticity of 11B ion beams on both 11Bdiffusion and electrical properties of the fabricateddevices are discussed. Energy contamination wasinternationally introduced by turning off theretarding potential to allow the 11B ions, with theenergy equal to the extraction voltages, to irradiatethe Si wafer directly. The study will provide anidea of how important the accuracy of implanta-tion energy control should be.

2. Experiments

Bare (100)-oriented n-type Si wafers, preamor-phized with 5 keV Ge ions (Rp � 12 nm), were im-planted with 0.2 keV 11B and 0.5 keV 11B,respectively, at a dosage of 5 · 1014/cm2. Boronimplantation was performed at AIBT�s high beamcurrent, ultra-low energy prototype ion implanter.The implanter employs special method to removeenergy-contaminated neutrals from decal beams.The ion implanter includes an ion source, beamextraction electrodes, a mass analyzer, a decelera-tion module, a plasma shower, a Faraday cup, abeam profiler, a process batch disk and a300 mm wafer transfer system. The beam trans-port optics was designed to deliver narrow and tallbeams with higher beam currents to wafers, in

both drift mode and decel mode, especially at en-ergy range from 5 keV to 100 eV. The decel mod-ule includes a set of decel electrodes and amagnet to prevent high-energy neutral particlesfrom reaching wafers in decel mode operations.

A 4 keV 11B beam is extracted and retarded bya potential of �3.5 keV for 0.5 keV 11B implanta-tion, or by a potential of 3.8 keV for 0.2 keV 11Bimplantation. Intentional beam contaminationwas introduced by turning off the retarding poten-tial to allow the 4 keV boron ions to irradiate theSi wafer directly. The percentage of contaminationis defined as the ratio of the 4 keV 11B dose to the0.5 keV or 0.2 keV 11B dose. Energy contamina-tion, at levels of 0.1%, 0.2% and 0.3% was intro-duced. Rapid thermal annealing (RTA) of all theimplanted samples was performed under N2 ambi-ent at 1050 �C for 1 s. The 11B atomic depth distri-bution profiles were obtained using secondary ionmass spectrometry (SIMS). An Atomika 4500SIMS tool was used with a 0.5 keV Oþ

2 primarybeam at 20� angle of incidence (from normal).The raster size was 220 lm. A laser was used tocharge neutralize using the OCE method (opticalcarrier enhancement). Standards were used toquantify the B concentrations. The crater depthswere measured using an Alpha-Step surface pro-filer. The depth scale was calibrated by assuminga uniform erosion rate.

3. Results and discussion

Fig. 1 shows calculated distribution of Si dis-placements produced by contaminated 11B implan-tation from 0.1% to 0.3%, which corresponds to5 · 1011, 1 · 1012 and 1.5 · 1012/cm2 4 keV 11Bbombardment, respectively. The displacementsper ion is calculated to be 58 for 4 keV 11B and 8for 0.5 keV 11B from transport of ion in matter(TRIM) code [4]. Even for low contamination le-vel, 4 keV 11B implants can produce considerableradiation damages. The arrow in Fig. 1 indicatesthe projected range of 5 keV Ge ions. One wouldexpect that, by increasing preamorphization en-ergy, the total displacements caused by contami-nated 11B implantation can be totally containedwithin the amorphous region. However, the benefit

0 10 20 30 40 50 601018

1019

1020 Si interstitials produced by 4 keV B with dose of

0.1% x 5x1014/cm2

0.2% 0.3%

Def

ect C

once

ntra

tion

(cm

-3)

Si Depth (nm)

Rp of 5 keV Ge

Fig. 1. TRIM simulation of Si displacements after 4 keV 11Bimplantations with a dosage from 0.1% to 0.3% of 5 · 1014/cm2.

0 20 40 60 80

1017

1018

1019

1020

20 40 60

1017

1018

1019

1020

10

b

0.2 keV B, 5x1014/cm2

Bor

on C

once

ntra

tion

(ato

ms/

cm3 )

Si Depth (nm)

after 1050°CRTA

0% 0.1% 0.2%

a0.5 keV B, 5x1014/cm2

after 1050°C RTA

0% 0.1% 0.2% 0.3%

Rp of 4 keV B

0

21

Fig. 2. SIMS profiles of 0.5 keV B11 implants (a) and 0.2 keV11B implants (b) with a dosage of 5 · 1014/cm2 after 1050 �C/1 sannealing. Effects of various percentages of energy contamina-tions on 11B diffusions are compared.

0.00 0.05 0.10 0.15 0.20 0.25 0.30

30

35

40

45

50

55

500 eV B 200 eV B

200 eV B

Energy Contamination (%)

500 eV B

Junc

tion

Dep

th X

j at

1x1

018/c

m3 (

nm)

Fig. 3. Junction depths measured at 1 · 1018/cm3 after anneal-ing at 1050 �C for 1 s for 0.2 keV and 0.5 keV 11B implants, as afunction of percentages of energy contaminations.

J. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 237 (2005) 155–159 157

of deep preamorphization is offset by the detri-mental increase in 11B TED and 11B segregation[5].

Fig. 2(a) shows the 11B atomic depth profiles ofsamples implanted with 0.5 keV 11B with energycontamination of 0.1%, 0.2% and 0.3%. As-im-planted profiles of energy-contaminated sampleshave visible enhanced tails with buried peaks ataround 17 nm deep, which corresponds to the pro-jected range of implanted 4 keV 11B. Effects of var-ious percentages of energy contamination onboron diffusion after RTA are surprisingly large.Energy contamination extends the junction depthsignificantly deeper than that obtained withmonoenergetic implants. For the 0.1% contami-nated sample, the junction depth measured at1 · 1018/cm3 is increased by around 10 nm. Junc-tion depth finally reaches 53 nm at 0.3% energycontamination, corresponding to a diffusion lengthof almost double that obtained from contamina-tion-free samples. Fig. 2(b) shows annealed 11Bprofile for 0.2 keV 11B implants. Energy contami-nation at 0.2% induces an additional profile shiftof 14 nm. Both Figs. 2(a) and (b) show anoma-lously enhanced boron diffusion with increasedpercentage of contamination. Fig. 3 summarizesthe junction shifts as a function of contaminationlevels. These results clearly show the importanceof beam energy monochromaticity on the ultra-shallow junction formations.

It has been shown that anomalous B11 diffusionpersisted even for sub-keV 11B implantation [6].

0 20 40 60

1017

1018

1019

1020

0.2 keV B, 5x1014

/cm2

Bor

on C

once

ntra

tion

(ato

ms/

cm3 )

Si Depth (nm)

after 1050°CRTA

0% 0.1% 0.2%

with MeV 0% 0.1% 0.2%

Fig. 4. SIMS profiles of 0.2 keV 11B implants with a dosage of5 · 1014/cm2. Effects of various percent of energy contamina-tion on boron diffusion after 1050 �C for 1 s, with or without1 MeV, 5 · 1015/cm2 Si ion implantation, are compared.

158 J. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 237 (2005) 155–159

One proposed mechanism is the so-called boride-enhanced diffusion (BED), in which self-intersti-tials injected from a silicon boride phase cause en-hanced 11B diffusion [6]. Another mechanism is thecoupled diffusion of 11B [7]. A simple explanationfor it is that the flux of 11B self-interstitial pairsfrom Si surface into the bulk drags Si interstitialsalong, creating a supersaturation of Si self-intersti-tials. However, both of these two mechanisms areoperative dominantly at high 11B concentrations,and should not be very sensitive to variation ofthe low 11B concentrations typical of the tail ofthe 11B profiles. The observed additional borondiffusion induced by energy contamination ismainly due to transient enhanced diffusion causedby additional implantation damage from 4 keV11B bombardment.

The energy contaminations can seriously detri-ment device performance. The well known short-channel effects, as the channel length decrease,the fraction of charge in the channel region con-trolled by the gate decrease, result in the increaseof threshold voltage. Si interstitial supersaturationresulted from the energy contamination increasethe lateral as well as the vertical diffusion dis-tances. The short-channel effects induced fromthe enhanced lateral diffusion need increases ofgate lengths to keep the same threshold voltages.Experimentally, it has been shown that the junc-tion shifts due to the energy contamination isaccompanied with the shifts in gate lengths, e.g.an increase of 9 nm in junction depth can shift gatelength by 15 nm [8]. Also, the extra damages mayincrease leakage currents. Studies in this aspectare underway.

We have used the technique of point defectengineering (PDE) to reduce boron diffusion, andthen to lower the requirements on energy mono-chromaticity of 11B ion beams. PDE that uses SiMeV Si ion implantation provides a unique meth-od to separate the spatial distribution of vacanciesand interstitials [9]. During ion bombardment, theforward momentum imparted to the lattice Sicauses the recoiled Si interstitial distributions tobe deeper than that of vacancies. Since spatiallyseparated Frenkel pairs recombine in nearby prox-imity, an excess vacancy rich region is formedclose to the surface and excess interstitials are left

in the deep range. Reduction of 11B TED and BEDwith co-implantation of MeV Si ions have been re-ported elsewhere [9]. In this study, a subset of thewafers additionally received a ‘‘PDE’’ MeV Si ionimplantation with a dosage of 5 · 1015/cm2 beforeRTA. Fig. 4 shows SIMS profiles of 0.2 keV 11Bimplants after 1050 �C RTA, with or withoutPDE. It shows that the Xj of 0.1% energy contam-inated sample is 5 nm deeper than that of contam-ination-free sample, while with PDE the incrementis reduced to around 1 nm only. The effects ofPDE on the Xj are manifested in the data depictedin Fig. 5. PDE significantly reduces B diffusion.Junction depth Xj of all contaminated 0.5 keV11B implants is systematically reduced by around10 nm if PDE was performed. However, PDE isunable to reduce the profile spreading of 0.2% en-ergy contaminated 0.2 keV 11B and 0.5 keV 11Bimplants to a depth close to the contamination-free sample. The study shows that even the mosteffective boron-diffusion-control method such asPDE cannot eliminate the effects of energy non-monochromaticity on 11B diffusion. Therefore, ahighly monoenergetic beam with energy contami-nation less than 0.1% is desired.

The rapid thermal processor AG 210T used inthe present studies has a typical ramp up rate of

0.00 0.05 0.10 0.15 0.20 0.25 0.30

30

35

40

45

50

55

200 eV B, with MeV

Energy Contamination (%)

with MeV

500 eV B

Junc

tion

Dep

th X

j at

1x1

018/c

m3 (

nm)

500 eV B, no MeV Si500 eV B, with MeV Si200 eV B, no MeV Si200 eV B, with MeV Si

Fig. 5. Junction depths measured at 1 · 1018/cm3 after anneal-ing at 1050 �C for 1 s for 0.2 keV and 0.5 keV 11B implants,with or without MeV Si co-implantation.

J. Chen et al. / Nucl. Instr. and Meth. in Phys. Res. B 237 (2005) 155–159 159

75–100 �C/s, which is much lower than a ramp uprate of around 400 �C/s of current state-of-the-artfurnaces. The low ramp up rate will increase thejunction depths. However, since all the sampleshave been annealed at the same time, the compar-ison among them is valid to shed light on the ef-fects of the energy contamination on B diffusion.As for the contamination-free sample, the ob-tained junction depth is deeper than that obtainedby boron layer deposition [9]. As we have dis-cussed above, the domination diffusion mechanismin the case of boron layer deposition is BED, whilein B implanted sample both BED and TED playroles. Furthermore, the step of preamorphizationis expected to reduce the amount of excessivevacancies created by PDE in the near surface re-

gion. Further optimizations on the implantationand annealing processes are demanded to get shal-lower junctions.

4. Conclusions

In summary, our studies have shown that en-ergy contamination during decelerated ion implan-tation can affect adversely the post-annealingdiffusion of 11B implants. In order to satisfy devicerequirements, highly monochromatic beams areneeded.

References

[1] The National Technology Roadmap for SemiconductorIndustry Association, 2002, Available from http://public.itrs.net.

[2] P.M. Fahey, P.B. Griffin, J.D. Plummer, Rev. Mod. Phys.61 (1989) 289.

[3] Lin Shao, John Chen, Jianming Zhang, D. Tang, SanjayPatel, Jiarui Liu, Xuemei Wang, Wei-Kan Chu, J. Appl.Phys. 96 (2004) 919.

[4] J.P. Biersack, L.G. Haggmark, Nucl. Instr. and Meth. B 174(1980) 257.

[5] M.I. Current, M.A. Foad, A.J. Murrell, E.J.H. Collart, G.de Cock, D. Jennings, J. Vac. Sci. Technol. B 18 (2000) 468.

[6] A. Agarwal, H.-J. Gossmann, D.J. Eaglesham, S.B. Herner,A.T. Fiory, T.E. Haynes, Appl. Phys. Lett. 74 (1999) 2331.

[7] A. Ural, S. Koh, P.B. Griffin, J.D. Plummer, Mater. Res.Soc. Symp. Proc. 610 (2001) B4.11.1.

[8] M. Kase, in: Varian Semiconductor Equipment AssociateTechnology Symposium 2004, San Francisco, July, 2004.

[9] Lin Shao, J. Liu, Y. Quark Chen, Wei-Kan Chu, Mater.Sci. Technol. Rev. 42 (3–4) (2003) 65.