effects of low dose silicon, carbon, and oxygen...

3516 J. Electrochem. Soc., Vol. 141, No. 12, December 1994 9 The Electrochemical Society, Inc.

REFERENCES i. S.P. Murarka, J. Steigerwald, and R. J. Gutmann, MRS

Bulletin, Vol. XVII, p. 46 (June 1993). 2. S.P. Murarka, in Tungsten and Other Advanced Metals

for VLS[ Applications in 1990, G. C. Smith and R. Blumenthal, Editors, p. 179, MRS, Pi t tsburgh (1991).

3. N. Stoloff, Rensselaer Polytechnic Institute, Private communication.

4. N. J. Brown, P. C. Baker, and R. T. Maney, Proc. SPIE, 306, 42 (1981).

5. Rodel Technical Release, Surfacetech Review, 1, No. 5, 3 (1988).

6. H. H. Uhlig, Corrosion and Corrosion Control, p. 28,

John Wiley & Sons, Inc., New York (1985). 7. F. B. Kaufman, D. B. Thompson, R. E. Broadie, M. A.

Jaso, W. L. Guthrie, D. J. Pearsons, and IV[. B. Small, This Journal, 138, 3460 (1991).

8. J.M. Steigerwald, S. P. Murarka, D. J. Duquette, and R. J. Gutmann, in Proceedings of the MRS 1994 Spring Meeting, MRS, Pit tsburgh, PA (1994).

9. M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, p. 384, NACE, Houston, TX (1974).

10. H.E. Johnson and J. Leja, This Journal, 112, 638 (1965). 11. Handbook of Chemistry and Physics, 71s~ ed., D. R.

Lide, Editor, p. 8-16-8.23, CRC Press, Boca Raton, FL (1990).

Effects of Low Dose Silicon, Carbon, and Oxygen Implantation Damage on Diffusion of Phosphorus in Silicon

Samir Chaudhry* and Mark E. Law** Department of Electrical Engineering, University of Florida, GainesvilIe, Florida 32611

ABSTRACT

As device dimensions shrink to submicron levels, good design of ultrashallow junctions has become increasingly important. It is in this context that the use of carbon/oxygen as a possible diffusion-suppressing agent for phosphorus has been suggested. To study this complex phenomenon, this experimental study looks at the effects of low dose silicon, carbon, and oxygen implantation damage on the diffusion of lightly doped phosphorus layers. The effects of a silicon and carbon coimplant on the diffusion of phosphorus are studied as part of a second experiment. Finally, lightly doped drain structure is annealed in the presence of a carbon implant. Carbon is the most effective diffusion-suppressing agent among the three species. Results from the second experiment suggest that carbon strongly affects the interstitial profile, and thereby the final phosphorus profile.

Recent studies have suggested the use of a carbon implant to suppress the diffusion of phosphorus3 '2 These looked at the formation of secondary defects formed at the original amorphous/crystal (a/c) interface and concluded that the density of dislocation loops decrease both in size and density with increasing carbon dose. In addition, criti- cal carbon doses which minimize the dopant diffusion were identified. Dislocation loop engineering by the use of a carbon coimplant for bipolar devices was reported recently, ~ as a way to improve device performance and yield.

Packan and Plummer observed substantially enhanced diffusion of low concentration boron in deep regions of silicon wafers implanted with low doses of silicon and showed its dependence on the implant dose and annealing temperature. 4 Their results showed that lower temperatures resulted in greater enhancements in profile shifts, implying that short time high temperature anneals were better for building devices which needed ultrashallow junctions.

Park and Law observed an enhanced diffusion of low concentration phosphorus in a silicon-damaged substrate. 5 This dopant profile redistribution was basically a transient short time process which occurred due to the interaction of dopant and point defects. 6 The effect was again more pronounced for lower anneal temperatures.

As the above-mentioned work indicates, extensive knowledge of transient diffusion of phosphorus due to ion- implantation damage in silicon during annealing is needed to design ultrashallow junctions in submicron devices. This transient diffusion is thought to be caused by the complex interaction of dopants, point defects, and extended defects whose concentrations are changing with both time and lo- cation during the thermal anneal cycle. 9-I~ Extensive work has already been done to investigate this phenomenon. ~'s However, the need exists to investigate innovative tech- niques to suppress this enhanced diffusion. It is in this con-

* Electrochemical Society Student Member. * * Electrochemical Society Active Member.

text that the use of certain types of damage species (e.g., carbon) to suppress the diffusion of dopants during the short thermal anneal has been suggested. 1'2 It is believed that this reduced net diffusivity is a strong function of the damage dose.

In contrast to earlier work, this study looks at the species, dose, and temperature dependence of the diffusion of phosphorus in silicon. The damage species under investigation are silicon, carbon, and oxygen, the later two of which are of special interest as regards their usefulness in forming ultrashallow junctions. Their relative effectiveness as interstitial traps is clearly brought out here. The dose range chosen was below the amorphization threshold for silicon and was lower than the one used in Ref. 1 and 2. Time and temperature effects are considered at two different temperatures (809 and 90O~ for two different times (15 and 30 rain). Direct comparisons for silicon-implant damage are made with Packan and Plummer, ~ as both boron and phosphorus are believed to diffuse mainly by an interstitial phenomenon. Also, results for silicon damage for the 1 x 10~4/cm 2 dose are compared with Park and Law's data? ,6

As part of a second experiment, phosphorus is made to diffuse in a silicon-damaged substrate both with and without a carbon coimplant. The dose, time, and temperature effects are investigated. A lightly doped drain (LDD) structure is then annealed in the presence of a carbon implant and the technological ramifications are studied.

Experimental Monte Carlo simulations in SUPREM-III 7'8 initially were

used to estimate the interstitial and vacancy profiles in silicon due to silicon implantation damage for three different doses (i • 1012, i • i0 ~3, and 1 • 1014/cm e) at 60 keV. These simulations indicated that the interstitials and va- cancies were produced in roughly equal numbers. Using these damage proliles as markers, equivalent doses and energies were iteratively obtained for carbon and oxygen us-

Downloaded 04 Apr 2011 to 128.227.135.101. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

J. Electrochem, Soc., Vol. 141, No. 12, December 1994 9 The Electrochemical Society, Inc. 3517 Table I. Equivalent implant conditions for the three species.

Silicon Carbon Oxygen Dose Dose Dose

Energy (cm -~) Energy (cm -2) Energy (cm ~)

60 keV le12 45 keV 1.3e12 40 keV 1.8e12 le13 1.3e13 1.8e13 le14 1.4e14 1.8e14

ing Monte Carlo simulations in SUPREM-III. Thus for each silicon dose (at 60 keV) a corresponding dose and energy (which produced the same damage as silicon) was obtained for carbon and silicon. The results of these simulations are summarized in Table I. Each row in the table corresponds to the same damage in the silicon substrate. This facilitated direct comparison of diffused profiles for the three implant species without any ambiguity regarding damage dependent diffusion effects. Figure i shows representative damage profiles for the three cases and clearly shows that the damage created by each species is similar for a set of doses and energies.

A 25 nm oxide was initially grown on <I00> oriented, n-type silicon wafers. All the wafers received a phosphorus implant at 60 keV with a dose of 1 x 10~/cm2. This set of implant conditions prevented any high concentration diffusion effects. To remove any point defects introduced by the implanted dopant and to activate the dopant, the wafers were annealed for i0 min at 900~ The walers were divided into three lots for each of the three implanted species. Each lot was then implanted at different doses and energies (Table I) with its corresponding species. In addition, undamaged substrate samples were left for annealing. These act as base lines for measuring net diffusion enhancements for the damaged samples. Each lot was then further subdivided and each dose and energy combination was annealed for two different times (15 and 30 min) and at two different temperatures (800 and 900~ The experiment is summarized in the form of a flow chart in Fig. 2. Phosphorus profiles were then extracted from the samples using spreading resistance profiling (SRP) measurements.

Data Analysis To compensate for minor fluetuations in the total phos-

phorus dose in each sample, the profiles were normalized to a base I • 10~S/em 2 dose. These profiles were then read into SUPREM-IV. The as-implanted profile was first diffused to the profiles with no damage implants (for all anneal conditions) and the diffusion time (t~) obtained for each anneal condition. For each profile, diffused in a damaged substrate (hereafter called the target profile), the net increase in profile movement ( ~ ( D . t)) was extracted by numeri-

60keY, I• p h o ~ implant lOmm, 900~

acttv2t~ot, anneal,

I lxl[It2-Ixl{Jl~/cmZ, I 1,3xl012~1 4xl01~/cm~, i'8x1012"1 gxl0l~/cm2, 6OteV SI danlage imp ] 45kcV C (St equiv.) imp. 40kcV O I SI equlv ) ~mp

[

I _

15~30. o~ aria.

t No damage ~plam~

Fig. 2. Flow chart of the first experiment.

cally diffusing the as-implanted profile to the target profile and obtaining the diffusion time (t2) as follows

4 ~ ( D . t) = ~ 5 ~ - ~ b . t~ SUPREM-IV was used o ~ extract the diffusion en-

hancement. The quantity "r 9 t), thus gave the net enhancement in the profile movements for eaeh case under study. The results brought out a clear damage species and dose dependence of the diffusion of phosphorus in silicon. Figure 3 shows a representative case; the profiles move to the right as the damage dose is increased. This profile movement is graphitieally illustrated as a function of dose in Fig. 4 to 7. The increase in the profile shift as a function of increasing dose, was seen consistently in almost all cases. The quantity on the x-ax is in Fig. 4-7 is the silicon equivalent damage dose, and not the actual damage dose for the case of carbon and oxygen (refer to Table I for the actual damage dose).

The results for silicon-damaged substrates concur with earlier experiments. 34 In all eases the profile shifts increase or remain the same as the dose is increased (Fig. 4). Paekan and Plummer 4 observed a qualitatively similar behavior for boron, though the profile shifts obtained by them are quan- ti tatively different. Results from Park and Law's experiment ~ are directly plotted in Fig. 4. The profile shifts are in good agreement, except for the 30 rain 900~ anneal. Lower temperature anneals result in greater enhaneements in profile shifts for all three damage doses. This is again in accor- dance with Ref. 4 and 5. The silicon-damaged samples exhibited the greatest enhancement in the profile shifts (Fig. 6 and 7). Thus (in the dose range under investigation), it is evident that with increasing damage dose the concen-

10 ~2 . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . ,

10 2o ~ I - - C a r b o n I 9 ~ ........... Oxygen

~ < ~ [ - - - S i l i c o n [

~; "-.-%. 3 ._ \ ' - . . .-'- 10,~ ", ~ . .

'-,2..

10~4 \ ~ ' : ' h \ 1

1012 ',,X ... . . . . . . . ,, 0 .00 0.10 0 ,20 0.30 0 .40 0.50

Depth in ~m

Fig. 1. Representative interstitial (damage} profiles generated for cordon, oxygen, and silicon using Monte Carlo simulations in SUPREM-III.

10 'a 9 9 ' . ...~, 9 . 9 , . 9 . , . 9 . , . . . , . . .

~ ~ r-- Asqmplan~ ~ " ~ . ~ \ I .......... le12 / c r n ~

1017 ~ ~ , . : , \ \ / . . . . 1 e 1 3 / c m 2 I

2 ',.. ' , \

10 's

10 is , , , , . . . . . . , , , , , , , , ~ ' - i - - ' 7 " - ' ; " - - : " 2 0 .0 0.1 0 .2 0 .3 0.4 0 .5

Depth in !.Lm

Fig. 3. Phosphorus profiles in the silicon-damaged substrate offer a 15 min/900~ anneal.



40.0

30.0

20.0

10,0

~-----O 1Smin.-800"C~ [ ] [ ] 30min.-O00~

J ~ - - ~ 15min..900~ [#- -- A 30min..900~

. . . . . . - ' / / 9 . ~ . f f ; / / I

. . . . . . . jlj/- i

10 ~2 10 ~ 10'* Damage Dose (/cm ~)

Fig. 4. Net enhancement of phosphorus dopant profiles for silicon- damaged substrates for various anneal conditions; the filled symbols

14 2 corresponding to the 1 • 10 /cm damage dose are Park and Law's data. 4

tration of interstitials increases. These interstitials aid in the diffusion of phosphorus which is known to diffuse mainly by an interstitial mechanism.

The effectiveness of carbon (when compared to the other two species) as a diffusion-suppression agent is clearly brought out in this experiment (Fig. 7a, b). In almost all cases, it results in shallower junctions vis-a-vis the other two damage species. As in the previous two cases, carbon- damaged samples exhibited an enhancement in the phosphorus profile shifts with increasing dose. For the 30 rain anneal at 900~ the carbon-damaged samples showed that increasing the dose beyond 1 • i013/cm 2 did not increase the net profile movement. Since this behavior was not observed for short time and lower temperature anneal, it is apparent that the damage dose becomes less of a factor for higher temperature/longer anneals.

An interesting result in the oxygen-damaged samples is clearly seen on examining Fig. 5. The profile motion is greater for a 15 min/900~ anneal than for a 30 min/900~ anneal for all three damage doses. This means that the profile has moved to the left after the initial 15 min of diffusion. The shape of the interstitial profile could be changing drastically with time in this particular case causing the interstitial concentration gradient to be larger in magni- tude when looking left in Fig. 3, as compared to when looking right. The diffusion of phosphorus not only depends on the concentration of the interstitials available, but also on the gradient of the interstitial concentration. Thus, in oxy-

4 0 . . . . , . . . . . . . . , . . . . . . . . ,

Io--o,s mi.-900"c I

30

J /

[] , i

2~ ' ' '1012 ' . . . . . . 1'0 TM ' . . . . . . 1'()" Silicon equi. damage dose (/cm 2)

Fig. 5. Oxygen uphill diffusion; total phosphorus dopant profile motions for oxygen-damaged substrates.

gen implantation a shorter anneal time did not reflect itself in a shallower junction.

Results for the oxygen-damaged samples also showed an enhancement in the net profile shifts with increasing dose in all cases (Fig. 6a, b). Almost all oxygen data points lie between carbon and silicon data points. Thus, the effectiveness of oxygen as a diffusion-suppressing agent for phosphorus lies below carbon but above silicon. This effect is more pronounced for longer anneals at higher temperatures (Fig. 6b) where oxygen is as effective as carbon. How- ever, at lower temperatures the net profile shifts for the oxygen-damaged samples are in close proximity to the silicon-damaged samples. This phenomenon is not clearly understood at this stage.

Discussion and Second Experiment The behavior of the diffusing phosphorus in silicon-dam-

aged substrates is qualitatively similar (as regards damage- dose dependence) to what was observed for boron. 4 This confirms earlier work 9 that suggested that both phosphorus and boron diffuse via an interstitial mechanism.

The use of carbon in reducing dislocation loops has been investigated. 14 These studies suggested that carbon re- duces the density of secondary defects, thereby suppressing boron diffusion. The results from this study suggest that carbon is more effective than oxygen and silicon as a diffusion-suppressing agent for phosphorus. That carbon consistently suppressed the dopant diffusion (when compared to the other two species) in our experiment, suggests that the behavior of the implanted carbon differs considerably from that of the implanted oxygen or silicon. This behavior is dose, temperature, and anneal-time dependent. Let us begin with the premise that after implanting the substrate with the three species, the same damage (interstitials) is generated in the silicon substrate. This assumption is based on the accuracy of the Monte Carlo simulations described earlier. Once the samples are annealed several factors come into play. The behavior of the three species is markedly different. Carbon and oxygen certainly act to suppress the diffusion of phosphorus when compared to silicon. The presence of the additional silicon atoms (among the multi- tude already in the lattice) does not adversely effect the properties of the substrate. However the additional carbon and oxygen atoms have a greater impact on the properties of the lattice. Since it is now generally accepted that phosphorus diffuses mainly by an interstitial mechanism, it is not unreasonable to conclude that these additional implanted carbon or oxygen atoms act as traps for the implant-generated interstitials. The similarity between carbon and oxygen ends here. It is most probable that these two elements diffuse differently and this reflects itself in the final phosphorus profiles of our experiment.

The use of carbon as a diffusion-suppressing agent from a technological viewpoint must be tested. In this context a second experiment was conducted where silicon-damaged substrates were implanted with phosphorus which was made to diffuse both in the presence and absence of a carbon coimplant. This replicates the real situation where phosphorus is made to diffuse in a damaged substrate.

As before, a 25 nm oxide was grown on <i00> oriented silicon wafers. Phosphorus was then implanted (60 keV, 1 • i013/cm2) through the screen oxide. A 15 min 900~ anneal was carried out to activate the dopant. The wafers were then divided into different lots. The first lot was left undamaged. All other lots received a silicon-damage implant (60 keV, 1 • i014/cm2). The damaged samples were then further subdivided. One lot received no carbon implant, while others were given carbon implants (45 keV, 1 • i0 I~ to i • 1014/cm2). The samples were then annealed for two different times (15 and 30 min) at two different temperatures (800 and 900~

A separate lot was used to carry out the LDD experiment. Phosphorus was implanted (45 keV, 5 • 1013/cm ~) through the 25 nm screen oxide into the undamaged substrate. This was followed by an arsenic implant (80 keV, 5 • 1015/cruZ). One sample then received no carbon implant, while two


,1. Electrochem. Soc., Vol. 141, No. 12, December 1994 9 The Electrochemical Society, inc. 3519

40

30

~ 20

10

. - ' " G , - ~ Si-15min. 1 e." /3-----El Si-30m in.

~ - -~O-15min , ~ - - - A O-30min.

40

30

~0

10

( ~ - ~ Si-15min. i~ El.----El Si-30min . . . . . . . "'" ~ - - s ~- -- ~. O--30min. ~.J"

//.--"

....... .-t~ f j

10 '2 10 ~ 10 '4 0 10,~ 10,3 "1-0~ , Silicon equi. damage dose (/cm ~) Silicon equi. damage dose (/era ~)

(a) (b)

Fig. 6. Net enhancements of phosphorus dopant profile motions for silicon- and oxygen-damaged substrates for various anneal times; (a) 800~ and (b) 900~

40

30

20

10

f..,1, j... ~/'f'j~E]

jl-"'" .................

G~--O Si-15 min. ~-.-.s Si-30 rain O- ,0 C-15 rain. Z~- -~ C-30 min,

0 , , , , i 10 '2 9 2 Silicon eqm. damage dose (Icm)

40

30

20

10

(9----O Si-15 rain. ~.D [:3-----.-El Si-30 min. / . . / / . / . . . . - .... O,-- r C-15 min. A-- - ~ C-30 rain.

/ j . . / ' / "

~ ~ ............ ~"~

'1'0 ~2 ' . . . . . . l 'd" ' ' | 'd" Silicon equi, damage dose (/cm =)

(a) (b)

Fig. 7. Net enhancements of phosphorus dopant profile motions for silicon- and carbon-damaged substrates for various anneal times; (a) 800~ and (b) 900~

other samples were given a carbon implant (45 keV, 1 • 1013 to 1 • 1014/cm2). All these samples were then annealed for 30 min at 900~ The entire second experiment is illustrated in the form of a flow chart in Fig. 8.

I 60 keV, le[d/cra 2 Si damage Lmplant.

~ n <I00> orlemed~

m~plant 15 min r176 c:~vadon ~nne~l

LDD Slluctur~ ? t

I 45 kCr 5el3/cm~ Pr~~ I ,rnplam.

15 mm 900~ m ,,, 9 POO~C A~ imp, s;

Fig. 8. Flow chart of the second experiment.

The profiles were extracted using SRP measurements and then normalized to a base dose. The normalized profiles were fed into SUPREM IV and the profile shifts extracted as described earlier in the Data Analysis section. Figure 9 shows profiles extracted from various samples after a 15 rain 800~ anneal. Figure 10 illustrates graphically the extracted profile shifts for all anneal conditions. Car- bon on the whole does not suppress the diffusion of phosphorus when compared to a damaged substrate without a carbon implant. However, a strong dose dependence in the profile shifts is observed for all anneals 9 For anneals at 800~ the diffusion of the dopant was suppressed as the dose of the carbon was increased; the profile motion peak- ing for a i • ]0~i/em 2 carbon dose. This phenomenon is not understood at this stage, but at such low carbon doses it is probable that not enough carbon exists to make a differ- ence in the interstitial concentration. For anneals at 900~ suppressed diffusion was observed for a few cases the numbers were within range of experimental error.

The results of the LDD experiment are illustrated in Fig. ii. The profiles for all cases lie close to each other, though diffusion with the lower carbon dose (i)< 10~3/cm z) resulted in a shallower junction depth. The numbers however are within the range of experimental error and we


1018

Fig. 9. Phosphorus profiles for various implant conditions after a 15 min/800~ anneal.

10 ~7

.~ 10 TM

0

10 TM

10 TM L_ 0.00 0.20

50.0

i ,~ As-implanted 9 St. damage, w/o C ~Si. damage + l e l l C OSi. damage + le12 C

St. damage + le13 C

0.40 0.60 0.80 Depth in kLm


70.0

60.0

50.0

40,00Lo

No carbon implant.

10 ~ 10 j~ 10 t~ 10 ~

Carbon damage dose {/cm;)

45,0

LV- 9 mn. 9o0~ ~ /

~f

= 40.0

35.0

30,00~0 ...... i;" ..... ;0 '~ ' ..... ~0'~' ..... i0 TM Carbon damage dose (/cm 2)

Fig. 10. Total phosphorus profile motions for various anneal conditions; (a, left) 800~ and (b, right) 900~

cannot say conclusively that the presence of carbon led to suppressed diffusion.

Conclusion The effects of implant species dependent diffusion have

been presented. Implantation by carbon/oxygen can reduce the net enhancement in the diffusion of phosphorus (as compared to a silicon-damaged substrate) under certain conditions. This reduction in the net enhancement was the most for carbon-damaged substrates. However, for longer and higher temperature anneals the effectiveness of oxygen is comparable to that of carbon. Silicon implantation damage almost always caused the maximum increase in the profile motion. The diffusion of phosphorus was shown to have a strong dependence on the dose of the damage species. In general there was an increase in the profile shifts as the dose of the damaged species was increased though this did not mean that doubling the dose resulted in twice the enhancement in the profile shifts. The enhancements in the net profile shifts consistently was greater at lower temperatures for a given damage dose. Finally, comparisons were

made with earlier work and good agreement was obtained for overlapping data points.

Once carbon was identified as the most effective diffusion-suppressing agent, its effectiveness was tested from a technological viewpoint as part of our second experiment. It was clearly brought out that the effectiveness of carbon in suppressing the diffusion in a predamaged substrate is limited to a few isolated cases which could well be at- tributed to statistical variations in the experiment. How- ever the strong carbon-dose dependence of the final dopant profile brought out some interesting points. An increase in carbon dose led to a shallower junction depth, indicating that the extra carbon could be acting as a sink for intersti- rials, as suggested earlier. Thus the use of carbon from a technological viewpoint for forming shallower junctions is limited. The results of the LDD experiment indicate that the presence of carbon does not appreciably effect the shape of the final profile.

Manuscript submitted April 4, 1994; revised manuscript received June 21, 1994. This was Paper 494 presented at the San Francisco, CA, Meeting of the Society, May 22- 27, 1994.


J. Electrochem. Soc., Vol. 141, No. 12, December 1994 9 The Electrochemical Society, inc. 3521

10 2 2

~.. 10 2o m,

E v

0

~ 10 TM

0 e- 0 0

0 10 T M

1014 0,00

Diff. with no C implant I - Diff with l e13 /cm-C

+ ...... • Diff with le14/cm 2 C

\

\ \ "..,

" \ ~ . \ ~'~§ ,

"----' +=r+

, I i I , I , I I

0.10 0.20 0.30 0.40 0.50 Depth in ~m

Fig. 11. Final phosphorus profiles for the LDD structure.

University of Florida assisted in meeting the publication costs of this article.

REFERENCES 1. S. Nishikawa, A. Tanaka, and T. Yamaji, Appl. Phys.

Lett., 60, 2270 (1992). 2. S. Nishikawa and T. Yamaji, ibid., 62, 304 (1993). 3. R. Liefting, R. C. M. Wijburg, J. S. Custer, H. Wallinga,

and F. W. Saris, IEEE Trans. Electron Devices, ED- 41, 50 (1994).

4. P. A. Packan and J. D. Plummer, Appl. Phys. Lett., 56, 1787 (1990).

5. H. Park and M. E. Law, ibid., 58, 732 (1991). 6. H. Park and M. E. Law, J. Appl. Phys., 72, 3431 (1992). 7. SUPREM-III user's Manual, TMA SUPREM-3, Version

8920 (1989).

8. T. L. Crandle, W. B. Richardson, and B. J. Mulvaney, Int. Elec. Dev. Meeting, p. 636 (1988).

9. P.M. Fahey, R B. Griffin, and J. D. Plummer, Rev. Mod. Phys., 61, 289 (1989).

10. A. E. Michel, W. Rausch, and P. A. Ronshiem, Appl. Phys. Let~., 51,487 (1987).

11. T. O. Sedgwick, A. E. Michel, V. R. Deline, S. A. Cohen, and J. B. Lasky, J. AppI. Phys., 63, 1452 (1988).

12. A. C. Ajmera, G. A. Rozgonyi, and R. B. Fair, AppI. Phys. Lett., 52, 813 (1988).

13. Y. Kim, H. Z. Massoud, and R. B. Fair, J. Electron. Mater., 18, 143 (1989).

14. Y. M. Kim, G. Q. Lo, D. L. Kwong, H. H. Tseng, and R. Hance, Appl. Phys. Lett., 55, 2316 (1989).

15. R. T. Hodgson, V. R. Deline, S. Mader, and J. C. Gelpey, ibid., 44, 589 (1984).


effects of low dose silicon, carbon, and oxygen...

Documents