the effect of implant energy, dose, and dynamic annealing on end...

The effect of implant energy, dose, and dynamic annealing on end-of-range damage in Ge+-implanted silicon

K. S. Jones and D. Venables Department of Materials Science and Engineering, University of Florida, Gainesville, Florida 3261 I

(Received 4 September 1990; accepted for publication 12 December 1990)

A study of the effect of Ge + implantation energy, dose, and temperature on the concentration of atoms bound by the extrinsic end-of-range dislocation loops in Si (100) wafers is presented. Plan-view and cross-sectional transmission electron microscopy observations of both the as-implanted and annealed (900 “C, 30 min) morphology were made. The implant energy was varied from 30 to 150 keV, the dose varied from 2x 1014 to 1 X 1016/cm2, and the temperature was varied by using three different wafer-cooling methods during the implantation. Increasing the implant energy, dose, or wafer temperature all resulted in significant increases (as much as two orders of magnitude) in the concentration of atoms bound by the end-of-range loops. Recent models have suggested that the concentration of end-of-range defects is related to the integrated recoil concentration beyond the amorphous/crystalline (a/c) interface. Correlation of TRIM-88 calculations with measured a/c depths show that the integrated recoil concentration beyond the a/c interface can explain both qualitatively and quantitatively the dependence of the “trapped interstitials” on the implant energy. However, this model can only qualitatively explain the temperature dependence of the defects, and it fails to account for the strong dose dependence when wafer heating is suppressed.

INTRODUCTION

It is well documented that the defect morphology of ion-implanted silicon can vary from a collection of point defects and clusters to amorphization of the silicon de- pending on the implant conditions. For silicon, this as- implanted morphology dictates the type and concentration of extended defects that form upon annealing. There are at least five different types of extended defects.’ However, subthreshold (category I) and end-of-range (category II) defects are the most common. Both common types of defects consist of extrinsic dislocation loops. However, the source and location of each defect type differs. As the name implies, subthreshold defects form upon implantation at doses below the amorphization threshold and typically occur at the peak of the implanted impurity distribution. In contrast, end-of-range (EOR) damage forms at doses above the amorphization threshold and is confined to the crystalline region adjacent to the amorphous/crystalline (a/c) interface.

These defects can have profound effects on dopant impurity profiles during annealing. For instance, interstitial fluxes created during both the formation and subsequent dissolution of extended defects can lead to enhanced diffu- sion of the implanted dopant species. Recent studies have attempted to better quantify these effects.2A However, it is necessary to obtain a fundamental understanding of the source of the extended defects if their effect on the implanted impurity profiles during annealing is to be quantitatively understood and modeled.

Although extensive studies on the nature, source, and stability of end-of-range defects have been reported,“5’6 the source of the interstitials for these extrinsic dislocation loops is still uncertain. We showed previously that the con-

centration of ions transmitted through the amorphous layer did not explain how the concentration of “trapped interstitials” bound by the end-of-range defects varied with dose for Ge implants at room temperature. Recently, Ga- nin and Marwick proposed that the integrated recoil concentration beyond the a/c interface, as determined from a TRIM (Monte Carlo) simulation, could explain the effect of implant energy on the end-of-range defect concentration. It was shown that decreasing the implant energy resulted in a lower integrated recoil concentration for Bi implants in silicon. This decrease was proposed to explain the experimental observation that no defects formed at low implant energies. Unfortunately, only two energies were investi- gated. In order to better understand the source of interstitials for the end-of-range defects in implanted Si and test the integrated recoil theory, an experiment exploring a greater range of parameters including implant energy, dose, and temperature was conducted. The results indicate that integrated recoil theory can explain the energy dependence of the end-of-range defect concentration, but it pro- vides only a qualitative explanation of the temperature dependence and it fails to explain the dose dependence when dynamic annealing is suppressed.

EXPERIMENT

The (100) 5-10-a cmp-type Si wafers were implanted with Ge + at doses of 2x 10’4/cm2 to 1 x 1016/cm2 and implant energies from 30 to 150 keV. The beam-current density was kept between 1.5 and 3 PA/cm2 to avoid possible dose-rate effects. The implant temperature was varied by attempting to control the amount of beam-induced wafer heating during the implants. This was achieved by using

2931 J. Appl. Phys. 69 (5), 1 March 1991 0021-8979/91/052931-07$03.00 @ 1991 American Institute of Physics 2931

Downloaded 04 Apr 2011 to 128.227.135.101. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

three different wafer-cooling methods: freon cooling the end station, water cooling the end station, and using no external cooling.

It has been well documented by Prussin, Margolese, and Tauber’ that a wayflow-type end station with no external cooling mechanism results in significant wafer heating and ion-beam-induced epitaxial crystallization’ (IBIEC) of the amorphous layer during ion implantation. When circulating freon is used to extract heat from the wafer, as in the waycool end station, substantially less wafer heating and IBIEC occurs.8*10 Based on the work of Prussin and co-workers, estimated wafer temperature ranges for the samples in the present experiment are 30 and 40 “C for the freon-cooled samples and 55 and 200 “C for the uncooled samples for doses of 2~ lOi and 1 x 10t6/ cm2 at 100 keV, respectively. Since water cooling is less efficient than freon cooling, the water-cooled samples rep- resent an intermediate wafer temperature between the freon-cooled and uncooled samples.

After implantation the samples were capped with a low-temperature (350 “C) chemical vapor deposition (CVD) oxide prior to annealing in order to eliminate the possibility of interstitial injection due to thermal oxidation during the anneal. Solid-phase epitaxial regrowth of the amorphous layers and development of the EOR defects was achieved with a furnace anneal at 900 “C for 30 min in flowing N,.

The amorphous-layer depths of the as-implanted samples were measured by cross-sectional transmission electron microscopy (XTEM) and by an optical technique using differential reflectometry.” The results of the “quick” optical method were calibrated with the more di- rect, but time-consuming, XTEM measurements to ensure consistency between methods. The XTEM measurements were performed on [l lo] cross-sectional specimens with the incident beam parallel to the [l lo] zone axis using bright-field, seven-beam imaging conditions on a JEOL 200CX TEM.

Both plan-view (PTEM) and XTEM specimens were prepared of annealed samples. The depth location of the EOR loops was obtained from the XTEM specimens using bright-field, two-beam g,,, and gc, imaging conditions. Quantitative analysis of the end-of-range damage was performed on weak-beam dark-field (g,,,) images of PTEM specimens. The mean dislocation loop diameter was obtained by averaging measurements of the longest dimen- sion of 20-30 loops per specimen. The area1 number density (loops/cm2) of loops was found by counting the number of loops in 2 pm2. As a measure of the “quantity” of end-of-range defects, the concentration of atoms bound by the dislocation loops (atoms/cm2) was calculated for each sample. This parameter was obtained by first calcu- lating the total area of the loops per square centimeter from the mean loop size and area1 number density, and then multiplying by the area density of atoms on { 111) planes. The EOR defects sometimes developed into a nearly planar dislocation network superimposed on discrete dislocation loops. In these cases, atoms bound by the discrete loops were calculated as above and then were

added to the atom density of a single plane to determine the total concentration.

Full-cascade Monte Carlo simulations of the Ge implants at 30, 60, 80, 100, and 150 keV were performed with version 4.3 of the TRIM-8812 code using a value of 13 eV for the displacement energy13 and the default value of 1 eV for binding energy. In order to ensure a good statistical repre- sentation of the tail region of the implants, at least 10 000 ions were calculated for each energy. The dose dependence of ion, vacancy, and recoil distributions was obtained from the normalized TRIM output by multiplying by the dose and appropriate unit conversion factors. The integrated recoil and integrated ion concentrations beyond the a/c interface were then calculated by summing the number of recoils (or ions) at depths below the experimentally determined amorphous-layer depth and multiplying by the 30-b; depth increment of the data. In addition, the damage density distribution was calculated from the TRIM data by multiplying the total vacancy distribution by the displacement energy 13 eV.

It is possible to determine the threshold damage density (TDD) or the amount of energy deposited into elastic collisions that is necessary to amorphize the Si. This is typically done by correlating the measured depth of the amorphous layer with the calculated damage density distribution. If the threshold damage density is known, it is possible to predict the amorphous-layer depth and thus the position of the end-of-range damage. If the amount of wafer heating is minimized (e.g., the freon-cooled samples), then the threshold damage density is found to be relatively constant for a given implant species.” However, if wafer heating is not suppressed and IBIEC results, the threshold damage density can vary over orders of magnitude, making modeling of the amorphous layer thickness much more difficult.

RESULTS

The amount of ion-beam-induced wafer heating was found to be a critical factor in these quantitative studies of end-of-range damage. For example, plan-view and cross- sectional TEM micrographs in Fig. 1 compare the effect of water cooling to no external cooling for otherwise identical 150-keV Ge implants at a dose of 2x 1015/cm2. Water cooling resulted in a lower wafer temperature, less dynamic annealing, and thus less regrowth of the amorphous layer.

For the uncooled sample, the additional IBIEC of the amorphous layer was only 100 A, yet it resulted in an increase in the amount of end-of-range damage of -98%. The integrated recoil and integrated ion concentrations beyond the a/c interface, as determined from the TRIM-88 simulations, showed corresponding increases of 42% and 73%, respectively. As a result of this sensitive dependence on the amount of IBIEC that occurred, a careful distinc- tion is drawn in the following discussion between effects associated with varying the implant dose or energy and effects associated with wafer heating during the implant.

In order to test the integrated recoil model as a function of energy, Ge + implants using the water-cooled end

2932 J. Appl. Phys., Vol. 69, No. 5, 1 March 1991 K. S. Jones and D. Venables 2932


XTEM

As Implanted Annealed As Implanted Annealed

30 keV 60 keV

Annealed Annealed

FIG. I. Bright field (gllO, B near [liO]) cross-sectional and weak-beam dark-field ([220] reflection, B near [ODI]) plan-view TEM micrographs of samples implanted with 150-keV Ge+ at 2~ 10’5/cmZ using (a) water- cooled end station and (b) no external means of cooling. The higher wafer temperature in (b) caused partial regrowth of the amorphous layer, leading to coarser end-of-range dislocation loops (i.e., more trapped interstitials) upon annealing.

station were performed over an energy range of 30-150 keV at a constant dose of 2X 1015/cm2. The amorphous- layer depth increased steadily with energy as shown in Fig. 2. TRIM-88 simulations indicate that the amorphous layer thickness as a function of implant energy can be modeled reasonably well by a single threshold damage density (TDD) of - 1.4X 102’ keV/cm3. This TDD value is a factor of 5 higher than that obtained from the freon-cooled implants as a function of dose (see below), indicating that

2000 , 1 I 1 I

u 1500 -

5 if cl : 1000 - 0 Ii z

% 500 -

A Water cooled .--- Constant TDD of

1.4x1 0” keV/cm’ ,‘i , I I

I

*, I

Dose = 2~10’~ cm-’

O”.‘.“““““““‘~ 0 50 100 150 200

Implant Energy ( keV )

FIG. 2. Amorphous-layer thickness as a function of implant energy using the water-cooled end station. Comparison with TRIM-U calculations shows that the data may be modeled reasonably well with a single threshold damage density value.

FIG. 3. Plan-view TEM micrographs of annealed samples showing the increase in end-of-range damage with implant energy at a constant dose of 2 x lO”/cm’ (weak-beam dark-field 12201 reflection, B near [OOl]).

some IBIEC did occur in these water-cooled implants. However, the fact that the TDD is relatively constant for all but the lowest energy suggests that the extent of IBIEC was approximately the same for each implant energy. This conclusion is consistent with the fact that the power input by the ion beam (which is the source of wafer heating and IBIEC) changed by a factor of only -2.5 as the ion energy increased from 60 to 150 keV.

Figure 3 shows PTEM micrographs of the annealed samples as a function of implant energy. Clearly, the loop concentration increases with increasing implant energy. A quantitative assessment of this increase is plotted in Fig. 4, which shows the atoms bound by end-of-range dislocation loops (i.e., trapped interstitials) as a function of implant energy. Shown for comparison in Fig. 4 are the integrated recoil and integrated ion concentration beyond the a/c interfaces as derived from TRIM simulations using the experimentally observed amorphous-layer depths (Fig. 2). The integrated recoil concentration models the trapped interstitial trend quite well, although the absolute values of the parameters differ substantially. However, since the TRIM code makes no attempt to account for Frenkel pair recombination, the absolutes values are not expected to match. The integrated ion concentration levels off slightly above 80 keV and therefore does not model the trapped interstitial trend quite as well as the recoil model. In addition, the concentration of atoms bound by the EOR dislocation loops is as much as a factor of 8 greater than the integrated ion concentration, and so the ion model fails quantitatively. Thus, the recoil model appears to best predict the variation of interstitials trapped in end-of-range defects



$ 10” Dose = ‘ye beyond

E

/I /

a/c interface

z E

4

” 1o16 1 t

1o15 7

10” T

L

Trapped interstitials J

\ ions beyond a/c interface

1o13 1 8 ’ s ’ ’ ’ ’ ’ ’ ’ . ’ ’ ’ ’ ’ * * ’ 1 0 50 100 150 200

Implant Energy ( keV )

FIG. 4. Comparison of experimentally determined concentration of interstitials trapped by end-of-range dislocation loops with integrated recoil and integrated ion models based on TRIM-C+ calculations. The recoil model follows the trapped interstitial trend quite well, although the absolute values differ substantially because of inherent limitations of the TRIM code.

with implant energy for IBIEC-influenced implants. To study the effect of dose on end-of-range damage,

IOO-keV Ge $- implants were performed over a dose range from 2X 1014/cm2 to 1 X 10’6/cm2. The effect of dynamic annealing due to wafer heating was also studied by com- paring water-cooled and freon-cooled samples at each dose.

The amorphous-layer thickness increased steadily with dose for the freon-cooled samples, as shown in Fig. 5. In contrast, the water-cooled a/c depths remained constant

2000

-2

5 1500 t-

D

: FI 5 ZG 1000

%

A Water cooled , Cl Freon cooled

‘--- Constant TDD of,/” 2.6x1 D2’ k&/cm’ ,’

, q

.d q

P’ ,

d’ Ll/

* -,-A\ /

A

i \ A

500 10”

Energy = 100 keV

I * * 1111.11 * I1 IVL

10’5 10’6 10” -2

Dose ( cm )

FIG. 5. Dose dependence of the amorphous-layer thickness for water- cooled and freon-cooled end stations. The freon-cooled results could be modeled by a constant TDD value only at doses below 2x 10iS/cm2.

2 x 10’41cm’ 1 x 10’%m2

5 x 10”/cm2 1 x 10”lcm2

FIG. 6. Plan-view TEM micrographs of annealed samples showing the increase in end-of-range damage with dose for the freon-cooled end station (weak-beam dark-field [220] reflection, B near [Ool]).

over much of the dose range and even decreased at the highest doses (Fig. 5). Clearly; water cooling was a less efficient means of stabilizing the wafer temperature, resulting in substantial dynamic annealing and IBIEC of the amorphous layer. As a result, it was not possible to model the amorphous depth of the water-cooled samples in terms of a threshold damage density. Figure 5 also indicates that a threshold damage density of -2.8 X 10” keV/cm3 did successfully model the freon-cooled results up to a dose of 1 X 10”/cm2. Beyond this dose, the TDD increased, sug- gesting that IBIEC effects might have occurred at high doses, even for the freon-cooled samples.

The plan-view results upon annealing for the two different cooling methods are shown in Figs. 6 and 7. In both cases the increasing dose produced a dramatic increase in defect density. As in Fig. 1, the data in Figs. 6 and 7 show that at a given dose, an increase in wafer temperature resulted in a larger concentration of trapped atoms in EOR dislocation loops. This difference is most dramatically il- lustrated for the 5 X lO”/cm* samples, where the higher wafer temperature (water cooling) changed the annealed structure from discrete dislocation loops (Fig. 6) to a dislocation network (Fig. 7).

Quantitative measurements of the end-of-range defects are shown in Fig. 8(a). The increase for both cooling methods as a function of dose is approaching two orders of magnitude. The IBIEC influenced (water-cooled) samples exhibited a larger number of trapped interstitials than the freon-cooled samples, particularly at doses above 5 X 1014/ cm2. The predictions of the integrated recoil and integrated ion models, using the experimentally determined



7 E 1o15 I I 0

A Water cooled (4

2 x 10”lcm’ Q&m 1 x 10’6/cm2

5 x 10’6/cm2 1 x 10’61cm2

FIG. 7. Plan-view TEM micrographs of annealed samples showing dose dependence of end-of-range damage for the water-cooled end station. Comparison with Fig. 6 indicates that the higher wafer temperature for the water-cooled samples resulted in more end-of-range damage (weak- beam-dark field [220] reflection, B near [OOl]).

amorphous-layer depths, are shown in Figs. 8(b) and 8(c), respectively. For the water-cooled samples, where substantial IBIEC occurred, both models correlate reasonably well with the experimental results at doses above 5 X 1014/cm2. However, the recoil model fails both qualitatively and quantitatively to explain the trapped interstitial concentration in the freon-cooled samples at the lower doses where IBIEC was suppressed. Indeed, if a constant threshold damage density is assumed, the integrated recoil concentration remains constant over the entire dose range studied, while the end-of-range defects increase by at least an order of magnitude. The dose dependence for the freon- cooled case also cannot be explained by the integrated ion model, which predicts a slight decrease in number of trapped atoms with increasing dose.

The integrated recoil model is slightly more successful in explaining the effects of wafer heating on end-of-range damage. The difference between the recoil concentration calculated for the water-cooled and freon-cooled samples increases steadily with dose. This trend parallels that of the trapped interstitials [Fig. 8(a)]. However, the correlation is, at best, only qualitative. For instance, at a dose of 1 x 1016/cm2, the recoil concentration difference is a factor of 20, while the trapped interstitials differ by a factor of only 3. This discrepancy becomes even more exaggerated with the integrated ion model.

DISCUSSION It is apparent that neither of the models considered

thus far can fully explain how the concentration of atoms

x s j 10’5 G 2 .E 0

E 10” 2

z 2 m

E o 10’3 I 2 10”

_ 10’9

: E ”

! y lo’e u 2 -

%

-F! rg 10”

$ m

.% h: LT

10’6 ,.

1 I A Water cooled (b) 0 Freon cooled

.--- Constant TDD of *

2.8x1 0” keV/cm’

/ i / /+

J / AC/ ______ ----. Energy = 100 keV

I I , ,,> ,,,, , , ,,,,

IO” 10’5 10’6 10”

0 Freon cooled

i

Energy = 100 keV

1 , , ,,, ,,,, , ,,,,,

10’5 lo’e 10” -2

Dose ( cm )

10’6

: E

v 10’5

i E 2 1o’4 %

10’2 I I II**III_ I I., 1*1*+ < . ..I, .-

lo’+ 10’3 10” 10” -2

Dose ( cm )

-2 Dose ( cm )

I I

. Water cooled (cl

Energy = 100 keV

FIG. 8. Comparison of the dose dependence of (a) experimentally derived trapped interstitial concentration with (b) integrated recoil and (c) integrated ion models. Both models failed to explain the dose dependence for the freon-cooled samples where IBIEC was suppressed.

bound by the end-of-range dislocation loops varies with dose, energy, and temperature. The integrated recoil model was more successful at explaining the energy dependence, and both models correctly predicted the dose dependence when partial regrowth of the amorphous layer occurred due to wafer heating during the implant. However, neither



model could explain the dramatic increase in bound-atom concentration with dose when IBIEC was suppressed by use of a freon-cooled implanter end station. In addition, both models failed to quantitatively account for the temperature dependence of the end-of-range defect concentration. This discrepancy between model and experiment could arise from two possible sources. First, the concepts expressed in the models could simply be erroneous, or, alternatively, the TRIM simulation upon which the models’ predictions depend may not accurately reflect the processes involved in the experiment.

The concept embodied by the integrated ion model is that a net surplus of interstitials must exist because of the inherently nonconservative nature of ion implantation. Simply put, there are more atoms than available lattice sites, and these surplus atoms may be expected to aggre- gate as extrinsic defects in the target. This concept has proven highly successful in explaining the dose dependence of subthreshold defects in ion-implanted silicon.’ The ex- tension of this idea to end-of-range defects proposes that the net surplus of atoms correspond to the number of ions that come to rest in the crystalline region beyond the a/c interface. Previous results assuming a Gaussian ion distribution’ and the present resulting using TRIM simulations of the ion distribution both show that the integrated ion concentration model does not explain the dose dependence of the end-of-range defect concentration. The integrated ion model also consistently underestimated the trapped interstitial concentration as the energy was varied. It is possible that some channeling occurred prior to amorphization, leading to a larger ion concentration beyond the amorphous layer than the TRIM simulations predict. However, over 50% of the dose would have had to have been chan- neled to regions below the amorphous layer before the integrated ion model would be able to quantitatively account for the observed differences. Experimental results from heavy ion implants show no such discrepancy.14 Thus, if the TRIM simulations are accepted as reasonably accurate, then the model must be rejected. Evidently, there are just too few ions beyond the a/c interface for this to be the sole source of excess point defects for the end-of-range dislocation loops.

The integrated recoil model proposes that a net surplus of interstitials exists beyond the a/c interface because re- coiling Si atoms are knocked from the amorphous layer into the crystalline region without an associated vacancy with which to recombine upon annealing. As noted earlier, the TRIM code does not account for Frenkel pair recombination so that the calculated integrated recoil values are much higher than the actual trapped interstitials in the annealed samples. Moreover, it is not possible to calculate the net interstitial concentration (recoils minus vacancies) from the TRIM results because of error in subtracting two large numbers. In this regard, TRIY may not be the best tool for modeling end-of-range defect concentration. Sim- ulations based on the Boltzmann transport equation have been used to predict net interstitial and vacancy concentrations in GaAs,15 and may prove better at predicting the absolute concentration of interstitials in these experiments.

Despite the error introduced by not accounting for Frenkel pair recombination, the integrated recoil model, as previously discussed, predicts the dose, energy, and implant temperature effects on the trapped interstitials better than the integrated ion model. However, there are still major discrepancies with the integrated recoil model, particularly when wafer heating is suppressed.

There have been previous speculations’6 that the morphology of the a/c interface is important and that interstitials released from regrowing amorphous islands for very rough interfaces are the source of interstitials for the EOR dislocation loops. Our observations do not support this model. For example, as a function of dose the roughest a/c transition region occurred for the lowest dose sample (fewest atoms bound by EOR loops) and the interface roughness decreased with increasing dose, whereas the concentration of trapped interstitials increased.

CONCLUSIONS

A careful comparison of the dependence of the atoms bound by the EOR dislocation loops as a function of implant energy, dose, and implant temperature was made. Increasing the energy, dose, or implant temperature resulted in a dramatic increase in the atoms bound by the EOR dislocation loops. It was shown that the integrated recoil model could qualitatively and quantitatively explain the energy and dose dependence of the EOR defects when IBIEC due to wafer heating occurred. The integrated ion model predicted qualitatively the energy dependence, but failed to quantitatively account for the number of trapped interstitials. However, neither the integrated recoil or integrated ion models could account for the strong dose dependence when wafer heating was suppressed. In addition, both models failed to quantitatively explain the temperature dependence. Thus, none of the models considered in the present work could fully explain the dependence of the trapped interstitial concentration on dose, energy, and temperature. It is proposed that modeling based on solving the Boltzmann transport equation may provide a better modeling tool for understanding the source of end-of-range damage in ion-implanted silicon.

ACKNOWLEDGMENTS

The authors wish to thank J. P. Jackson of IICO, Inc. for performing the water-cooled implants, Wei Xi of the University of Florida for performing the differential reflec- trometry measurements, and Jeff Fetzko for his able assis- tance in performing these experiments.

‘K. S. Jones, S. Prussin, and E. R. Weber, Appl. Phys. A 45, 1 ( 1988). ‘R. B. Fair, IEEE Trans. Electron Devices ED-35, 285 (1988). ‘Y. Kim (unpublished). 4R. Deaton, D. Gosele, and P. Smith, J. Appl. Phys. 67, 1793 (1990). s A. C. Aimera and G. A. Rozonyi, Appl. Phys. Lett. 49, 1269 ( 1986). ., ‘S. Mader, in Ion Implantation: Scierkand ?echnology, edited by J. F. Ziegler (Academic, New York, 1984), p. 109.

‘E. Ganin and A. Marwick, in Ion Beam Processing of Advanced Elec- tronic Materials, edited by N. W. Cheuna, A. D. Manvick. and J. B. Roberto (Materials Research Society, Pittsburgh, PA, 1989), Mater. Res. Sot. Symp. Proc. Vol. 147, p. 13.



‘5. Frussin, D. I. Margolese, and R. N. Tauber, J. Appl. Phys. 54, 2316 (1983).

‘R. G. Elliman, J. S. Williams, D. M. Maher, and W. L. Brown, in Beam Solid Interactions and Phase Transformations, edited by H. Kurz, G. L. Olson, and J. M. Poate (Materials Research Society, Pittsburgh, PA, 1986), Mater. Res. Sot. Symp. Proc. Vol. 51, p.127.

‘OS. Prussin, D. I. Margolese, and R. N. Tauber, J. Appl. Phys. 57, 180 (1985).

” R. E. Hummel, W. Xi, P. H. Holloway, and K. A. Jones, J. Appl. Phys. 63, 2591 (1988).

‘*J P. Biersack and L. G. Haggmark, Nucl. Instrum. Methods 174, 257 .~ (1980).

“J M Poate and J. S. Williams, in Ion Implantation and Beam Process- . ing, edited by J. S. Williams and J. M. Poate (Academic, New York, 1984), p. 13.

14J. F. Ziegler, J. P. Biersack, and U. Littmark, The Stopping and Range of Ions in Solids (Pergamon, New York, 1985).

“L. A. Christel and J. F. Gibbons, J. Appl. Phys. 52, 5050 (1981). “N. R. Wu, P. Ling, D. K. Sadana, J. Washburn, and M. I. Current, in

Defects in Silicon, edited by W. M. Bullis and L. C. Kimerling (Elec- trochemical Society, Pennington, NJ, 1983), p. 366.



the effect of implant energy, dose, and dynamic annealing on end...

Documents