- Email: [email protected]
Channeling effect of P implantation in Si(100)
Nuclear Instruments North-Holland
~~anneling
and Methods
in Physics Research
1061
(1991) 1061-1066
effect of P implantation
Ning Yu and Wei-Kan
in Si( 100)
Chu
Texas Center for Superconductioity. and Department
Bijoy Patnaik
B59/60
of Ph_v.ws, Uniuer.~it.vof Ilouston. Houston, TX
77?04, O’SA
and Nalin Parikh
D~~urt~e~tof Physics and Astronon!v: ~njf~e~sit~of North Carolinu. Chapel Hill. NC -77514. USA
Sean Corcoran
and Charles
Kirschbaum
Charles Eoans And Associates, Redwood C’1t.v.CA 94063. USA
Kody Varahramyan IBM Corporatron, Essex Junction, VT 05452 USA
With the objective of obtaining shallow junctions, we have made a comparative study of the distribution profiles of phosphorus ions implanted into Si(100) at 5 to 200 keV. along different crystalline directions. viz. the [OOl] axis, the (310) plane with 7O tilt away from [OOl]. and a random equivalent direction, defined by 5.2O tilt. and 7’ rotation from the (100) plane. The P implants were also performed in amorphous Si, and through a thin SiOz layer at the same energies for comparison. The P depth profiles in Si were measured by SIMS with 14.5 keV Cs ions. The damage profiles in SitlOO) crystals were studied by RBS/channeling of 2 MeV He ions. The role of the implants along the selected orientations is discussed in light of the resulting depth of the tails of the implanted dopants distribution.
1, Introduction Ion implantation in single crystals is often unintentionally accompanied by the channeling effect either due to misalignment of the crystals or due to the “feed-in” process during the slowing down of ions in crystals, producing a long tail of implanted dopants. Some efforts have been made for boron implants in Si(100) to study the channeling tail as functions of tilt and rotation angles of the crystals, both from theoretical calculation [1,2] and experiment [1,3]. Normally there are two different approaches to minimize the channeling tail. One is by preamorphizing the crystal through Si or Ge ion implants [4]. An alternative is to carefully align the crystal in a “random equivalent” direction prior to the dopant implantations. Cho et al. have developed a criterion to sort out the random equivalent directions for Si(100) wafers along which the B implantations give shallower profiles, as compared to implantations with a tilt angle of 7O regardless of the rotation angle [l]. Similarly to B in Si(lOO), the angular dependence of the P doping profiles in Si has not been systematically investigated [5,6]. Recently, we have applied the same 016&583X/91/$03.50
Publishers
channeling criterion to P implantations in Si(100) crystals to determine the random equivalent directions. The implantations of 3*P have been made along the [OOl] axis, the (310) plane with 7’ tilt away from [OOl]. the random equivalent direction. and through amorphous Si and through thin SiOz layers (with thicknesses of 160 and 290 A). The SIMS and RBS/channeling measurements were carried out to determine the P distribution and the damage profiles in Si, respectively.
2. Calculation The existence of the critical angles for axial and planar channeling effects were demonstrated quite well by RBS/channeling measurements of light energetic ions on a variety of single crystals [7]. Simple formulae for these angles were obtained from Lindhard’s channeling theory [8-111 and Barrett’s formulae through computer simulation of the channeling effect [7]. Based on those formulae and the analytic method employed by Cho et al. [l] for boron implants in Si(100). the critical angles of P in Si(100) for different axial and planar channeling were calculated. A tilt-
B.V. (North-Holland)
VIII. SEMICONDUCTORS
N. Yu et al. / Channeling effect of P implantation in Si(100)
1062
rotation map was constructed with the channeling regions extended from the corresponding axial channeling points or planar channeling lines into circular area or bands. The radius of the circle or the width of the band represents the amount of critical angle at different implant energies. The channeling maps for P implanted in Si(100) at 10, 25, 50 and 100 keV are plotted in fig. 1 with the unchanneling regions darkened. The coverage of the map with channeling regions at 10 keV in fig. la indicates that, based on our analysis, there are no random equivalent directions for implantation. At 25 keV, several small darkened islands show up in fig. lb. The darkened area, corresponding to the regions where the channeling effect is minimized, increases with increasing implantation energy, as seen from figs. lb to Id. We could choose the center of one of the darkened regions close to the (100) plane with a tilt of 5.2’ and rotation of 7” as the random equivalent direction for all implants at different energies.
3. Experimental Si(lO0) wafers, B-doped to a resistivity of 14-22 Dcm, were used for all the implantations. Some of the Si(100) wafers were oxidized at 95O’C in dry 0, (with 4.5% HCl) ambient to grow 160 A and 290 A SiO,. For P implantations in Si(100) crystals, we first aligned the crystal with respect to the incident 200 keV He beam along the specific directions, viz. the [OOl] axis, the (310) plane with 7O tilt away from [OOl], and the random equivalent direction. This was accomplished by measuring backscattering yield versus the tilt and rotation angle of a two-dimensional goniometer. The alignment along the (310) plane or random equivalent direction was obtained by determining the coordinates of the [OOl] axis, the (100) and (310) planes, and adjusting the goniometer correspondingly. For P implants through thin SiO, layers, the [OOl] direction of Si(100) substrates was also aligned along the beam direction. For P implantations in a-Si, the
(b) 25 keV
(a) 10 keV
/
0 rb ‘0
G
,
,
[0011 Tilt Angle
2
4
6
Tilt Angle
(degree)
8
,$a
8
.
-0
10
(degree)
(c) 50 keV
. [OOll
2
4
Tilt Fig. 1. Tilt-rotation
6
Angle
8
10
(degree)
maps of P ion channeling
[OOll
2
4
6
Tilt Angle
8
-0
10
(degree)
in Si(100) at (a) 10, (b) 25, (c) 50 and (d) 100 keV, showing the channeling effect is minimized.
the darkened
regions
where
N. Yu et al. / Channeling effect of P implantaiion
1063
in Si(lO0)
Table 1 The energies, doses [atoms/cm21 and alignments of P implantation in Si
W11 (310) Random a-%
10 keV Pz
25 keV P2
25 keV P
50 keV P
100 keV P
200 keV P
1 x 10’4 1 x 10’4 1 x 10’4 _
4x10’4 4 x 10’4 4x10’4 4x10’4
5x 5x 5x 5x
1.1 x 1.1 x 1.1 x 1.1 x
9x10’4 9x10’4 9 x 10’4 9x10’4
1.5x10’5 1.5x10’5 1.5 x 10’5 1.5 x 10’5
10’4 10’4 10’4 10’4
Si(lO0) wafers were preamorphized to a depth of 2500 A by Ar ion implantation at 50 and 200 keV with a dose of 2 x 10r5/cm2 (for P ion implant at 200 keV, a Si wafer preamorphized by Ar ion implantation up to 400 keV was used). After alignment or preamorphization. the P ion beam was tuned on within the same ion source, and implanted into the aligned Si samples. The P ions were implanted at 25, 50, 100 and 200 keV with doses varying between 1 x lOI and 1.5 x 1Or5 atoms/cm*. Molecular P2 ions
10’5 10’5 10’5 10’5
at 10 and 25 keV (equivalent to 5 and 12.5 keV of atomic P ions) were also implanted for the purposes of comparison. The implantation parameters are listed in table 1. An X-Y angular scanner was used to provide uniform implantation across the implanted region. The angular variation from the edge to the center of the implanted area was about 0.13’, estimated from the distance (2.7 m) of the scanner to the sample, and the dimension (12 mm) of the implanted area. In order to obtain the P concentration depth pro-
ll.“I”.‘I.‘l
[ool] Implant
~
- - -(310)Implant
,.g- 1020
o
Random Implant
& 4 1019
In a-S
8 ‘Z $ 1018
5x1014at./em
25 keV P
5 x 0 101’ u & 1016 0
500 Depth
” 1000 (A)
0
1500
500
1000
1500 20000 2500 Depth (A)
0
~
m-lo2
-
-(310)Implant Random Implant
100
8 ._ z 101 I: B B 5 lo1
3500
[OOl] Implant o
>-&lo1 J
3000
9xlOl
keV
P
4at./cmi
b
6
lo1 0
0.1
0.2
0.4 Deptk3(pm)
0.5
0.6
0
0.8
0.2
Keith
1
&f
Fig. 2. P profiles in Si(100) from SIMS analysis, implanted along the [OOl]axis, the (310) plane with 7 o tilt away from the [OOl], the random equivalent direction and in a-Si; at (a) 10 keV P2. (b) 25 keV P, (c) 50 keV P and (d) 100 keV P. VIII. SEMICONDUCTORS
files. secondary ion mass spectrometry (SIMS) analysis was performed by a CAMECA IMS-3f using 14.5 keV Cs ion bombardment and negative secondary ion detection. The implantation damage distribution of Si(100) has been analyzed by Rutherford backscattering spectrometry (RBS)/~hanneling of 2 MeV He ions. The grazing incidence geometry along the [llO] crystal axis was employed to enhance the depth resolution for analyzing the samples implanted at energies lower than 100 keV. while the normal geometry along [OOl] was used for samples implanted at higher energies. The RBS detector was placed at 145” with respect to the incident He beam.
~
The P profiles in Si from SIMS measurements are plotted in figs. 2a-2d separately, for different implantation energies. In each subfigure, the profiles of P implanted along the [OOI] axis, the (310) plane (with 7O off [OOl]). random equivalent direction (with S.2* tilt and 7O rotati~~n) of Si(100) crystal. and in a-% are plotted together for comparison. Overall. from figs. 2a-2d, it is obvious that the P profiles along the (310) plane direction are almost identical to those along a random direction, within SIMS accuracy. This feature could be explained in terms of the damage created by the nuclear energy deposition of ion i~lplantation. Because the nuclear energy deposited by P inlplantation is much higher than that by B implantation, the channeling effects of the planes with high Miller indices such as (310) and (210) could consequently be obliterated very quickly. It may also be noted that the differences in the P profiles between the samples implanted along the [OOI] direction and those implanted along a random equivalent direction are very large for incident energies higher than 25 keV. When the implant energy is reduced from ZOOto 12.5 keV. the ratio of the depth (for a concentration level of 1 X lOI atoms/cm”) of the P distribution resulting from the [OOl] implantation to that from the random implantation is decreased from 1.9 to 1.1. On closer examination. we can distinguish among the three distinct energy regimes in which the differences between the P profiles from the [OOI] direction implantati~~ll and the random direction implantation do not hehave in the same manner. For incident energies higher than 50 keV. a small shift in the position of the concentration peak could be discerned, in addition to the large differences in the respective tails. With implantation energies ranging from 12.5 to 50 keV, there remains only the differences of the two tails, while the peak positions of the two P profiles coincide. For energies lower than 12.5 keV, even the difference between the two tails vanishes. This is consistent with the notion
Random Implant
*
Thru 160 8, Si02
-----Thru
290.& Si02
IO’ 0
4. Results
[OOI] Implant D
0.1
0.2
0.4
0.5
0.6
Fig. 3. P profiles for implants through 160 A and 290 A SKI, layers along the Si [OOl] axis at 50 keV, along with those for implants along the (OOl] axis and the random equivalent direction without the Si02 layer. and in a-Si.
that the channeling effect is enhanced by lowering the implantation energy (as is evident from fig. la). When the implantation energy is lowered. the critical angle for channeling becomes large enough so that the random equivalent directions no longer exist. The SIMS analysis was performed on samples of SiO,/Si(lOO) implanted by P ions without stripping off the SK& layers. Fig. 3 gives the P profiles for implantation along the [OOl] direction through 160 A or 290 A SiO, layers into Si(lO0) at 50 keV. referred to those for implants in a-Si, along the [OOl] axis and along the random equivalent direction. The P profile for implantation through 160 ,& SiO, is very similar to that for implantation along the random equivalent direction, without the oxide layer. The implantation through 290 A SiO, gives a shallower profile. This is mainly caused by the increase of the angle divergence of ions passing through this thicker amorphous layer, which reduces the channeling effect. Thus. in the sense of reducing the channeling tail of the P distribution at 50 keV implantation, a 160 A SiO, thin layer at the surface of a Si(100) crystal has an effect equivalent to aligning the single crystal along the random equivalent direction. The damage profiles in the Si crystals. caused by P implants under the experimental conditions. were obtained by the RBS/channeiing analysis with a 2 MeV He beam and are displayed in figs. 4a-4d. These correspond to the P profiles shown in figs. 2a-2d, respectively. As can be seen, the damage profiles for implantation along the channeling and the random directions are almost identical at implant energies lower than or equal to 50 keV. However, for a P implant ion energy reaching 100 keV. a large shift of about 220 A of the damage distribution towards greater depths is observed for the [OOl] implants relative to that for the random
1065
N. Yu ei al. / Channeling effect of P i~l~la~fation in Sr{fOOt
2 MeV He - Si impl. by 25 keV P
0
900
800
1000 1100 E (keV)
1200
250 2 MeV He+
- Si
-I
./cd
II .Y sr ~1500
~e,,thiA,-
.E $ s m1000
:
%
2
900
1000 1100 E (keV)
.i 1200
1 Implsnt
-
-
1200
1300
2 MeV He+ _ Si impl. by 100 keV
P
-Virgin [OOl]
j: M 1500 .5 % Z m 1000 u %
5: Y u @a 500
800
1000 1100 E (keV)
2000
: : : : : I : : : : 0I loo0 dil
-.-,-‘---,-,-..-l-,~i,~
900
--[@II
keV P
2000
0
800
1300
1300
500
---01’ ’ 800
s
’
-__ r-r ._I ’ * ’ r. j x .-+900
1000
1100 E (keV)
Fig. 4. R3S/channeling spectra of 2 MeV He ions channeled along the [llO] axis of Si(100) wafers implanted [OOI]axis and the random equivalent direction at (a) 10 keV Pz, (b) 25 keV P and (c) 50 keV P: and the spectrum
[OOl] axis for implantation
implants. These features are consistent with the SIMS results, confirming that it is difficult to get rid of the channeiing effect for low energy implants of P in Si(100) because of the opening up of the channeling region in all directions.
5. Conclusions For P implantation at 25-200 keV in Si(lOO), a pre-alignment of the Si crystal along the random equivalent direction (with a tilt of 5.2” and rotation of 7 o ) results in a large decrease in the tail of the P profiles as well as in crystal damage, as compared to those when the crystal is aligned along the [OOlJ axis. However, the channeling effect could not be avoided for implant energies lower than 25 keV. This is due to the disappearance of regions corresponding to absence of channeling in the tilt-rotation map. In other words, there is
~
.
’
,zbs”
1---T
1200
1300
by P ions along the channeled along the
with (d) 100 keV P.
no random equivalent direction to be chosen at these energies. For heavy ion implants such as P. planes with high Miller indices such as (310) no longer affect the channeling tail in the P profile in Si(100). This feature is totally different from that of light ion implant such as B, which causes significantly less damage in Si. As far as the reduction of the tail of the P distribution in Si(100) is concerned, a 160 A SiO, thin layer has the same effect as pre-aligning the Si crystal along the random equivalent direction. at a 50 keV implantation of P ions.
Acknowledgements This work was mainly supported by IBM Corporation. The RBS/channeling analysis was done at University of North Carolina at Chapel Hill. A small part of the SIMS analysis was carried out at North Carolina VIII. SEMICONDUCTORS
1066
N. Yu et al. / Channeling effect of P implantation in Si(IO0)
State University. This work is also supported by the Texas Center for Superconductivity at the University of Houston which is supported through grants from DARPA, the State of Texas and private foundations.
References [l] K. Cho, W.R. Allen, T.G. J.J. Wortman, Nucl. Instr. [2] K.W. Brannon and R.F. Meeting, San Diego, CA, [3] A.E. Michel, R.H. Kastl, J.A. Gardner, Appl. Phys.
Finstad, W.K. Chu, J. Liu and and Meth. B7/8 (1985) 265. Lever, Electrochem. Sot. Fall 1986. S.R. Mader, B.J. Masters and Lett. 44 (1984) 404.
[4] M.C. Ozturk, J.J. Wortman, CM. Osbum, A. Ajmera, G.A. Rozgonyi, E. Frey, W.K. Chu and C. Lee, IEEE Trans. Electron. Devices ED-35 (1988) 659. [5] R.J. Schreutelkamp. F.W. Saris, J.F.M. Westendorp. R.E. Kaim, G.B. Odlum and K.T.F. Janssen, Mater. Sci. Eng. B2 (1989) 139. [6] C. Bresolin, C. Zaccherini, M. Anderle and R. Canteri, Nucl. Instr. and Meth. B51 (1990) 122. [7] J.H. Barrett, Phys. Rev. B3 (1971) 1527. [8] J. Lindhard, K. Dan. Vidensk. Selsk. Mat.-Fys. Medd. 34 (1965) no. 14. [9] D.V. Morgan and D. Van Vliet, Radiat. Eff. 5 (1970) 157. [lo] D.V. Morgan and D. Van Vliet, Radiat. Eff. 8 (1971) 51. [ll] D. Van Vliet, in: Channeling, ed. D.V. Morgan (Wiley, London, 1973) chap. 2. p. 52.
~~anneling
and Methods
in Physics Research
1061
(1991) 1061-1066
effect of P implantation
Ning Yu and Wei-Kan
in Si( 100)
Chu
Texas Center for Superconductioity. and Department
Bijoy Patnaik
B59/60
of Ph_v.ws, Uniuer.~it.vof Ilouston. Houston, TX
77?04, O’SA
and Nalin Parikh
D~~urt~e~tof Physics and Astronon!v: ~njf~e~sit~of North Carolinu. Chapel Hill. NC -77514. USA
Sean Corcoran
and Charles
Kirschbaum
Charles Eoans And Associates, Redwood C’1t.v.CA 94063. USA
Kody Varahramyan IBM Corporatron, Essex Junction, VT 05452 USA
With the objective of obtaining shallow junctions, we have made a comparative study of the distribution profiles of phosphorus ions implanted into Si(100) at 5 to 200 keV. along different crystalline directions. viz. the [OOl] axis, the (310) plane with 7O tilt away from [OOl]. and a random equivalent direction, defined by 5.2O tilt. and 7’ rotation from the (100) plane. The P implants were also performed in amorphous Si, and through a thin SiOz layer at the same energies for comparison. The P depth profiles in Si were measured by SIMS with 14.5 keV Cs ions. The damage profiles in SitlOO) crystals were studied by RBS/channeling of 2 MeV He ions. The role of the implants along the selected orientations is discussed in light of the resulting depth of the tails of the implanted dopants distribution.
1, Introduction Ion implantation in single crystals is often unintentionally accompanied by the channeling effect either due to misalignment of the crystals or due to the “feed-in” process during the slowing down of ions in crystals, producing a long tail of implanted dopants. Some efforts have been made for boron implants in Si(100) to study the channeling tail as functions of tilt and rotation angles of the crystals, both from theoretical calculation [1,2] and experiment [1,3]. Normally there are two different approaches to minimize the channeling tail. One is by preamorphizing the crystal through Si or Ge ion implants [4]. An alternative is to carefully align the crystal in a “random equivalent” direction prior to the dopant implantations. Cho et al. have developed a criterion to sort out the random equivalent directions for Si(100) wafers along which the B implantations give shallower profiles, as compared to implantations with a tilt angle of 7O regardless of the rotation angle [l]. Similarly to B in Si(lOO), the angular dependence of the P doping profiles in Si has not been systematically investigated [5,6]. Recently, we have applied the same 016&583X/91/$03.50
Publishers
channeling criterion to P implantations in Si(100) crystals to determine the random equivalent directions. The implantations of 3*P have been made along the [OOl] axis, the (310) plane with 7’ tilt away from [OOl]. the random equivalent direction. and through amorphous Si and through thin SiOz layers (with thicknesses of 160 and 290 A). The SIMS and RBS/channeling measurements were carried out to determine the P distribution and the damage profiles in Si, respectively.
2. Calculation The existence of the critical angles for axial and planar channeling effects were demonstrated quite well by RBS/channeling measurements of light energetic ions on a variety of single crystals [7]. Simple formulae for these angles were obtained from Lindhard’s channeling theory [8-111 and Barrett’s formulae through computer simulation of the channeling effect [7]. Based on those formulae and the analytic method employed by Cho et al. [l] for boron implants in Si(100). the critical angles of P in Si(100) for different axial and planar channeling were calculated. A tilt-
B.V. (North-Holland)
VIII. SEMICONDUCTORS
N. Yu et al. / Channeling effect of P implantation in Si(100)
1062
rotation map was constructed with the channeling regions extended from the corresponding axial channeling points or planar channeling lines into circular area or bands. The radius of the circle or the width of the band represents the amount of critical angle at different implant energies. The channeling maps for P implanted in Si(100) at 10, 25, 50 and 100 keV are plotted in fig. 1 with the unchanneling regions darkened. The coverage of the map with channeling regions at 10 keV in fig. la indicates that, based on our analysis, there are no random equivalent directions for implantation. At 25 keV, several small darkened islands show up in fig. lb. The darkened area, corresponding to the regions where the channeling effect is minimized, increases with increasing implantation energy, as seen from figs. lb to Id. We could choose the center of one of the darkened regions close to the (100) plane with a tilt of 5.2’ and rotation of 7” as the random equivalent direction for all implants at different energies.
3. Experimental Si(lO0) wafers, B-doped to a resistivity of 14-22 Dcm, were used for all the implantations. Some of the Si(100) wafers were oxidized at 95O’C in dry 0, (with 4.5% HCl) ambient to grow 160 A and 290 A SiO,. For P implantations in Si(100) crystals, we first aligned the crystal with respect to the incident 200 keV He beam along the specific directions, viz. the [OOl] axis, the (310) plane with 7O tilt away from [OOl], and the random equivalent direction. This was accomplished by measuring backscattering yield versus the tilt and rotation angle of a two-dimensional goniometer. The alignment along the (310) plane or random equivalent direction was obtained by determining the coordinates of the [OOl] axis, the (100) and (310) planes, and adjusting the goniometer correspondingly. For P implants through thin SiO, layers, the [OOl] direction of Si(100) substrates was also aligned along the beam direction. For P implantations in a-Si, the
(b) 25 keV
(a) 10 keV
/
0 rb ‘0
G
,
,
[0011 Tilt Angle
2
4
6
Tilt Angle
(degree)
8
,$a
8
.
-0
10
(degree)
(c) 50 keV
. [OOll
2
4
Tilt Fig. 1. Tilt-rotation
6
Angle
8
10
(degree)
maps of P ion channeling
[OOll
2
4
6
Tilt Angle
8
-0
10
(degree)
in Si(100) at (a) 10, (b) 25, (c) 50 and (d) 100 keV, showing the channeling effect is minimized.
the darkened
regions
where
N. Yu et al. / Channeling effect of P implantaiion
1063
in Si(lO0)
Table 1 The energies, doses [atoms/cm21 and alignments of P implantation in Si
W11 (310) Random a-%
10 keV Pz
25 keV P2
25 keV P
50 keV P
100 keV P
200 keV P
1 x 10’4 1 x 10’4 1 x 10’4 _
4x10’4 4 x 10’4 4x10’4 4x10’4
5x 5x 5x 5x
1.1 x 1.1 x 1.1 x 1.1 x
9x10’4 9x10’4 9 x 10’4 9x10’4
1.5x10’5 1.5x10’5 1.5 x 10’5 1.5 x 10’5
10’4 10’4 10’4 10’4
Si(lO0) wafers were preamorphized to a depth of 2500 A by Ar ion implantation at 50 and 200 keV with a dose of 2 x 10r5/cm2 (for P ion implant at 200 keV, a Si wafer preamorphized by Ar ion implantation up to 400 keV was used). After alignment or preamorphization. the P ion beam was tuned on within the same ion source, and implanted into the aligned Si samples. The P ions were implanted at 25, 50, 100 and 200 keV with doses varying between 1 x lOI and 1.5 x 1Or5 atoms/cm*. Molecular P2 ions
10’5 10’5 10’5 10’5
at 10 and 25 keV (equivalent to 5 and 12.5 keV of atomic P ions) were also implanted for the purposes of comparison. The implantation parameters are listed in table 1. An X-Y angular scanner was used to provide uniform implantation across the implanted region. The angular variation from the edge to the center of the implanted area was about 0.13’, estimated from the distance (2.7 m) of the scanner to the sample, and the dimension (12 mm) of the implanted area. In order to obtain the P concentration depth pro-
ll.“I”.‘I.‘l
[ool] Implant
~
- - -(310)Implant
,.g- 1020
o
Random Implant
& 4 1019
In a-S
8 ‘Z $ 1018
5x1014at./em
25 keV P
5 x 0 101’ u & 1016 0
500 Depth
” 1000 (A)
0
1500
500
1000
1500 20000 2500 Depth (A)
0
~
m-lo2
-
-(310)Implant Random Implant
100
8 ._ z 101 I: B B 5 lo1
3500
[OOl] Implant o
>-&lo1 J
3000
9xlOl
keV
P
4at./cmi
b
6
lo1 0
0.1
0.2
0.4 Deptk3(pm)
0.5
0.6
0
0.8
0.2
Keith
1
&f
Fig. 2. P profiles in Si(100) from SIMS analysis, implanted along the [OOl]axis, the (310) plane with 7 o tilt away from the [OOl], the random equivalent direction and in a-Si; at (a) 10 keV P2. (b) 25 keV P, (c) 50 keV P and (d) 100 keV P. VIII. SEMICONDUCTORS
files. secondary ion mass spectrometry (SIMS) analysis was performed by a CAMECA IMS-3f using 14.5 keV Cs ion bombardment and negative secondary ion detection. The implantation damage distribution of Si(100) has been analyzed by Rutherford backscattering spectrometry (RBS)/~hanneling of 2 MeV He ions. The grazing incidence geometry along the [llO] crystal axis was employed to enhance the depth resolution for analyzing the samples implanted at energies lower than 100 keV. while the normal geometry along [OOl] was used for samples implanted at higher energies. The RBS detector was placed at 145” with respect to the incident He beam.
~
The P profiles in Si from SIMS measurements are plotted in figs. 2a-2d separately, for different implantation energies. In each subfigure, the profiles of P implanted along the [OOI] axis, the (310) plane (with 7O off [OOl]). random equivalent direction (with S.2* tilt and 7O rotati~~n) of Si(100) crystal. and in a-% are plotted together for comparison. Overall. from figs. 2a-2d, it is obvious that the P profiles along the (310) plane direction are almost identical to those along a random direction, within SIMS accuracy. This feature could be explained in terms of the damage created by the nuclear energy deposition of ion i~lplantation. Because the nuclear energy deposited by P inlplantation is much higher than that by B implantation, the channeling effects of the planes with high Miller indices such as (310) and (210) could consequently be obliterated very quickly. It may also be noted that the differences in the P profiles between the samples implanted along the [OOI] direction and those implanted along a random equivalent direction are very large for incident energies higher than 25 keV. When the implant energy is reduced from ZOOto 12.5 keV. the ratio of the depth (for a concentration level of 1 X lOI atoms/cm”) of the P distribution resulting from the [OOl] implantation to that from the random implantation is decreased from 1.9 to 1.1. On closer examination. we can distinguish among the three distinct energy regimes in which the differences between the P profiles from the [OOI] direction implantati~~ll and the random direction implantation do not hehave in the same manner. For incident energies higher than 50 keV. a small shift in the position of the concentration peak could be discerned, in addition to the large differences in the respective tails. With implantation energies ranging from 12.5 to 50 keV, there remains only the differences of the two tails, while the peak positions of the two P profiles coincide. For energies lower than 12.5 keV, even the difference between the two tails vanishes. This is consistent with the notion
Random Implant
*
Thru 160 8, Si02
-----Thru
290.& Si02
IO’ 0
4. Results
[OOI] Implant D
0.1
0.2
0.4
0.5
0.6
Fig. 3. P profiles for implants through 160 A and 290 A SKI, layers along the Si [OOl] axis at 50 keV, along with those for implants along the (OOl] axis and the random equivalent direction without the Si02 layer. and in a-Si.
that the channeling effect is enhanced by lowering the implantation energy (as is evident from fig. la). When the implantation energy is lowered. the critical angle for channeling becomes large enough so that the random equivalent directions no longer exist. The SIMS analysis was performed on samples of SiO,/Si(lOO) implanted by P ions without stripping off the SK& layers. Fig. 3 gives the P profiles for implantation along the [OOl] direction through 160 A or 290 A SiO, layers into Si(lO0) at 50 keV. referred to those for implants in a-Si, along the [OOl] axis and along the random equivalent direction. The P profile for implantation through 160 ,& SiO, is very similar to that for implantation along the random equivalent direction, without the oxide layer. The implantation through 290 A SiO, gives a shallower profile. This is mainly caused by the increase of the angle divergence of ions passing through this thicker amorphous layer, which reduces the channeling effect. Thus. in the sense of reducing the channeling tail of the P distribution at 50 keV implantation, a 160 A SiO, thin layer at the surface of a Si(100) crystal has an effect equivalent to aligning the single crystal along the random equivalent direction. The damage profiles in the Si crystals. caused by P implants under the experimental conditions. were obtained by the RBS/channeiing analysis with a 2 MeV He beam and are displayed in figs. 4a-4d. These correspond to the P profiles shown in figs. 2a-2d, respectively. As can be seen, the damage profiles for implantation along the channeling and the random directions are almost identical at implant energies lower than or equal to 50 keV. However, for a P implant ion energy reaching 100 keV. a large shift of about 220 A of the damage distribution towards greater depths is observed for the [OOl] implants relative to that for the random
1065
N. Yu ei al. / Channeling effect of P i~l~la~fation in Sr{fOOt
2 MeV He - Si impl. by 25 keV P
0
900
800
1000 1100 E (keV)
1200
250 2 MeV He+
- Si
-I
./cd
II .Y sr ~1500
~e,,thiA,-
.E $ s m1000
:
%
2
900
1000 1100 E (keV)
.i 1200
1 Implsnt
-
-
1200
1300
2 MeV He+ _ Si impl. by 100 keV
P
-Virgin [OOl]
j: M 1500 .5 % Z m 1000 u %
5: Y u @a 500
800
1000 1100 E (keV)
2000
: : : : : I : : : : 0I loo0 dil
-.-,-‘---,-,-..-l-,~i,~
900
--[@II
keV P
2000
0
800
1300
1300
500
---01’ ’ 800
s
’
-__ r-r ._I ’ * ’ r. j x .-+900
1000
1100 E (keV)
Fig. 4. R3S/channeling spectra of 2 MeV He ions channeled along the [llO] axis of Si(100) wafers implanted [OOI]axis and the random equivalent direction at (a) 10 keV Pz, (b) 25 keV P and (c) 50 keV P: and the spectrum
[OOl] axis for implantation
implants. These features are consistent with the SIMS results, confirming that it is difficult to get rid of the channeiing effect for low energy implants of P in Si(100) because of the opening up of the channeling region in all directions.
5. Conclusions For P implantation at 25-200 keV in Si(lOO), a pre-alignment of the Si crystal along the random equivalent direction (with a tilt of 5.2” and rotation of 7 o ) results in a large decrease in the tail of the P profiles as well as in crystal damage, as compared to those when the crystal is aligned along the [OOlJ axis. However, the channeling effect could not be avoided for implant energies lower than 25 keV. This is due to the disappearance of regions corresponding to absence of channeling in the tilt-rotation map. In other words, there is
~
.
’
,zbs”
1---T
1200
1300
by P ions along the channeled along the
with (d) 100 keV P.
no random equivalent direction to be chosen at these energies. For heavy ion implants such as P. planes with high Miller indices such as (310) no longer affect the channeling tail in the P profile in Si(100). This feature is totally different from that of light ion implant such as B, which causes significantly less damage in Si. As far as the reduction of the tail of the P distribution in Si(100) is concerned, a 160 A SiO, thin layer has the same effect as pre-aligning the Si crystal along the random equivalent direction. at a 50 keV implantation of P ions.
Acknowledgements This work was mainly supported by IBM Corporation. The RBS/channeling analysis was done at University of North Carolina at Chapel Hill. A small part of the SIMS analysis was carried out at North Carolina VIII. SEMICONDUCTORS
1066
N. Yu et al. / Channeling effect of P implantation in Si(IO0)
State University. This work is also supported by the Texas Center for Superconductivity at the University of Houston which is supported through grants from DARPA, the State of Texas and private foundations.
References [l] K. Cho, W.R. Allen, T.G. J.J. Wortman, Nucl. Instr. [2] K.W. Brannon and R.F. Meeting, San Diego, CA, [3] A.E. Michel, R.H. Kastl, J.A. Gardner, Appl. Phys.
Finstad, W.K. Chu, J. Liu and and Meth. B7/8 (1985) 265. Lever, Electrochem. Sot. Fall 1986. S.R. Mader, B.J. Masters and Lett. 44 (1984) 404.
[4] M.C. Ozturk, J.J. Wortman, CM. Osbum, A. Ajmera, G.A. Rozgonyi, E. Frey, W.K. Chu and C. Lee, IEEE Trans. Electron. Devices ED-35 (1988) 659. [5] R.J. Schreutelkamp. F.W. Saris, J.F.M. Westendorp. R.E. Kaim, G.B. Odlum and K.T.F. Janssen, Mater. Sci. Eng. B2 (1989) 139. [6] C. Bresolin, C. Zaccherini, M. Anderle and R. Canteri, Nucl. Instr. and Meth. B51 (1990) 122. [7] J.H. Barrett, Phys. Rev. B3 (1971) 1527. [8] J. Lindhard, K. Dan. Vidensk. Selsk. Mat.-Fys. Medd. 34 (1965) no. 14. [9] D.V. Morgan and D. Van Vliet, Radiat. Eff. 5 (1970) 157. [lo] D.V. Morgan and D. Van Vliet, Radiat. Eff. 8 (1971) 51. [ll] D. Van Vliet, in: Channeling, ed. D.V. Morgan (Wiley, London, 1973) chap. 2. p. 52.