Channeled implantation with a parallel scanned ion beam

Channeled implantation with a parallel scanned ion beam

Nuclear Instruments and Methods North-Holland, Amsterdam CHANNELED J.F.M. IMPLANTATION WESTENDORP, R. SCHREUTELKAMP FOM-Instiiute K.T.F. in Phys...

370KB Sizes 0 Downloads 44 Views

Nuclear Instruments and Methods North-Holland, Amsterdam

CHANNELED J.F.M.

IMPLANTATION

WESTENDORP,

R. SCHREUTELKAMP FOM-Instiiute

K.T.F.

in Physics

Research

B37/38

WITH A PARALLEL

R.E. KAIM

357

(1989) 357-360

SCANNED

ION BEAM

and C.B. ODLUM

and F.W. SARIS

for Atomic and Molecular Physics, Kruislaan 407, 1098 SJ Amsterdam,

The Netherlands

JANSSEN

Direct axial channeling, with the major crystal axis a!igned with the incident ion beam, can produce deep penetration of dopants at modest ion energies. In addition, damage levels are significantly reduced. When the ion beam is scanned parallel over the wafer, this deep penetration can be achieved at every location on the wafer, provided the change of the angle of incidence across the wafer is smail compared to the critical angle for channeling. Also, if implants in a random direction are done, the implant will be random all across the wafer, resulting in a homogeneous dopant depth profile. Si wafers of up to 150 mm have been impianted under channeling and random conditions. Dopant and lattice damage profiles were measured with Rutherford backscattering and SIMS. The reproducibility and applicability of the axial channeling implant technique are discussed.

1. Introduction One of the principal requirements of ion implantation is that the implant be uniform in dose, with VLSI processing trends imposing stringent limits on allowable dose variation across a wafer. However, an equally important parameter that characterizes an implant is the dopant depth distribution. Theory can adequately predict implantation profiles into amorphous materials [l-3], but in practice most implants in VLSI fabrication are done in single-crystalline Si and GaAs. In singlecrystal lattices there are certain directions along which incident ions have a greatly reduced probability of colliding with lattice atoms. This phenomenon, called channeling, enables ions to be steered into an open axial or planar direction and can give rise to penetration depths as much as an order of magnitude larger than for ions implanted into amorphous material. In addition, the amount of damage to the crystal lattice caused by channeled ions is much less than for ions incident in a nonch~neled direction. The possibility of reducing lattice damage, or of obtaining deeply implanted junctions at modest ion energies, makes deliberate channeling an attractive prospect for some implants. The principal requirement for successful applications of this channeled implant technique is that the degree of channeling should be uniform at all points on the wafer. The probability of an ion being channeled depends on the angle between the ion trajectory and the crystallographic direction: for 0168-583X/89/$03.50 0 Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

angles larger than the “critical angle” the probability is greatly reduced. The requirement that channeling should be uniform in a channeled implant therefore implies that the change of angle of incidence across the wafer surface be much less than the critical angle for channeling and that the alignment of the Si wafer with respect to the beam be accurate and reproducible. Ion implanters employ different methods of spreading ions uniformly over the surface of a wafer and the consequent variation of the angle of incidence depends on the type of implanter, as is illustrated in fig. 1. In an implanter where the beam is rastered electrostatically in two dimensions over a single wafer (fig. la) the variation in implant angle across a 150 mm wafer is typically 5 O. In batch implanters employing mechanical rotation of a disc or drum the wafer is tilted away from the plane normal to the rotation axis to achieve clamping of the wafer by the resulting centrifugal force. If the ion beam is also tilted with respect to the axis of rotation (fig. lb), the implant angle changes as the wafer is rotated past the beam. The side-to-side implant angle variation is dependent on the angles involved and on the ratio of the radii of the wafer and the disc, but is typically 2-3” for a 200 mm wafer. Critical channeling angles for the ions used in VLSf implantation range typically from 2O to 4O. Fig. 2 gives calculated [4] values of the critical angle for (100) axial channeling in Si as a function of energy for B, P and As ions. The variation of the implant angle across the wafer for the implanters illustrated in figs. la and lb is of the IV. ION IMPLANTATION

358

J. F. M. Westendorp et al. / Channeled implantation with a parallel scanned ion beam directions, implanted

the maximum wafer diameter that could uniformly amounted to only 38 mm.

Recently,

a new ion implanter

to the market parallel, neled

which, because

has the potential

on wafers

is a serial

trostatic/mechanical zontal

the

the beam

implant

scanned than

(c)

magnet

angle by

beam.

0.35O

than 0.5O

side-to-side

same order

of magnitude

as the critical

with these implanters

tion is not maintained of a wafer.

while

An implant

angles.

Conse-

a well channeled

condi-

scanning

performed

to axial channeling harness

the effect

As a result,

in the industry

common

practice

the incident

for axial

An attempt doping

channeling ASM-220

of seeking

to

the strategy

150 mm, n-type

which

between

uniformity

was

scan system using double

eliminates

ion implantation et al. [6]. Although

achieved

electrostatic

with

a parallel

deflection

in both

CRITlCAL ANGLE FOR AXIAL (100) CHANNELNG

at various while

the twist

on the wafer plantation

axial in the

orienta-

of energies

while for the B+ implants The implants

corresponds being

defined

and the wafer

or rota-

to the major

horizontally.

damage

profiles

backscattering

as

normal

to the azimuthal

positioned

vari-

were performed

the tilt being

the ion beam

ion-induced

flat

After

im-

were measured

and channeling

with 1

angle of 170 O. Other wafers

were annealed

at 950 o C for 30 minutes

dry nitrogen

after implant and

sheet

resistance

on a Prometrix

performed

using

profiles

were

primary

beam

rastered

profiles

were

recorded

beam rastered analysed

RS-30

measurements system.

a Cameca determined

(M/AM

SIMS

IMS with

in

were analysis

3f instrument. a 10.5

keV

0:

over 350 km X 350 pm, while P using

a 14.5

keV

Cs+

primary

over 250 pm x 250 pm. In both cases

area of 60 pm diameter

phosphorus

I

substrate

MeV He ions at a detection

Boron

DEV,AT,ON FROM PARALLELI%

of (100)

tional angle. A twist of 0 o corresponds

was

40

of uniform implanted

P+ and B+ ions

tilt and twist angles,

performed

/-

in Si (see

100 and 200 keV. The dose for the P+ implants

using Rutherford

D0

with

the angle between channeled

less

channeling

(100)

Si wafers

tion were implanted

ous doses were employed.

by Nishi

and

less than

2. Experimental

away from

to perform

the

implanter.

is to tilt the wafer normal

been reported

of

to be less

[9], i.e. much

150 mm Si wafers

was 2 X 1Ol4 ions/cm2,

by 5-7”

has been shown

we will give evidence

across

it [5]. The

direction

is only

for a 150 mm wafer

to avoid

beam

has been

due

axial channeling.

has already high

condi-

on the wafer and

instead

of direct axial channeling,

adopted

direct

these

depth variations

at some locations

less or none at others.

wafer

parallelism

over the surface

under

tions will give rise to large junction

dimension this system,

fig. 2). In this article

quently,

the

from

The deviation

angle

of a nonuni-

With

across

deviation

for a 200 mm wafer

the critical

electrostatically

by means

translated.

elec-

In the hori-

[8]. In the vertical

variation

the

chan-

as 200 mm.

with a hybrid

is scanned

is mechanically

determined

Fig. 1. Schematic representation of different ion beam scanning techniques employed in (a) an X-Y electrostatically scanned implanter, (b) a batch implanter and (c) the ASM-220.

implanter

scan is achieved

form field dipole the wafer

applying

as large

scan [7] (see fig. lc).

direction

and a parallel

introduced

its beam scan is accurately

of successfully

ion implantation

The ASM-220

has been

be

measurements

> 4000)

was

was chosen.

high

necessary

mass

to separate

For

an the

resolution “P

from

30Si1H. SO

120

160

2w

IDN ENERGY (keb’)

Fig. 2. Critical angle for axial (110) charmeling in Si as a function of ion energy, for As, P and B ions. The deviation from parallelism in the ASM-220 is shown for comparison.

3. Results Fig.

3 shows

spectra

measured

1 MeV

He Rutherford

on wafers

implanted

backscattering with 100 keV P+

J. F.M. Westendorp et al. / Channeled implantation

1OOkeV

*---_

t

DEPTH

P++Si

with a parallel scanned ion beam

359

(100).6’

(urn)

Fig. 5. SIMS depth profiles for random and channeled phosphorus implants into silicon.

B--__.c_-___-_____

Fig. 3. 1 MeV He Rutherford backscattering spectra for random aad channeled phosphorusimplants into silicon.

ions under different tilt and twist angles. The random implant was performed at a tilt angle of 10 o and a twist of 150, while in the case of the channeled implant both angles were 0 O. The difference between the two cases is striking: the random implant yields a damage profile that peaks at a depth of about 1000 A, which is consistent with a TRIM [3] calculation for a 100 keV P+ implant into amorphous Si, while for the channeled implant the bulk of the damage is produced much deeper in the substrate, with a peak at roughly 1650 A. In addition, the overall damage level is significantly reduced for the channeled implant. This implies that for the 0” implants the bulk of the implanted ions are well

channeled and are deposited at greater depths. Most importantly, however, channeling spectra measured at three locations along a line parallel to the horizontal electrostatic scan direction (see fig. lc.) show identical damage depth profiles. This means that the well channeled condition is maintained all across the wafer. Sheet resistance measurements constitute a very sensitive measure of junction depth uniformity on implanted wafers. Therefore wafers were implanted under the same channeled and random conditions as above and sheet resistance maps were recorded. The maps in fig. 4 show that the mean sheet resistance of the channeled implant (fig. 4a) is 20% lower than for the random implant (fig. 4b). However, there is no degradation of uniformity on the channeled implanted wafer of the

WNNfLEDMPVINT 2 1OOke” P .2E14 0NSlCM c MAN

=.71x = 206.6

J-L/~

Fig. 4. Sheet resistance maps for random and channeled phosphorus implants into silicon. The sheet resistance is measured on 81 sites on the wafer, (I is the standard deviation in percent. IV. ION IMPLANTATION

360

J.F.M.

Westendorp

et al. / Channeled

implantation

with a parallel

scanned

ion beam

4. Discussion The

above

results

demonstrating It has been

shown

a significant

while

yields

still

step

of channeled

that channeled

150 mm Si wafers depths

are

the feasibility

much

ion implantation larger

maintaining

towards

implantation.

good

uniformity.

Since

the parallelism

of the beam

scan

in the ASM-220

been

over

field

of 250

qualified

reason

to expect

a scan

that similar

on

ion penetration

uniformity

mm,

has

there

is

can be attained

for 200 mm wafers. For channeled cable

ion implantation

technique

and

in VLSI

wafer-to-wafer

These graphic

orientation

standard enting

wafers)

have

surface

perpendicular

study of the effects

conducted.

reproducible the

from wafer

dependence

measurements about dopant profiling performed Fig. 6. SIMS depth

profiles for random and channeled boron implants into silicon.

wafers.

type

expect

References

one

contour

would

Finally,

Si wafers

ions under identical as above

profiles

by SIMS.

observed. plants

penetration

depth along

B+ (b).

From

profiles,

depth

species for

for the chan-

Fig.

6 shows

with 2 x 1015 ions/cm’ at

three

are identical.

channeled

condition

This

again

is maintained

these of 200

that SIMS

different

to the horizontal

locations

electrostatic

means across

is im-

B+ implant

fig. 6b it is also evident

direction,

profile

for channeled

5 X 1015 ions/cm*

measured

a line parallel

conditions

tail in the depth

B + ions.

with

for a 100 keV,

(a) and for an implant keV

P+ and B+

of the implanted

tail is not observed

performed

profiles

with

and random

Fig. 5 shows depth profiles

a profound

This

(i.e.

on

results

been

do not provide depth profiles.

measurements on

channeled

are

ion

in ori-

beam.

look

very prom-

therefore

because

sheet

unambiguous Additional

for

to be extremely

However,

mobility,

and

An

is now

and uniformity

found

to wafer.

carrier

for SEMI

information

SIMS

and C-V

presently

randomly

of

resistance

being

implanted

flat).

with a dose of 2 X 1014 ions/cm*.

the enhanced implant,

in parallelism

were implanted

a 100 keV P+ implant Besides

errors

to the major

channeled

and depth

were recorded

neled

for

lines perpendicular

have

(lo

of these tolerances

Preliminary

implants

addressed. in crystallo-

tolerances

to the

ising since the mean sheet resistance channeled

to be

by the tolerance

of the wafer

an appli-

of reproducibility

and the mechanical

the wafer

extended being

uniformity

will be determined

to become

the issues

scan

that the well the wafer.

[l] J. Lindhard, M. Scharff and H. Schiott, K. Dan. Vidensk. Selsk. Mat. Fys. Medd. 33 (1963) 1. [2] K.B. Winterbon, Ion Implantation Range and Energy Deposition Distributions, vol. 2 (Plenum, New York, 1975). [3] J.F. Ziegler, J.P. Biersack and U. Littmark, The Stopping and Ranges of Ions in Solids, vol. 1 (Pergamon Press, New York, 1984). [4] MI. Current, N.L. Turner, T.C. Smith and D. Crane, Nucl. Instr. and Meth. B6 (1985) 336. [5] K. Cho, W.R. Allen, T.G. Finstad, W.K. Chu, J. Liu and J.J. Wortman, Nucl. Instr. and Meth. B7/8 (1985) 265. [6] H. Nishi, T. Inada, T. Sakurai, T. Kaneda, T. Hisatsugu and T. Furuya, J. Appl. Phys. 49 (1978) 608. [7] ASM-220 Ion Implanter, ASM Ion Implant, Beverly, MA, USA. [8] U.S. Patent No. 4,276,477. [9] R.E. Kaim and J.F.M. Westendorp, to be published.