Sheet resistance uniformity monitoring of a medium-current ion implanter

Sheet resistance uniformity monitoring of a medium-current ion implanter

Nuclear Instruments and Methods North-Holland, Amsterdam in Physics Research B37/38 (1989) 361-364 361 SHEET RESISTANCE UNIFORMITY MONITORING OF...

379KB Sizes 0 Downloads 40 Views

Nuclear Instruments and Methods North-Holland, Amsterdam

in Physics

Research

B37/38

(1989) 361-364

361

SHEET RESISTANCE UNIFORMITY MONITORING OF A MEDIUM-CURRENT ION IMPLANTER

Donald Eaton

R. ZRUDSKY

Corporation,

* and

Semiconductor

Robert

B. SIMONTON

Equipment

Diorsion,

2433

Ruhnd

Drive,

Austin,

TX 78758,

USA

Sheet resistivity measurements on 100 mm (100) silicon, implanted at 100 keV, for B+, As+ and P+ as a function of screen oxide thickness and twist orientation are presented. We identify the optimum orientation and screen oxide thickness for these implants to avoid variations in the measurement from channeling and dopant capture by the oxide. For a given species and energy there is an optimum oxide thickness range to monitor implanter uniformity performance; thinner oxides will degrade uniformity due to increased channeling effects while thicker oxides will degrade uniformity due to variable dose capture in the nonuniform oxide film. For moderate to high energy implants there is a useful optimum screen oxide thickness which, in conjunction with wafer orientation, will adequately minimize these channeling and oxide-induced sheet resistivity non-uniformities.

1. Introduction A modem production

2. Experimental method electrostatically

ion implanter

(“twist”)

control

implant.

However,

is capable

the crystal

of achieving

across

Appropriate

of the scanned

sheet resis-

in the substrate,

from

ion beam

to

due to channeling

ing variations.

Screen

oxide modifies

attack

of an incident

attack

angles at the oxide-crystal

scanned

This strate formity

article

presents

implants

orientation

so that revealed.

ate sheet effects

data

which

surface

of

thus reducing

The monitor those

oxide

thickness

and

of suband uni-

This information

an implant

process

hardware

varia-

performance

is

must be able to differentiby

caused

by process

implanter

hardware

malfunction.

its (100)

caused

at the University

30”

Duluth,

MN 55812,

Physics

0 Elsevier Publishing

Science

Publishers

Division)

B.V.

(1 x

with fixed

10”

tilt

for phosphorus)

uniformities

[%, lo]

Sheet

for

each

with the twist angle used to implant planes

being

so that O” results

aligned

of the implanter;

implanted

with the verti-

positive,

rotation

surface

scanning

increas-

of the wafer

normal.

The

varia-

tilt setting,

of the ion beam,

are

in fig. 1 for a 100 mm wafer with a 7O tilt (a

twist is also noted

and twist variation

in this figure).

is reduced

The range

with increased

of tilt

tilt angle

values. from

test: wafers with oxide

100 to 6000

phorus

and arsenic

optimum

twist

1 X lOi phorus

at both 0”

boron

used 30°

thickness

A were implanted

settings.

cm-2;

determined

twist

All

and 7”

were

used 23”

and arsenic

by the previous plotted

against

during

the implantation.

thickness

and uniformity

was

independent

of

keV,

twist,

used 45’

and the sheet resistance

oxide

using

100

phos-

twist,

as

test. The oxide was stripped

after implantation were

phos-

tilt angles

implants

implants

ranging

with boron,

uniformities

ples

0168-583X/89/$03.50 (North-Holland

of Minnesota,

and

by electrostatic

present * Presently USA.

[llO]

In a second

monitor

and process

nonuniformities caused

arsenic

and arsenic,

mm silicon

tions of tilt and twist angles about a nominal

the degree

the effect

measurement

the implanter

resistance from

boron,

uniformity.

whether

control

for

(1 x 1Or4 cme2)

ing twist angles result in clockwise about

100

phosphorus

The twist angle is defined

cal scan coordinate

by small-angle

(100)

twist angle (O-45 o ) were performed.

averages

these channel-

the single angle of

examine

and screen

on sheet resistance

can adequately clearly

resistance

visualized

can help determine tions

(7O for boron

and variable

bare

(1 x 1013 cmm2),

and arsenic

in the wafer’s

of channeling. phosphorus

angle

into

boron

with an over-

ion beam into a cone

ions,

1013cm-2)

the wafer.

oxide can serve to reduce

implants

using

wafer were plotted

the wafer [l]. tilt and twist orientation

of the dopant

100 keV wafers

a very uniform

layer of screen

scattering

current

can result in many implants

angle of attack planes

variations

medium

tilt and rotation

when using bare wafers,

tance nonuniformities the variable

scanned with wafer

the

averages

oxide

The magnitude variations

thickness

for

and

thickness of the

in the samthe

range

IV. ION IMPLANTATION

of

D. R. Zmdsky,

362

TWIST

ANGLE

100

VARIATION

MM

R. B. Simonion

(Ati)

WAFER

Fig. 1. Variation of tilt and twist angles from nominal settings (7 o tilt, 30 o twist) during scanning for a 100 mm wafer.

100-6000

A; a frequency

in angstroms, scribed

for

by

frequency

a

Poisson

at 4-6

All implants

distribution

ion implanter

digital,

all solid-state

Fig. 3. Boron implant average sheet resistivity (lower plot) and uniformity (upper plot) vs screen oxide thickness.

deviation,

was with

well

de-

maximum

on an Eaton [2] using

electrostatic

of tilt gradient

modeled

as purely

sic diffusion

at 1150 CJC/20

NV-6200

scanner

boron

at 1150 CJC/10

and arsenic

at 1125 o C/10

developed

to predict

3. Results

sheet

model

resistance

(1) a simple

of these implants.

Gaussian

tion using ranges (2) annealing methods,

was

for silicon

redistribution

as appropriate

closed-form

reflection

of the diffusion

The model

(two-moment)

and its oxide determined

(normalized solution

differential

implant

equation),

featured

to identify

the onset

effects

deviations

rela-

function were

from

not

model

of uncontrolled

chan-

and discussion

The sheet resistance rotation

from ref. [5],

by one of three error

as an inverse

Channeling allowing

intrin-

Arrhenius

the

distribu-

function,

or a numerical

from

neling.

S.

A computer

mobility

in the model,

predictions

(4) simple,

barriers,

calculated

concentration.

included

featur-

reflective

constants

and (5) carrier

of carrier

an extremely

dose errors [3,4]. Rapid

was used for all species,

s, phosphorus

loo0 2ooo 3000 0 SCREEN OXIDE THICKNESS(A)

tionships

were performed

ing compensation annealing

thicknesses

A.

medium-current linear,

plot of the standard

all oxide

uniformity monitoring

/ Sheet resistance

a

solution

(3) oxide surfaces

gives

(twist)

the boron

arsenic.

All

minimum uniformity The

plots

of

uniformity

angle data, these

between

of average

in fig. 2;

fig. 2b phosphorus figures

20”

for boron

plots as a function

are presented

and

show 30°

and arsenic sheet

a local rotation;

and

(not

fig. 2c

uniformity very

also appears

resistance

of

fig. 2a

good

at 45 O.

shown)

re-

piGq 3

7’ TILT

2 i

;]I+&9-y--j 0

0

10

Fig. 2. Sheet resistivity

20

30

40

10 ROTATION

20

30

40

(DEGREES)

uniformities vs twist angle: (a) Boron implant (100 keV, 1 X lOI cm-‘, 7 o tilt), (b) Phosphorus keV, 1 x lOI cme2, 10 o tilt), (c) Arsenic implant (100 keV, 1 X lOI cm-‘, 7 o tilt).

implant

(100

D.R. Zmdsky, R.B. Simonton

vealed

that

curred

at 4S”

the

ity minimum indicates

highest

value

for all species; significant

at the highest channeling

the entire range of rotations. in good agreement

resistance

oxide

from

average

value over

The data from this test are

with previous

boron

thicknesses

plot (average data points

reports

[l].

through

and uniformvarying

screen

for both 7O and O” tilt. In the lower

sheet resistance).the thickness

gradual

prediction

values

loss of control

havior

divergence

of

(the solid curve)

is an indication

over channeling.

The

of the same be-

at a higher screen oxide thickness value 0 1000 A) for the 0” tilt data. This is expected at

tilt because

control

effect

of the removal

by Myers

boron

implants

adequately

of the axial

of the 7 o tilt angle;

report

it is consistent

et al. [6], that (150

keV)

controlled

channeling

channeling

at 0”

with a

effects

for

were

not

in silicon

unless the screen oxide was greater

The

upper

uniformity

plot

versus

as the lower much

larger

implants,

of fig. 3 gives oxide

plots.

Here

effects

low thickness increase

tilt implants

than those

be expected

from

all yield

the greater

axial

at 0 O. The larger non-uniformities

values

are due to channeling,

of the nonuniformity

nesses

is due to oxide

effects

yield

with larger

thickness

a uniformity

while oxide

variations;

minimum

at the

thick-

these

two

at 1000-1500

A.

The solid curve on the right of the upper plot is a model prediction

of the nonuniformity

sources;

&lo

A (lo)

apparent

oxide

arising

oxide

thickness

contribution

uniformity

changes

from

variations

as seen

from its angle of attack

two and

by the ion

variation

across

a

100 mm wafer. Data screen

from oxide

manner

the

to the boron

the lower

plot

from

sheet resistance identifies earlier

arsenic

thicknesses

the

onset

the model of

for boron,

as expected

be expected

and straggle

at a screen

than arsenic

oxide

but less than

by the intermediate

posi-

have at 100 keV relative

to

those two species. Since

these

results

selecting

an

computer

model

of

screen

wafer energy

screen

thickness

nonuniformity

to

in sheet resistivity

thickness,

maximum

achieve

variations

no

across

for all three

a

an upper

tions:

f 10 A and (2)

(1)

uniformity

and lower

variations.

variation

due

to

variations

over the wafer

The resultant

the

mm

over the were used

oxide

apparent

scanned

curves

than

limit for these calcula-

f 20 A (la)

The

values

100

species

of the

more

range of 10 to 200 keV. Two criteria

to create

cases.

the importance oxide

was used to generate

oxide

*0.25%

demonstrate

appropriate

beam

surface

thickness

oxide

thickness

angle

of

was included

are shown

attack in both

in fig. 6.

oxides

channeling; relative

(SO-100

the

to the 7O

in the upper

for arsenic

much lower screen oxide thickness

as would

minimum

A) greater

of average

Also indicated

uniformity

a uniformity

(300-400

varying

for very thin screen

uncontrolled

is again observed.

figure indicate

of the data in

prediction

for the 0 o implants

plot is that the best

through

800

in fig. 4 in similar

data. The departure

(solid curve)

departure

implants

implants

are plotted

6bO

SCREEN OXIDE THKZKNESS (A,

tion its range

of the 7O tilt

400

Fig. 4. Arsenic implant average sheet resistivity (lower plot) and uniformity (upper plot) vs screen oxide thickness

boron,

resistivity

for the same samples

the 0”

non-uniformities

as would

channeling

thickness

the sheet

1400

1OOCI

,

200

thickness

than 1000 A.

beam

ii’

0

occurs

(about O”

implants

from the model

at low oxide gradual

363

oc-

is occurring

Shown in fig. 3 are the sheet resistance ity data

monitoring

the fact that the uniform-

did not occur

that

of sheet

/ Sheet resistance uniformity

occurs

at a

A) than that

from its much shallower

range at

100 keV. Data

from

ing oxide

in the previous uncontrolled indicated

the phosphorous

thicknesses

implants

are presented

figures.

Again

channeling

by the departure

at

through

vary-

in fig. 5, plotted

as

in fig. 5 is the onset

of

thin

screen

oxide

values

of the data from the average

sheet resistance

model prediction

sheet resistance

uniformity

in the lower plot. The

data in the upper plot of the

0

400

800

1200

1800

SCREEN OXIDE THICKNESS

2000

300

6,

Fig. 5. Phosphorus implant average sheet resistivity (lower plot) and uniformity (upper plot) vs screen oxide thickness. IV. ION IMPLANTATION

D.R. Zrudsky, R.B. Simonton / Sheet resistance uniformity monitoring

364

of the species, occurring at higher energies masses. Therefore, to eliminate channeling low energies, species

% g

500:

=.

300-

8

100:

E k?

50

Sheet

resistivity

optimum

30

8

i

twist

Fig. 6. Model-generated

curves

of maximum

oxide

thickness

due to oxide varia-

curves for lo oxide thickness

A. Top curves for lo oxide thickness varia-

and

energy

The computer

model was also used to generate

of

screen

no more

sheet resistance they

keV.

than

These

that the oxide

oxide

thickness

+0.5%

variations

are nearly

curves

selection

wafer-to-wafer

uniformity

criteria

and f 0.5% average

for the indicated

These implant tures

from

behavior this

theoretical

have

work.

serious

It

only

been

for low energies

implant

energies

channeling oxide

control

required

because

dependent much

energy pends

region

to control uniformity

become

oxides

ineffective

control

layer and

The

specific

are inapplicable implanted,

tilt and twist

[8]; again,

for loss of channeling

angle of

is primarily

with the lower

energy.

for higher mass species. energies

(fig. (figs.

[7], so the effective-

degrades

on the mass of the species

at low implant

effects

oxide

for

of screen

dose capture

of the screening energy

in

more

is dependent

de-

occurring

Furthermore,

orientation

the specific

channeling

control.

At

control

channeling,

of a neutral implant

results

which

and energy,

channeling

to

optimum is that

well enough (I

Thicker

lo)

oxides

for

will de-

capture

the uniformity implant

to

0.5%

and

due to loss of energies

and orientation

screen

is ineffective

so the use of amorphizing

species prior to the electrically

to

implants

active species’

is recommended.

These when

that, for moderate

uniformity),

low

are inapplicable

the im-

is revealed.

due to nonuniform

oxides will degrade

oxides

so that

species

measurements.

the uniformity

thickness to control

is a predictable

(and

obtain

resistivity

oxide

effects

on implant

can be employed sheet

arthe

performance

there

thickness

excellent

with

screen

data indicate

dependent

boron,

which identify

considerations

designing

planter

must

sheet

dose repeatability

difficult

to determine

be

resistance

taken

into

monitors

and uniformity,

whether

an unacceptable

trend in

is due to ion implanter

ware

process

or other

when the ion implant ned to control allows

monitor

more effective

effects.

hard-

Furthermore,

has been correctly

measurement

and process

diagnosis

im-

or it will be

dose or dose uniformity problems

account

for ion

desig-

variations,

it

if there is an implanter

malfunction.

ions. At low

channeling

by decreasing

where screen

at higher energies

is far

in the solid scattering

layers rapidly

required

resistance

are inapplicable

by the screen

on the thickness

ness of screening

depar-

the lower thickness

control

ion beam

for all

at 100 keV

channeling

layers

less on the ion beam

thickness

sheet

to avoid unacceptable

The value achieved

the incident

* 0.25%

of oxide

and for heavy-mass

6) does not adequately 3-5).

values

characterized

screening

use

repeata-

channeling-induced

that

but

of oxide variations.

(unchanneled)

is known

both

sheet resistivity

magnitude

because

here,

in fig. 5, indicating

in fig. 5 will permit

are not the optimum conditions,

from 10

are not presented to those

to

average

for the three species

identical

maxi-

variation

of the same value of screen oxide to obtain bility

a neutral

of this study)

capture

implants,

oxide

strongly

thinner

values

oxide

dose and uniformity

grade

tions of + 10 A.

and

channeling

screen

nonuniformity

tions on 100 mm wafers. Bottom of k20

orientation

The experimental

for f 0.25% sheet resistance

for implants

were presented

(for the conditions

high ENERGY (keV)

to 200

data

parameters planter

‘“ikT7XX&0

achieve

with

4. Conclusions

senic and phosphorus

variations

implant

[l].

10007

::z

n

mum

an amorphizing

is recommended

for higher effects at

energy

also region

on the mass

References [l] N.L.

Turner,

Simonton, [2] Eaton 2433

M.1. Current,

Proc. SPIE

Corporation, Rutland

T.C.

530 (1985)

Semiconductor

[5] J.F.

Division,

USA.

Nucl. Instr. and Meth. B21

and D.D. Myron, Nucl. Instr. and Meth. B26

614.

Ziegler,

Ranges

and R.

410.

[4] D.R. Zrudsky (1987)

D. Crane

Equipment

Drive, Austin, TX 78758,

[3] D.D. Myron and D.R. Zrudsky, (1987)

Smith, 55.

J.P.

of Ions

Biersack in Matter

and U. Littmark, (Pergamon

The Stopping

Press,

New

York,

1985). [6] D.R. Myers, (1981)

J. Comas

and R.G. Wilson,

J. Appl. Phys. 52

3357.

[7] L. Meyers,

Phys. Status Solidi 44 (1971)

253.

[8] K. Cho et al., Nucl. Instr. and Meth. B7/8

(1985)

265.