Non-Contact Electrical Doping Profiling
D. Marinskiy, J. Lagowski, J. D'Amico, A. Findlay, L. Jastrzebski
Semiconductor Diagnostics, Inc., Tampa, FL 33612, USA
Abstract. Monitoring of dopant concentration in the near surface region is very important in semiconductor manufacturing,
especially for epi-technology and ion implantation. Two relevant techniques which have been used are Mercury probe and Elastic
probe. Both of them allow profiling of dopant concentration by measuring the capacitance of the depletion layer, CD, versus the
applied bias. These techniques are contact in nature. In addition Mercury probe uses Hg, which may be considered undesirable in
cleanroom environment.
The method being presented is non-contact and non-destructive. A deep depletion layer is created by corona charging of the
wafer surface. This depletion layer decays to an equilibrium value due to thermal generation of minority carriers. Two transients
are simultaneously monitored during this process: 1) the small signal ac surface photovoltage that measures the depletion layer
capacitance, and 2) the contact potential difference that measures the voltage drop across the depletion layer. The set of
corresponding C-V data is used to calculate the dopant concentration profile. In silicon, the technique is applicable for dopant
concentrations in the range from Iel4to Iel8 cm"3. The probing depth is limited at the upper end by avalanche breakdown in the
semiconductor and at the lower end by the minimum surface barrier. Measurements of dopant profiles on bare and oxidized
surfaces using this technique are presented for epitaxial p/p+ and n/n+ substrates, n/p structures, and implanted wafers, covering
probing depths from 0.05 urn to 7 Jim. This non-contact technique can be realized in a simple configuration that may be of
interest for universities and research and development centers.
Knowing CD is not sufficient for determination of
doping. A voltage drop across a depletion layer
corresponding to a given capacitance must also be
known. This was overcome with a wafer treatment that
created deep inversion. The corresponding voltage
drop across the depletion layer is given by the Lindner
approximation [4]:
INTRODUCTION
Control of the dopant concentration profile in the
near surface region is very important in semiconductor
manufacturing, especially for epi-technology and ion
implantation. A variety of techniques have been
developed for measuring of dopant concentrations [1].
Two commonly used techniques are based on
capacitance-voltage
measurements
of
metalsemiconductor contacts formed with a Mercury probe
or Elastic probe.
A non-contact technique for measuring dopant
concentration near the surface was introduced by
Kamieniecki et al. [2]. In this approach, a depletion
capacitance is determined from the small ac-SPV
signal, VSPV, using a relationship derived by
Nakhmanson [3],
^
const • leff
= — [2.1 ln(N A /ni) + 2.08]»
q
(2)
where kT is the thermal energy, q is the elementary
charge, NA is the dopant concentration, and nf is the
intrinsic carrier concentration.
An extension of Kamieniecki's approach to doping
profiling was done using a ramping voltage bias
applied to a transparent capacitive SPY electrode
placed on the wafer surface. It was assumed that all
induced charge was reflected in a semiconductor as a
depletion layer charge.
This is a questionable assumption for a bare silicon
surface with a large density of surface traps. In reality,
only a portion of the induced charge should be
expected to be imaged in the semiconductor space
charge region, while the rest will be imaged in the
surface or interface traps.
n\
where const is the calibration constant, leff is the
effective photon flux, (D is the light modulation
frequency, and Im VSpy is the imaginary component of
SPY signal
CP683, Characterization and Metrology for VLSI Technology: 2003 International Conference,
edited by D. G. Seiler, A. C. Diebold, T. J. Shaffner, R. McDonald, S. Zollner, R. P. Khosla, and E. M. Secula
© 2003 American Institute of Physics 0-7354-0152-7/03/$20.00
802
The refined SPY method by Lagowski et al. [5]
involved an additional measurement of the
semiconductor surface barrier, VSB, that complemented
the measurement of CD by ac-SPV. From CD and VSB
the Near Surface Doping, NSD, was obtained as:
2-(V s b -
^--|t:
10-
——Avalanche breakdown theory
ED Experiment
— — W minimum
^SS^....:.....:,
1-
kT
""^1?
(3)
CD
Q.
0
D
"*" *••» ^
•
;;
^Y^
0.1 .
...
...
... ...; ..,: ...
...
...
;..
. ... ... .. ... L ... I. ... U
where e0 is the permittivity of free space and € is the Si
dielectric constant.
The concentration measured by this method
corresponds to an average dopant concentration over
the depletion width, W.
0.01 - 1E+15
1E+16
1E+17
1En
Concentration, [cm"3]
FIGURE 1. Probing depth in Si for dynamic C-V technique
is limited in the upper end by the avalanche breakdown in the
semiconductor and in the lower end by the minimum surface
barrier.
(4)
After a large pulse of the corona charging, the
depletion layer collapses with time from a maximum
value determined by the silicon breakdown voltage
until it reaches an equilibrium value with the voltage
drop across the depletion layer given by the Lindner
approximation. These two limiting voltages determine
Extension of this non-contact NSD technique to
profiling of dopant concentration is done using corona
charging in order to vary the depletion layer width.
PRINCIPLE OF THE METHOD
the range of depth profiling. The time constant of the
corresponding transients depends on the generation
lifetime. In Fig. 1, the maximum and minimum probing
depths are given versus dopant concentration. A
theoretical value of the upper limit is calculated using
the silicon avalanche breakdown field [6]:
The technique relies on a non-equilibrium deep
depletion transient condition that is created by charging
the surface with ions from a corona discharge in the
air. Negative corona is used for n-type and positive
corona for p-type silicon, respectively.
4-l(r
E=-
(5)
TABLE 1. Basic equations used in the dynamic CV and in standard MS - CV methods.
C-V
Dynamic CV
Measured
Vbias; CD
VCPD J VSpv
parameters
Calculated from VSPV.
Capacitance of
depletion layer,
Directly measured CD
const -I ff i
CD
0)
Depletion width,
W
Dopant
concentration,
NA(W)
£ 0 8- A
W
W
CD
SPV
S08
V
const • leff
SPV
-1
XT f\*r\
8o fcE -A
A
£
2
<*V
803
fconst'Ieffl
I
<* J
2
{(VSPV)2]
qeo6
dVCPD
The depletion layer relationship at avalanche
breakdown [6]:
Measurements on n/p Epi Wafers
An n-type epi-layer was prepared on a p-type
substrate. The substrate resistivity was between 4 - 8
Q-cm. The target epi-layer concentration was about 1.4
e!5 cm"3 and the epi thickness 4.0 |nm. The structure
has a p/n junction at the epi-substrate boundary and an
associated p/n junction depletion layer. This depletion
layer width is 0.9 Jim. The depletion width on the nside of the junction is 0.6 (im. The structure is shown
schematically in Fig. 4.
For a structure with a p/n junction, the
measurements will be limited by the junction depth
rather then the condition of avalanche breakdown
corresponding to epi-doping. A large dose of negative
corona ions deposited on the surface would deplete the
entire n-type layer before reaching the breakdown limit
of 15 |um for 1.4 ell cm"3 doping (see Fig. 1). When
the depletion layer reaches the n/p junction, the hole
injection will take place from the p-type substrate to
the epi-layer accelerating the depletion layer collapse.
This condition can be prevented by reducing the dose
of negative corona ions. Such measurements are given
in Fig. 5.
(6)
q-N A
The lower limit is obtained from the Lindner's
approximation. Experimental values were calculated
from the maximum value of the initial contact potential
difference attainable for a given dopant concentration.
If necessary, the lower limit of probing depth can
be reduced by decreasing the depletion width using
corona charging with an opposite polarity.
During a depletion layer transient, following
cessation of corona charging, two parameters are
simultaneously recorded, the contact potential
difference, VCPD, and the small signal ac-surface
photovoltage, VSPV. We refer to this technique as "the
dynamic CV method".
The equations used to calculate the dopant depth
profile (DDP) from the two measured quantities, are
summarized in Table 1, together with corresponding
equations used in the MS-CV profiling technique.
RESULTS
Measurements on p/p+ Epitaxial Wafers
A typical VSpy2 versus VCPD plot is shown in Fig. 2
for a p/p+ wafer with an epi-layer thickness of about
6.0 \im and a nominal epi-layer concentration around
1.1 el5 cm"3. The corresponding dopant depth profile
(DDP) is shown in the upper portion of Fig. 3. For a
comparison, the depth profile obtained with a Hg-probe
is shown in the lower portion of Fig. 3.
„ 1.00E+16 IT
lE^EEEEEfEEEEEEEEEi
. _
.„! _. ... .... ... _
.1 ._ ...
...
... .... }_ ...
_. .... ,
1.00E+15
1.00E+14
7E-06 T
1.0
6E-06 \
2.0
3.0
4.0
5.0
6.0
Depth, [um]
cT1 5E-06 \
E. 4E-06 :
"> 3E-06 j
> 2E-06 j
1E-06 i
OE+00 :
10
20
30
40
50
60
1.0
Vcpd.tV]
FIGURE 3. Dopant depth profile for a p/p+ wafer. Top measured with non-contact dynamic CV method; bottom measured with Hg-probe. Average dopant concentration at a
depth between 2.0 to 3.0 urn is 1.17 e!5 and 1.27 e!5 cm"3,
respectively.
2
FIGURE 2. VSPV versus VCpD plot for a p/p+ wafer. Change
of the slope shows the transition from the lightly doped epilayer to the heavier doped substrate.
804
Measurements of Ion Implanted Wafers
Repeatability of the Method
This method can be used to measure doping
concentration up to lei8 cm"3. Therefore, it can be
used to monitor low and medium implant dose. In this
experiment, a p-type wafer was used with a 150A thick
oxide. The boron implanted at 25 keV was activated by
a RTF. The results of DDP are shown in Fig. 6. The
implanted peak is observed at a depth of 0.08 um. The
concentration at the maximum is about 2el7 cm"3,
while the background dopant concentration measured
at larger depth was 3el5 cm"3. The obtained 0.08 |0m
depth position of the implantation peak and the peak
concentration is in good agreement with the selected
implant conditions. The wafer was also measured with
the NSD method (i.e., without corona charging) that
relies on a native surface depletion with a measurement
of VSB and ac-SPV. The NSD measurement of dopant
concentration was 137 el? cm"39 which is an average
over a depth 0.053 jim.
Repeatability was studied on a wafer with a thermal
oxide. The wafer had uniform doping level (Fig. 7).
The average concentration was calculated between 3.0
to 5.0 jim. The data in Table 2 was obtained in a series
of 10 measurements repeated on the same site. A
standard deviation was 0.26% of the average value of
7.95 e!4 cm"3. A repeatability study on various nonoxidized wafers is underway.
1.00E+18 q
1.00E+18 3
FT
:
,0,
1.00E+17 :
•.v :.v v.. ..:. :::. .;i .v. v:. ::. :.v :.v.l*/*.tt.*'
~ 1.00E+17 :
0
'-
1
•-•"q^-J.-
rv:::]^::
:
1.00E-H6 -
S 1.00E+16 ;
0.04
;
" ~ ••" '" "" '^ * * '
o
o
:
0.09
0.14
* ° :*i o o::.* «, > * o o *!« o o 0 1
-|— - - f - - - - 1 — -]
0.0
0.5
1.0
1.5
2.0
2.£
Depth, [um]
FIGURE 6. DDP measured on a wafer implanted with Boron
with 25 keV energy and 4el2 cm"2 dose.
„ 1.00E+16 T
H
CO
'•g 1.00E+15
0
O
~i—N
8
1.00E+14
1.0
5.0
6.0
7.0
TABLE 2. Results of DDP repeatability measurements.
Concentration, [cm"3]
7.96E+14
runOl
run02
7.98E+14
7.97E+14
runOS
run04
7.93E+14
7.95E+14
run05
7.96E+14
run06
7.93E+14
nm07
7.93E+14
runOS
7.93E+14
run09
7.92E+14
runlO
7.95E+14
Avg=
StDev=
2.1E+12
%=
0.26%
1.00E+16
o
o
To 1.00E+15
I
O 1.00E+14
2.0
4.0
8.0
FIGURE 7. An example of a DDP of a p-type wafer with a
thermal oxide.
FIGURE 4. Schematic diagram of n/p structure.
1.5
3.0
Depth, [um]
.6 put
1.0
2.0
2.5
3.0
3.5
4.0
Depth, [um]
FIGURE 5. Results of DDP measurements on an n/p
structure. Average doping concentration between 1.5 to 2.5
umis 1.28 el5 cm"3.
805
CONCLUSIONS
A non-contact dynamic CV method for dopant
depth profiling has been developed based on corona
charging into deep depletion and simultaneously
measuring two transients during the collapsing of the
depletion layer. Contact potential voltage transients are
used to monitor the voltage across the depletion layer,
while small signal ac-surface photovoltage transients
are used to determine the corresponding value of the
depletion layer capacitance. The method was applied to
p/p, n/n, n/p epi-wafers, and ion implanted wafers with
dopant concentrations in the range from Iel4 to lei8
cm"3. Repeatability of the method was demonstrated to
be better than 0.3% for oxidized bulk silicon wafers
with a uniform concentration of 7.95 e!4 cm"3. The
method offers advantages of a quick turnaround of
results due to a preparation-free non-contact approach.
The method employs corona charging; however,
the results are obtained without using the corona dose
or any corona parameter. This simplifies very much the
experimental approach and is distinctly different from
previous doping measurements based on corona
charging.
REFERENCES
1.
2.
3.
4.
5.
6.
806
D. Schroder, Semiconductor Material and Device
Characterization, J. Wiley & Sons, 1998, Chapter 2.
E. Kamieniecki, 1 Vac. Sci. Technol., 20, 811-817
(1981).
R. Nakhmanson, Solid State Electron., 18, 617-626
(1975).
R. Lindner, Bell Syst. Tech. J., 41, 803-806 (1962).
D. Marinskiy, J. Lagowski, M. Wilson, A. Savtchouk, L.
Jastrzebski, D. DeBusk, "Small Signal ac-Surface
Photovoltage Technique for Non-Contact Monitoring of
Near Surface Doping and Recombination-Generation in
the Depletion Layer" in Nondestructive Methods for
Materials Characterization, edited by G. Baaklini et al.,
Mat. Res. Soc. Symp. Proc., Vol 591, Warrendale, PA,
2000, pp. 225-230.
S.M. Sze, Physics of Semiconductor Devices, Wiley,
1981, Chapter 2.