Thin Solid Films, 213 (1992) 257-264

Study of sputtered

257

HfO 2

thin films on silicon

C. T . K u o a n d R . K w o r
Department of Electrical and Computer Engineering, University of Colorado at Colorado Springs, Colorado Springs, CO 80933 (USA)

K. M. Jones
National Renewable Energy Laboratory, Golden, CO 80401 (USA)

(Received September 3, 1991; accepted December 3, 1991)

Abstract
Thin HfO2 films were deposited on Si(100) wafers by magnetron sputtering, followedby a high temperature anneal in oxygen.
Structure evolution and/or allotropic transformation occurred during the annealing process. A very thin well-definedSiO2 layer
was also formed at the HfO2-Si interface as a result of annealing. The HfO2/SiO2film showed a high dielectricconstant, very
low leakage current, extremelyhigh dielectricstrength, low dielectricloss, well-behavedcapacitance-voltage characteristics and
good stability, making it a viable candidate for applications in very-large-scaleand ultralarge-scale integration circuits.

1. Introduction
In view of the low dielectric constant and the physical
thickness limit of thin SiO2 films, high dielectric constant materials are being evaluated for applications in
very-large-scale integration (VLSI) and ultralarge-scale
integration (ULSI) circuits and scaled-down dynamic
random access memories (DRAMs) with densities beyond 4 Mbits. Refractory oxides have attracted considerable attention because of their high dielectric
constants, stability and structure simplicity. To date, a
substantial amount of research has been done on Ta205
[1, 2], ZrO2 [3] and Y 2 0 3 [4, 5] films, whereas work on
HfO2 [6] and La20 3 [7] has been somewhat limited.
Previous research indicated that these films have a good
potential and thus more study is needed. In this work,
a comprehensive investigation of HfO2 films has been
conducted. In particular, the properties of the films
have been studied to determine the deposition parameters necessary to create the optimum HfO2 films.

2. Experimental procedure
HfO2 films (200-300/~) were formed on 4 in Si(100)
wafers with resistivity in the range 1-10 ~ cm by either
r.f. magnetron reactive sputtering (in pure oxygen) of a
hafnium target with 99.9% purity (method A) or by r.f.
sputtering (in argon) of a HfO2 target with a purity of
99.95% (method B). In both cases, the sputtering was
followed by a high temperature oxygen anneal to densify the as-deposited films. Either dry or wet oxygen can
be used as the annealing ambient. Suitable post-

0040-6090/92/$5.00

deposition oxygen annealing temperatures were determined by comparing the refractive index data, accumulation capacitances (obtained from capacitance vs. bias
voltage ( C - V ) plots by the mercury probe technique)
and X-ray diffraction patterns (using Cu K0q radiation at 35 kV). The choice of annealing temperature
and time will be discussed in some detail in the
next section. The structure of the resultant films was
investigated by cross-sectional transmission electron microscopy (TEM). Metal-insulator-semiconductor capacitor (MIS-C) devices were fabricated to evaluate the
electrical properties of these HfO2 films, including the
dielectric constant, interface state density, dissipation
factor, dielectric strength and resistivity. After HfO2
film formation, an aluminum film with a thickness of
8000 A was sputtered and then patterned into four
differently sized circular dots (areas equal to
2.3 x 10 - 3 cm 2, 1.3 x 10 -3 cm 2, 5.76 10 - 4 cm 2 and
1.44 10-4cm 2) as the front-side electrodes. Subsequently, the back-side oxide was removed by dipping
the substrate in 10% H F (front-side aluminum patterns
were protected with photoresist). Aluminum was then
sputtered to form the back-side electrode. Finally, the
wafer was subjected to 350-450 C sintering in argon
for 30 min to improve the contacts.

3. Results and discussion

3.1. Oxygen annealing effects

In this experiment, the oxygen annealing effect was

evaluated by examining the refractive index of the
resultant film (measured by ellipsometry), because it is

1992-- Elsevier Sequoia. All rights reserved

258

C. T. Kuo et al. / Sputtered HfO 2 on Si

TABLE 1. Refractive index dependence on oxygen annealing temperature and time for r.f. reactive sputtering of hafnium target (as-deposited film thickness, about 255 .~)
Sample

Vl-0a
Vl-1
VI-2
VI-3
VI-4
Vl-5
VI-6
Vl-7

Annealing
ambient

Wet 0 2
Wet 02
Wet 02
Wet 02
Wet 0 2
Wet 0 z
Wet 02

Annealing
temperature
(of)

Annealing
time
(h)

Refractive
index

-400
500
600
700
800
800
900

-1
1
1
1
1
1

1.756 + 0.008
1.765 + 0.008
1.778 ___0.004
1.790 _ 0.015
1.815 _+0.005
1.809 4- 0.018
1.745 + 0.008
1.708 + 0.012

TABLE 2. Refractive index dependence on oxygen annealing temperature and time for r.f. sputtering of HfO2 target (as-deposited film
thickness, about 325/k)

V5-0a
V5-1
V5-2
V5-3
V5-4
V5-5
V5-6
V5-7

Annealing
ambient

Annealing
temperature
(of)

Annealing
time
(h)

Dry
Dry
Dry
Dry
Dry
Dry
Dry

400
500
600
700
800
900
900

1
1
1
1
1

02
O2
02
02
02
02
02

AnnealingTime = 1 hr.
1.0

o
o

-o

aRefractive index value of as-deposited film is listed here for reference.

Sample

1.5

Refractive
index
1.837 + 0.005
1.934 + 0.001
1.942 + 0.006
1.952 + 0.002
1.964 -I-0.006
1.927 ___0.001
1.920 + 0.002
1.913 ___0.001

aRefractive index value of as-deposited film is listed here for reference.

well k n o w n that denser films always have a higher

refractive index, dielectric constant and dielectric
strength. Tables 1 and 2 present the dependence o f the
refractive index on the oxygen annealing temperature
and time for the HfO2 films by methods A and B
respectively. It can be seen that, for the films formed by
r.f. reactive sputtering o f the hafnium metal target, the
highest value o f the refractive index is found from the
samples annealed at 700 C (for 1 h) or 800 C (for ~h),
where the thickness o f as-deposited films is about
255/~. As for the films p r o d u c e d by the r.f. sputtering
o f HfO2 target ( m e t h o d B), 700 C was the annealing
temperature that gives the highest refractive index for
a b o u t 325/k as-deposited films. F o r the films deposited
by m e t h o d A and then followed by oxygen annealing, a
higher refractive index corresponds to higher accumulation capacitance, and thus higher dielectric constant
(Fig. 1). F o r the films deposited by m e t h o d B, there is
a similar correlation. However, the matching is not as
close. According to the X-ray diffraction ( X R D ) data
(see Section 3.2.1), an allotropic transformation o f

[]

[]

0.5
o Method A -255/~
i

u Method B -325/~
I

o
400

500

600

700

800

900

10o0

Annealing Temperature, C
Fig. 1. Normalized accumulation capacitance vs. annealing temperature for the films deposited by methods A and B.

HfO2 occurred during the annealing process following

m e t h o d B. This is p r o b a b l y the reason why the dielectric constant and refractive index o f the films do not
closely correspond to each other.
F r o m Tables 1 and 2, it is clear that the refractive
index increases with increasing annealing temperature
until it reaches a m a x i m u m value. Then the refractive
index decreases with increasing temperature. The HfO2
film densities during annealing but, at the same time, a
very thin layer o f SiO2 also grows between the HfO2
film and substrate (see T E M m i c r o g r a p h in Fig. 2). The
densification o f the HfO2 film dominates in the lower
temperature range, while the interfacial SiO2 film
growth becomes significant near the temperature when
the refractive index is a m a x i m u m , causing a d r o p in
the overall dielectric constant.
3.2. Physical and chemical properties o f H f 0 2 f i l m s
3.2. I. Film structure
The X R D patterns indicated that the final (stable)
form o f HfO2 is a polycrystaUine monoclinic substance
with lattice constants a = 5 . 2 8 5 , ~ , b = 5 . 1 8 2 , ~ and
c = 5.116/k, and a preferred orientation in the ( - 1 , 1, 1)
plane.
F o r the films deposited by m e t h o d A, when the
as-deposited film thickness is a r o u n d 255/k (Figs. 3 and
4), the X R D patterns change only slightly between
annealing temperatures of 400 and 600 C, ; however, at
a r o u n d 700 C the diffraction peak becomes sharper
and its intensity also becomes higher; then, above
700 C (up to 900 C), the temperature again has little
effect on both the peak shape and the intensity. The
mean grain sizes calculated f r o m X R D peak broadening or the peak widths at full width at half-maximum

C. T. Kuo et al. / Sputtered HfO 2 on Si

CPS 5.90
100

259

(-1 3)

2.25

1.43

%
100

90

HfO 2

SiO 2
Si

(a)

500 h

90

80

80

70

70

60

60

50

50

40

40

30

30

20

20

10

10
I

15

HfO 2

SiO 2

Si

(b)

I
250 A

Fig. 2. TEM photographs showing a SiO 2 layer formed at the

HfO2-Si interface during annealing; the annealing temperatures are
(a) 900 C and (b) 800 C.
CPS 5.90
100
90

2.25

1.43

%
100
90

(-1,1,1)

80

80

70

70

60

60

50

50

40 ~

40

30

30
20

lO

10

15

40

65

40

Fig. 4. X R D patterns of HfO 2 films from method A after annealing

700-900 C (film thickness, about 255 tk): CPS, counts s 1.

In the case of the films obtained by method A, apart

from the grain size affected by the temperature, only
one phase (i.e. the monoclinic phase) is involved. However, the as-deposited films formed by method B are
amorphous, and their structure remains amorphous as
long as the annealing temperature is below 400 C (1 h
annealing). However, when the annealing temperature
is increased to between 500 and 600 C, the film structure changes from amorphous to a mixture of the
monoclinic and orthorhombic phase. Then from 700 to
900 C, only one phase (the monoclinic structure) dominates. The grain sizes are about 325 ,~, after annealing
at 800 C for 1 h (the same value as obtained by Balog
et al. [6] for their 800/k HfO: films deposited by
chemical vapor deposition (CVD)) for HfO: films
325 ,~ thick.
From Figs. 3-6, it can be seen that the temperature
at which the diffraction peak is narrower and higher is
consistent with that at which the accumulation capacitance (or dielectric constant) of the film is highest (Fig.
1), except for the temperatures above the point where
CPS 5.90
200
180

2.25
J

1.43
I

%
/ 100
L 90

M(-I,I,1)

65

Fig. 3. X R D patterns of HfO 2 films from method A after annealing

at 400-600 C (film thickness, about 255/~): CPS, counts s -t.

vary from about 160,~ (400-600 C) to about 210215 A (700-900 C), using the Scherrer formula [8]

160
140

I~

I~ 80
70

I,I,I)

100 _,...,_,3 v .k.~.C/lh,~ o 2 ~ e ~ g

50

60-8

3040

;Clhr

0.92
t - (f12 _ B 2) 1/2 c o s 0B

(1)

where t is the mean grain size, 2 is the X-ray (Cu K~q

line) wavelength (1.5405 ,~), fl is the broadening of the
diffraction line measured at half its maximum intensity,
B is the machine broadening and 0B is the Bragg angle.

20 0 ~
15

As deposited
, . . . I.
',
40

10

_
i

~1

65

Fig. 5. X R D patterns of HfO 2 films from method B after annealing

at 400-600 C (film thickness, about 325/~): CPS, counts s - L

C. T. Kuo et al. / Sputtered H f 0 2 on Si

260
CPS 5.90
500

2.25

1.43

%
100

(-1,1,1)

450

90

400

80

350
300

(1,1,1)

(-2,2,2)

250

too0,

0.08 -

8001

0.10

= 1 MHz

Frequency

C-V

70

~ 0.06 -

60

.g

50

-~ 0.O4 -

6001
40(

200
0.02 -

150
100

50
0

J t tk,x8 "c1hrdry~ armealinag ~

,

15

40

I 10
,

20C "I

-5 -4 -3 -2 -1

Bias (V)

65

Fig. 6. X R D patterns of HfO 2 films from method B after annealing

at 700 900 C (film thickness, about 325 ~): CPS, counts s - L

the dielectric constant is highest and the growth of the

interfacial SiO2 film is taking place. The monoclinic
structure is believed to be the stable form of HfO2 films,
while the orthorhombic phase is a metastable phase.
This monoclinic phase of HfO2 has been reported by
Balog et al. [6] who used CVD to prepare the HfO2 and
Thomas [9] who prepared HfO2 using anodic oxidation.
However, a triclinic structure was mentioned by Smith
[10], who deposited HfO 2 films by reactive sputtering.
3.2.2. Film stress
The as-deposited HfO2 films possessed a compressive
stress, which is between - 1 x 108 and - 3 108pa
for method A films and between -2.5 108 and
-4.5 x 10s Pa for method B films. After annealing, the
HfO2 films showed a tensile stress in the range (25.5) x 10s Pa. The reason that the film stress changes
from compressive to tensile after the high temperature
annealing could be due to the density changes and/or
grain boundary movements [11]. For the sputtered
films, the stress value of (2-5.5) 10s Pa is comparatively low.
3.2.3 Film etching rate
The H f O 2 films etch slowly in 10% HF solution. The
etching rates for the films produced from methods A

Fig. 7. Typical 1 MHz C - V and D - V characteristics of A I / H f 0 2 /

SiO2/Si capacitors.

and B are about 25/~min -1 and 20/kmin -1 respectively. As expected, the rate is affected by the annealing
temperature and time, as summarized in Table 3. Generally, the etching rate decreases with increasing annealing temperature, indicating the film-densifying effect of
the annealing (although the etching rate from reactive
ion etching (RIE) does not seem to depend very much
on film density). From the table, it can also be seen that
the etching rate of the HfO2 film made by method B is
lower than those made by method A, indicating a
higher density for the films obtained by method B.
Apart from HF, the HfO2 formed by both methods are
quite insensitive to many acids or bases. They cannot be
etched by hot H2SO 4 or HNO3, boiling HC1, hot
CH3COOH or cold NH4OH.
3.3. Electrical properties of HfO 2 films
A1/HfO2/SiO2/Si MIS capacitors with an HfO2 thickness ranging from 200 to 300 ,~ have been characterized. Figures 7 and 8 indicate the typical 1 MHz C - V
and dissipation factor-voltage ( D - V ) curves and
quasi-static C - V curve. A typical current-voltage ( I V) curve for A1/HfO2/SiO2/Si capacitors is shown in
Fig. 9. Through the C-V, D - V, and I - V curves, such
parameters as the dielectric constant e, mobile ion, fixed
surface charge and interface state density, dissipation

TABLE 3. Etching rate of HfO 2 films vs. annealing temperature

Etching
agent

Film formation
method

CF 4 and H2(RIE )
CF4 and H2(RIE )

A
B

10% H F
10% HF

A
B

Etching rate (/k min ~) at the following annealing temperatures

400 C

500 C

700 C

800 C

900 C

~28
~22

.~28
~22

~20
~ 18

~20
~ 18

---

~ 120

30 - 40

20 - 30

~ 20

260

C. T. Kuo et al. / Sputtered HfO2 on Si

600

higher density and/or higher refractive index have a

higher dielectric constant. Compared with the known
values of dielectric constant for HfO2 films, the values
obtained in this work are lower, which might be attributed to the prominent polycrystalline nature of our
HfO2 films [ 12].
The capacitance densities calculated were 0.250.291aFcm -2 and 0.32-0.37~tFcm -2 for the HfO2
films with a thickness of 250/k from methods A and B
respectively. These values are higher than any other
published data for HfO2 films.

500
400
300

"~ 200
1oo

-4

-2

0
Voltage(V)

3.3.2. Mobile ion, fixed surface charge and interface

state density
Mobile ion drift is one of the main concerns relating
to instabilities of metal-oxide-semiconductor (MOS)
devices. The bias-temperature stress (BTS) technique
[13] has been developed for detecting mobile ions.
Typical results from BTS measurements for HfO2 films
by methods A and B show a negligible presence of hole
trapping and mobile positive ions in the oxide.
The average density of the effective surface states
(including both the oxide fixed charges and the interface traps) of the HfO2/SiO 2 films was found to be
approximately 2 x 10 '~ cm -2, which is close to the average surface state density in the SiO2/Si system
(5 10 l0 cm-Z).

Fig. 8. Quasi-static C-V curve for HfO2 film.

-1

10

165 Capacitor Area

10 .9

1613

-5

-10

-15

-20

-25

-30

(v)

Fig. 9. Typical I - V characteristics of AI/HfO2]SiO2/Si capacitors.

factor D, leakage current density, dielectric strength,

resistivity, and charge storage capacity can be deduced.
3.3.1. Dielectric constant and capacitance density
Based on the 1 MHz C - V curves (measured at
20 C), the dielectric constant of HfO2 films was obtained from

1
%A
dsio2
eHfO2 CaccdHfo2 dHfo2~sio2

261

(2)

where Ca~ refers to the accumulation capacitance observed from 1 MHz C - V curves, CHfo2 pertains to the
capacitance of HfO2 films ei~fo2and dHfo2 are the dielectric constant and thickness of HfO2 films respectively, A
is the capacitor area, e0 is the permittivity in vacuum,
and esio2 and dsio2 indicate the dielectric constant and
thickness respectively of SiO2 layers.
T h e dielectric constants obtained were 12 + 1 and
17 + 2 for HfO2 film produced by methods A and B
respectively. It was found that the dielectric constant of
HfO2 films provided by method B is higher than that of
HfO2 films by method A, verifying that the films with a

3.3.3. Dissipation factor and resistivity

Generally speaking, the dissipation factor is an indication of the structural imperfections associated with
defects in an oxide film. The value of dissipation factor
usually refers to that obtained at 1 kHz and at room
temperature. Since our investigated frequency range
varied from 10 kHz to 10 MHz, owing to the limitation
of available equipment, here the dissipation factors of
HfO2/SiO2 films indicate the values measured at 10 kHz
and 20 C. They were between 0.0002 and 0.005 for
films provided by either method A or method B. Those
values satisfy the criteria for high capacitance capacitors, where moderate loss (tan 6 < 1%) is needed [14]
and also meet the requirements of MOS devices, where
a good-quality dielectric is desired (tan 6 ~ 0.1%) [12].
There are two ways of obtaining the resistivity of
HfO2/SiO2 films. By employing tan 6 = 1/coRC, one can
obtain R, if t a n 6 (at 1 kHz), co and C are known
[15, 16]. The second method of calculating the film
resistivity is based upon the resistance of the MIS
capacitor obtained from d.c. conductivity (or l - V )
measurement [ 17, 18]. On the assumption that the value
of tan c5 is 0.0002 at 1 kHz, the resistivity was found to
be on the order of 10~2f~ cm. The resistivity derived
from the I - V curves varied from 5 x 10" to
5 1012f~cm, which agrees well with that obtained
from method A. The average resistivity of HfO2/SiO2

262

C. T. Kuo et al. / Sputtered H f O 2 on Si

films (about 1012 ~ cm) is lower than the typical value

of SiO2 and Si3N 4 at room temperature (about
1014~ cm) [19]. The lower resistivity may be caused by
the grain boundaries present in HfO2 films which are
polycrystalline. The grain boundaries act as the sites for
preferential absorption of impurities [20], resulting in a
lower film resistivity.

3.3.4. Leakage current density and dielectric strength

Figure 9 shows the typical I - V characteristics of
HfO2/SiO2 films from either method A or method B.
Here we define an effective thickness as the thickness of
a SiO2 film for which the MOS capacitor has the same
capacitance as that possessed by the A1/HfO2/SiO2/Si
MIS capacitor. A very low leakage current with
a density of lower than 10-9A cm -2 under an effective applied field of 4 M V c m -1 was observed (see
Fig. 8; there is no tilting phenomenon occurring in
the quasi-static C - V curve). The effective dielectric
strength of HfO2/SiO2 films varied between 11.5 and
15.5 MV cm-~, which is higher than that of the conventionally grown SiO2 films (less than 10 M V c m -1 [19]).
This high dielectric strength could be attributed to the
very high quality homogeneous interfacial SiO2 film.
3.3.5. Charge storage capacity
The charge storage capacity per unit area is calculated by multiplying the accumulation capacitance per
unit area with the statistically average breakdown
voltage of the A1/HfO2/SiO2/Si capacitors. The charge
storage capacity of our HfO2/SiO2 films is in the range
4 - 5 laC cm -2, and is higher than the storage capacities
for SiO2 (2.7 I.tCcm -2) and Si3N 4 (3.1 taCcm-2), as
previously reported in ref. 21. It also satisfies the requirement imposed by ULSI D R A M cells (at least for
the 64 Mbit level, in which 1-7 ~tC cm -2 storage charge
density is needed for the trench type of capacitors) [22].
3.3.6. Frequency and temperature effect
Dielectrics have losses due to electronic, ionic and
dipolar contributions to polarizability [14]. All these
losses are frequency dependent. In the frequency range
from d.c. to kilohertz, the loss may also come from a
small contribution from the d.c. conductivity of the
material. Since the HfO2/SiO2 films have a very low
leakage current, they should have negligible conduction
loss [5]. To investigate the other dielectric losses of
these films, the change in accumulation capacitance as a
function of frequency varying between 10kHz to
4 MHz was measured. Figure l0 shows a typical plot of
normalized accumulation capacitance vs. frequency
measured at 20 C with an a.c. voltage of 100 mV for
the A1/HfO2/SiO2/Si capacitors. It can be seen that the
accumulation capacitance is nearly constant from
10kHz to 1 MHz. Above 1 MHz the accumulation

1.5

1.0

)-

0.5

10

HfO 2

~290 ,~,

SiO 2

~ 50 /~

105

106

10

Frequency in Hz

Fig. 10. Normalized accumulation capacitance

HfO2/SiO2/Si capacitors.

vs.

frequency for A1/

capacitance decreases, which has the same trend as

observed by Manchanda and Gurvitch [5] for their
Y203/5iO2 films. They related this phenomenon to a
more resistive substrate at high frequencies, and the
capacitance-measuring instrument cannot correct for the
high dissipation factors associated with high resistivity
[5]. It seems that at high frequencies the energy loss
caused by series resistance of the measuring circuit and
the inductive reactance from the probe wire can also play
a prominent role [23]. The loss at high frequencies is not
believed to be associated with the insulator. We can
conclude here that, because of a lack of appreciable
frequency dependence of the accumulation capacitance,
there is no measurable dielectric loss in the frequency
range from l0 kHz to 4 MHz for HfO2/SiO 2 films.
The influence of temperature on the accumulation
capacitance (or dielectric constant) and the dielectric
loss can be explained by the model suggested by
Goswami and Goswami [24]. In this case, the equivalent circuit of a capacitor element comprises a parallel
combination of a resistance R and an inherent capacitance C, which is, in turn, in series with a small
electrode resistance r.
The dielectric constant (e.g. e = 17 for H f O 2 films
from method B) is due to a contribution of different
polarization mechanisms. The contribution to the
dielectric constant from electronic polarization was calculated to be about 3.8 (~o = n 2, n was chosen to be
1.95; see Table 2). The remaining contribution may be
attributed to ionic polarization due to the ionic nature
of the oxide films. Since the frequency and temperature
ranges studied here are not low enough (the relaxation
effect usually is observed in the low temperature region
at low frequencies), contributions from dipolar and
possibly interfacial polarization cannot be completely
ruled out, even though our results did not show any
associated loss peaks.

c. T. Kuo et al. / Sputtered HfO2 on Si

3.3. 7. Thickness effect
For many dielectric films, there is a variation in
dielectric constant with thickness due to defects such as
voids, stress, inhomogeneity and discontinuities [23].
However, the dielectric constant of our HfO2 films
shows little thickness dependence (e of the HfO2 films
from method B are 16 _ 1 and 19 _ 1 for thicknesses of
around 1300 A and 3000 ]k respectively). The dissipation factor and the breakdown field E B of HfO2/SiOz
films also have little thickness dependence (for instance,
EB is in the range 14.5-16MVcm -t for HfO2 films
about 1000 A thick), and E B does not follow the Forlani-Minnaja equation [24,25] (i.e. the dielectric
strength is a power function of the film thickness). This
might be because these dielectrics have a double structure and are not just a single layer. However, when the
thickness of the HfO2 film was reduced to about 200 ]k
(corresponding to an effective SiO2 thickness of 8090/~), the dielectric strength decreased to about
8 MV cm-~. Thus there is a limit for which the dielectric strength is independent of the thickness.

3.3.8. Conduction mechanism

The typical conduction behavior of the HfOz/SiOz
structure is shown in Fig. 11. This figure reveals a
Fowler-Nordheim plot for the AI/HfO2/SiO2/Si capacitors. Fowler-Nordheim tunneling conduction has the
following relationship between the current and the applied voltage [ 19]:

I = aV2 e x p ( - ~ )

(3)

where a and b are constants.

-0.032-

263

By Gauss's law, when an electric field is applied to a

HfO2/SiO2 film, nearly all the field is dropped across
the SiOz layer because of its lower dielectric constant
(i.e. Esio2/EHfo2= e H f O 2 / / ~ S i O 2 ) . This indicates that the
conduction mechanism through the A1/HfO2/SiO2/Si
capacitors is predominantly determined by the tunneling current through the SiO2 film at higher electric fields
(Ivl > 8 v, for HfO2( ~260 ]k)/SiO2( ~40 .~) films).
As is known, at high fields the most common conduction mechanism which takes place in dielectric films is
Poole-Frenkel or space-charged-limited conduction
[6, 21, 26, 27]. Here Fowler-Nordheim tunneling mechanism is observed mainly because of the influence from
the SiO2 interlayer. Coincidentally, Fukumoto et al. [3]
and Nishioka et al. [28] also proposed the FowlerNordheim tunneling mechanism for the current transport through their Zr02/Si02 and Ta2Os/SiO2 films
respectively.

3.3.9. Interface charge trapping and hysteresis effect

Charge trapping at the metal-oxide interface has
been measured to evaluate the oxide reliability. When a
repetitively ramped voltage is applied to a capacitor,
the I - V curve corresponding to the second voltage
ramp is shifted due to interface trapping. Figure 12
shows a typical sequential-ramp I - V curves for A1/
HfO2/SiO2/Si capacitors (with the HfOz/SiO2 film
formed by method A). The second-ramp I - V curve
shows little shift from the initial curve, indicating the
existence of very little charge trapping. A decrease in
leakage current observed from the second-ramp I - V
curve is considered to be caused by hole trapping.
When the third voltage ramp is applied, the I - V curve
shifts only slightly from the second-ramp I - V curve,
showing the filling of the traps.
10 -7

0.001V
|

10 -s
-0.034 10 -9

m Third R a m p
Second Ramp
First Ramp

,
j

-0.036
1040

10-it

-0.038

10-,2

.m0

llmm

-0.040
10-13

-0.042

10"t4
8

10

12

14

16

18

20
10

Voltage (V)

2O

Fig. l 1. Fowler-Nordheimtunnelingconductionthroughthe HfO2/ Fig. 12. Typical sequential ramp I-V curves for AI/HfO2/SiO2/Si
SiO2 layer.
capacitors.

264

C. T. Kuo et al. / Sputtered HfO 2 on Si

The charge trapping at the oxide-semiconductor interface or in the oxide bulk is also negligible, as revealed by the hysteresis observation of both the high
frequency and the quasi-static C - V curves. Here, either
a non-hysteretic or a slightly hysteretic C - V curve was
seen for the HfO2/SiO2 films.

3.3.10. Stability
A comparison of C - V and D - V characteristics measured shortly after the MIS capacitor was made with
those obtained 3 months later showed that there is no
dielectric degradation. I - V characteristics also showed
no sign of deterioration. All these phenomena lead to
the conclusion that the HfO2 films formed by methods
A and B have a good stability.

4. Conclusion
In the present work, H f O 2 films with thickness between 200 and 300 A were deposited on silicon substrates by either r.f. sputtering of hafnium in oxygen or
r.f. sputtering of HfO2 in argon, followed by a high
temperature anneal in oxygen. A very thin well-defined
SiO2 layer was formed at the HfO2-Si interface during
annealing. At the same time, a structure evolution for
the HfO2 film also occurred. Comprehensive characterization has revealed that these films have very good
physical, mechanical and electrical properties. The
HfO2/SiO2 system has high dielectric constant (compared with SiO2), high dielectric strength, low dielectric
loss, low leakage current, well-behaved C - V characteristics and good stability. The film fabrication process is
simple and compatible with the conventional integrated-circuit processing. The HfO2/SiO2 system thus
deserved further study and consideration as a potential
candidate for applications in VLSI and ULSI circuits.

Acknowledgments
The authors wish to thank Dr. L. Kammerdiner of
Ramtron Corporation for his help with the XRD analysis, and Dr. T. L. Hwang, Dr. L. W. Yang and Mr. Y.
Chang of Ford Microelectronics, Inc., for their support
in reactive ion etching. Thanks are also due to Mr.
Steve Jernigan for his technical assistance in the sputtering processes.

This work was mainly supported by a scholarship

grant from the Institute of Nuclear Energy Research,
Taiwan.

References
1 S. G. Byeon and Y. Tzeng, IEEE Trans. Electron Devices, 37
(1990) 972.
2 M. Matsui, S. Oka, K. Yamagishi, K. Kuroiwa and Y. Tarui, Jpn.
J. Appl. Phys., 27 (1988) 506.
3 H. Fukumoto, M. Morita and Y. Osaka, J. Appl. Phys., 65 (1989)
5210.
4 M. Gurvitch, L. Manchanda and J. M. Gibson, Appl. Phys. Lett.,
51 (1987) 919.
5 L. Manchanda and M. Gurvitch, IEEE Electron Device Lett., 9
(1988) 180.
6 M. Balog, M. Schieber, M. Michman and S. Patai, Thin Solid
Films, 41 (1977) 247.
7 T. Mahalingam, M. Radhakrishnan and C. Balasubramanian,
Thin Solid Films, 78 (1981) 229.
8 H. P. Klug and L. E. Alexander, X-ray Diffraction Procedures for
Polycrystalline and Amorphous Materials, 2nd edn., Wiley, New
York, 1974.
9 M. T. Thomas, J. Electrochem. Soc., 117 (1970) 396.
10 F. T. J. Smith, J. Appl. Phys., 41 (1970) 4227.
l 1 D. S. Campbell, in L. I. Maissel and R. Glang (eds.), Handbook
of Thin Film Technology, McGraw-Hill, New York, 1970, p. 12-3.
12 P. J. Harrop and D. S. Campbell, in L. I. Maissel and Glang
(eds.), Handbook of Thin Film Technology, McGraw-Hill, New
York, 1970, p. 16-1.
13 D. G. Ong, Modern MOS Technology: Processes, Devices and
Design, McGraw-Hill, Ne'~ York, 1984.
14 P. J. Harrop and D. S. Campbell, Thin Solid Films, 2 (1968) 273.
15 C. R. Dutta and K. Barua, Thin Solid Films, I00 (1983) 149.
16 H. Oetzmann, H. J. Krokoszinski, H. Gernoth and C. Schmidt,
Thin Solid Films, 140 (1986) 209.
17 T. A. Raju and A. A. Talwai, J. Appl. Phys., 52(1981) 4877.
18 C. K. Campbell, Thin Solid Films, 6(1970) 197.
19 S. M. Sze, Physics of Semiconductor Devices, 2nd edn., Wiley,
New York, 1981.
20 R. W. Wilson and L. E. Terry, J. Vac. Sci. Technol., 13 (1976)
157.
21 W. P. Noble and W. W. Walker, IEEE Circuits Devices, 1 (1985)
45.
22 L. H. Parker and A. F. Tasch, IEEE Circuits Devices, 6 (1990)
17.
23 E. H. Nicollian and J. R. Brews, MOS Physics and Technology,
Wiley, New York, 1982.
24 A. Goswami and A. P. Goswami, Thin Solid Films, 16 (1973) 175.
25 A. Singh, Thin Solid Films, 105 (1983) 163.
26 H. Birey, J. AppL Phys., 49 (1978) 2898.
27 B. E. Deal, P. J. Fleming and P. L. Castro, J. Electrochem. Soc.,
115 (1968) 300.
28 Y. Nishioka, S. Kimura, H. Shiniki and K. Mukai, J. Electrochem. Soc., 134 (1987) 410.