Diffusion of Chromium in Thin Hydrogenated Amorphous Silicon

[Watchlist]

Физика и техника полупроводников, 2004, том 38, вып. 3
Diffusion of Chromium in Thin Hydrogenated Amorphous Silicon Films
© S.K. Persheyev {, P.R. Drapacz, M.J. Rose, A.G. Fitzgerald
Carnegie Laboratory of Physics,
Electronic Engineering and Physics Division,
University of Dundee, Dundee, DD1 4HN, Scotland, UK
(Получена 17 июля 2003 г. Принята к печати 18 июля 2003 г. )
The diffusion of chromium bottom contact has been studied through thin
10 nm amorphous silicon film. The
concentration of the diffused impurity has been analysed by an X-ray
photon spectroscopy technique and the
diffusion coefficient was estimated. Diffusion annealing was carried
out in vacuum (10−6 mTorr), the temperature
was kept at 400 C and annealing time varied from 0−300min. The
authors propose that diffusion of chromium in
thin hydrogenated amorphous film limited by silicide formation at the
metal−silicon interface.
1. Introduction
While microelectronics device production is moving into
nanoscale region chromium diffusion in amorphous silicon
(a-Si :H) is a significant area of study for number of reasons.
Good electrical contacts to semiconductors are crucial for
the proper functioning of circuits because signals enter and
leave devices through them. One requirement for contacts is
chemical stability. Furthermore, it is essential that a contact
does not introduce unwanted electrical characteristics such
as signal rectification or high resistance into the circuit.
Since a-Si :H consists of a random network of atoms,
distorted considerably from the minimum energy sites of
its crystalline counterpart, it can easily interact with other
materials. Interaction at a metal/a-Si :H contact can cause
irreversible interfacial degradation; however, it may also be
used to some advantage.
Memory devices based on a-Si :H [1] that act as two
state digital devices or analogue devices with a continuum
of states depending on the nature of their metal contacts
(e. g. chromium or vanadium) are finding applications as
non-volatile switching materials. The nature of the top
metal interaction is of crucial importance to the switching
characterictics.
Amorphous silicides formed in hydrogenated amorphous
silicon are not well understood despire the fact that
silicides are widely used in contacting technologies for the
thin film transistors needed in active matrix liquid crystal
displays. The most widely studied metal is chromium,
and it has been shown that silicide layers at most a few
nanometres thick are formed after sputtering chromium onto
hydrogenated amorphous silicon even with no subsequent
annealing treatment. Recent work [2] suggests that using
an ultra-thin metal-like chromium silicide in an entirely
amorphous structure naturally forms a hot electron device
with a high electron barrier emitter and low electron barrier
collector. The aim of this work is further understanding of
chromium diffusion mechanism and metal/a-Si :H interface
reactions mediated by thermal annealing.
{ E-mail: s.per...@dundee.ac.uk
FAX: +44{1382{348313
Tel.: +44{1382{344563
2. Experimental
Hydrogenated amorphous silicon layers were deposited
onto Corning 7059 glass preliminarily coated with sputtered
chromium as show in Fig. 1. Films were deposited using
plasma enhanced chemical vapour deposition (PECVD)
from silane at a substrate temperature of 300 C, applied RF
power of 6.5W, a silane flow rate 40 sccm, diving a growth
rate of 0:9 A/s. Typical layers had a thickness of 10 nm
for Cr diffusion experiments and 600 nm films on silicon
wafers for Fourier transform infrared spectroscopy (FTIR).
Chromium was sputtered on glass substrates to a
thickness of 30 nm and immediately amorphous silicon
was grown to avoid Cr surface oxidisation. Films for
thickness measurements were patterned using an interdigital
photomask and after removing the photoresist film the
surface was profiled using thickness profiling technique. To
study the thermal diffusion of chromium, samples on glass
were annealed at 400 C in vacuum (10−6 mTorr) and then
samples were analysed for presence of chromium in a X-ray
photon spectroscopy (XPS) system. The spectra are shown
in Fig. 2.
All the XPS measurements we have done were for the
chromium diffusion studies using Mg-K-alpha radiation. The
apparatus (a VG HB 100 adapted to incorporate XPS) was
operated in constant analyser energy mode with the X-ray
gun itself running at 130W. The pass energies were 50 eV
for survey scans and 20 eV for region scans. Spectra were
collected using VGX900 software and the system was kept
calibrated using the procedures outlined in ISO/DIS 15472.
Figure 1. Schemical representation of layers for diffusion study.
358
Diffusion of Chromium in Thin Hydrogenated Amorphous Silicon Films 359
Figure 2. XPS signal from amorphous Si: 1 | before and 2 |
after annealing at 400 C, 5 h.
Figure 3. Concentration of chromium (1) and hydrogen (2) after
annealing in vacuum. Dashed lines are our theoretical calculations
for diffused chromium.
3. Discussion
The diffusion ot the contact metals in PECVD amorphous
hydrogenated films shows unusual behaviour of diffusing
materials and is quite different than that found in crystalline
silicon [3]. The diffusion coefficient of the impurity in
amorphous silicon depends only weakly on the impurity
itself, and the diffusion coefficient and is activation energy
are nearly equal to those of hydrogen. It was found that
boron diffuses quite fast with a diffusion coefficient of
about 10−13 cm2/s and the antimony diffusion coefficient
was found 10−14 cm2/s at 400 C. Activation energies are
the same for both and are about 1.5 eV.
A. Polman and co-authors [4] studied the diffusion of
copper in annealed and unannealed a-Si :H in the tem-
perature range 150−270 C. The diffusion rate in annealed
amorphous silicon is a factor of 2−5 times higher than
in unannealed a-Si : H. The diffusion activation energy
Ea = 1:39 eV in annealed a-Si :H is not significantly dif-
ferent than in unannealed a-Si : H, where it is measured to
be Ea = 1:25 eV.
The diffusion of silver has been studied in previous
work [5] and the authors conclude that silver diffuses
in undoped amorphous silicon interstitialy and through
hydrogen vacancies with activation energies of 1.3 eV
and 1.7 eV respectively. Interstitial diffusion described as
Di = 4:4 10−3 exp(−1:3=kT) cm2/s and diffusion through
vacancies as Dv = 70 exp(−1:7=kT) cm2/s.
In our experiments an annealing temperature 400 C has
been chosen for two reasons: it is known that diffusion of
metals in a-Si :H at temperatures lower than 400 C is small
and requires very long diffusion annealing times, and second
hydrogen effusion occurs at temperatures higher than 400 C
and changes the amorphous structure and its network. The
experimental results are represented in Fig. 3. As we can
see from the grahp the hydrogen concentration decreases
from 15 to 8.3 at% mainly at the beginning of the anneal
where the initial drop in the first 30 min is obviously due
to hydrogen effusion from the film. At the longer times
the change in the hydrogen content is relatively small and
after 5 h of annealing the remaining hydrogen is about 6 at%.
The change of hydrogen content in the beginning can also
be explained due to relaxation processes in the amorphous
matrix. With the releasing of some hydrogen we are
creating additional
”
hydrogen vacancies“, enabling diffusion
of chromium impurities. The XPS surface analysis technique
allowed detection of Cr atoms diffused through the very
thin 10 nm a-Si :H layer. The time dependence of the Cr
concentration is also plotted in Fig. 3. For purpose of
theoretical calculations a thick layer of Cr (30 nm) can be
considered as an unlimited source, and diffusion through
a thin film described as the diffusion in a body with one
impermeable border [6]. The impurity concentration at the
first boundary of a body (x = 0) has a constant value C0
which does not change with time. The second boundary
is regarded as impermeable, with dC=dx = 0 at x = l. For
such a body the impurity concentration as a function of
coordinate x and time t is given by
C(x; t) = C0 1 −
4

1X
k=0
(−1)k
(2k + 1)
exph−(2k + 1)2 2
4
Dt
l2 isin(2k + 1)
2
x
l : (1)
At the boundary x = l the concentration changes with time
C(l; t)=C0 1−
4

1X
k=0
(−1)k
(2k+1)
exph−(2k+1)2 2
4
Dt
l2 i :
(2)
After a transient period of time the terms with large k will
decay away and only the first term (with k = 0) will have
to be considered
C(l; t) = C0 1 −
4

exp −
2
4
Dt
l2 : (3)
Calculation (dashed line on Fig. 3) gives a good match
with the experimental results except for the last point.
Физика и техника полупроводников, 2004, том 38, вып. 3
360 S.K. Persheyev, P.R. Drapacz, M.J. Rose, A.G. Fitzgerald
Simulation results allowed to estimate of Cr diffusion in thin
a-Si :H films with a diffusion coefficient of 4 10−17 cm2/s,
which is quite slow compared to impurities such as Cu,
Sb, B and Ag. The reason is likely the processes at the
silicon and metal interface. Initially, during even short
deposition of a-Si :H on metal, Cr forms a silicide layer,
which creates a barrier for the diffusion and subsequently
prevents penetration of Cr atoms into the bulk. The diffusion
process is followed apparently with the growth of a silicide
layer, which gives a fast increase of Cr concentration at the
end of the annealing (t = 300 min).
The same experiments were carried out using vanadium
metal as the diffusion source, but even after 5 h annealing in
vacuum vanadium was not found to diffuse through 10 nm
thick amorphous silicon.
4. Conclusions
We have studied diffusion (at 400 C) of chromium
in 10 nm a-Si :H films by means of XPS and found that
it includes two processes. First of all silicide formation and
growh during annealing at the metal−silicon interface and
secondly silicide limited diffusion of Cr atoms through the
thin amorphous silicon film. We propose a new XPS based
method of diffusion analysis of contact materials on thin
films.
The authors acknowledge R.Sh. Malkovich for helpful
discussions. This work partially was supported by Royal
Society NATO Advanced Fellowship.
References
[1] J. Hu, J. Hajto, A.J. Snell, A.E. Owen, M.J. Rose. Phil. Mag. B:
Phys. Condens. Matter., 74 (1), 37 (1996).
[2] A. Kovsarian, J.M. Shannon, F. Cristiano. J. Non-Cryst. Sol.,
276, 40 (2000).
[3] H. Matsumura, M. Maeda, S. Furukava. J. Non-Cryst. Sol.,
59&60, 517 (1983).
[4] A. Polman, D.C. Jacobson, S. Coffa, A. Poate. Appl. Phys. Lett.,
57 (12), 1230 (1990).
[5] M.S. Ablova, G.S. Kulikov, S.K. Persheyev, K.K. Khodzhaev.
Semiconductors, 24 (11), 1208 (1990).
[6] J. Crank. The mathematics of diffusion (Oxford, Clarendon
Press, 1956) p. 58.
Редактор Т.А. Полянская
Физика и техника полупроводников, 2004, том 38, вып. 3