Spring meeting; Mat. Res. Soc. Symp. Proc., vol.762, 2003
Photoelectron spectroscopic investigations of very thin a-Si:H layers
M. Schmidt, A. Schoepke, O. Milch, Th. Lussky, W. Fuhs
Hahn Meitner Institut Berlin, Abteilung Silizium-Photovoltaik,
Kekuléstr. 5, D-12489 Berlin, Germany
ABSTRACT
We report on a detailed study on gap-state distribution in thin amorphous silicon layers (a-Si:H)
with film thicknesses between 5 nm and 20 nm on c-Si wafers performed by UV excited
photoelectron spectroscopy (UV-PES). We measured how the work function, the gap state
density, the position of the Fermi-level and the Urbach-energy depend on the layer thickness and
the doping level of the ultra thin a-Si:H(n) layers. It was found, that for phosphorous doping the
position of the Fermi level saturates at EF–EV=1.47 eV. This is achieved at a gas phase
concentration of 10000 ppm PH3 in the SiH4/H2 mixture which was used for the PECVD
deposition process. The variation of the doping level from 0 to 20000 ppm PH3 addition results
in an increase of the Urbach energy from 65 meV to 101 meV and in an increase of the gap state
density at midgap (Ei-EV= 0.86eV) from 3·1018 to 2·1019 cm-3eV-1.
INTRODUCTION
The energetic distribution of gap states, of interface states and the band offsets play a crucial role
in heterostructures. The detection of these quantities is possible by photo-excited electron
emission [1,2] if the escape depth corresponds to the top layer thickness. The excitation with
photon energies below 10 eV results in an increase of the escape depth up to 10nm [3]. Very thin
amorphous silicon layers play a fundamental role in hetero emitter solar cells of the TCO/aSi:H/c-Si type [4]. The thickness of such a-Si:H emitter layers amounts to about 5 nm [5]. They
are doped with boron or phosphorus depending on the type of the c-Si substrate. Additionally, a
undoped a-Si:H layer is often used for a better interface passivation of c-Si [4,5]. This layer is
located among the doped amorphous emitter layer and the c-Si absorber. Therefore, the
distribution of the gap state density of a thin a-Si:H layer which covers the c-Si surface, and their
change with doping and layer thickness are of outmost interest.
The basic properties of doped and undoped a-Si:H layers are well known [6,7] and are based on
many years research. But, the knowledge concerning the electronic properties of such very thin
amorphous layers grown on crystalline silicon substrates is patchy.
EXPERIMENTAL DETAILS
We used c-Si wafers (FZ) of p-type (75 Ohm cm) with a (111) surface orientation as substrates.
Prior to deposition of the amorphous layer the wafers were cleaned by the standard RCA process
followed by a HF-dip in 2% HF acid for 30 sec. The last step leads to a passivation of the Si
surface by hydrogen (H-termination) which is stable for about 45 minutes. This allows to insert
the samples into the deposition chamber without oxygen contamination of the silicon surface.
The deposition of the thin amorphous silicon layers was realized by the conventional 13.56 MHz
plasma enhanced chemical vapor deposition (PECVD) technique in a high vacuum system (10-7
mbar basic pressure) with a loadlock. Semiconductor grade silane (SiH4) with gas flow rates of
2.5 sccm, phosphine diluted in hydrogen (0.5%) and pure hydrogen were used as source gases.
The summarized gas flow rate amounts 7.5 sccm. This allowed a doping variation between 0 and
10000 ppm phosphine at nearly constant deposition conditions. The deposition was carried out
at a chamber pressure of 0.5 mbar, rf power of 55 mW cm-2 , and a substrate temperature of 170
°C. The layer thicknesses range between 2 and 20 nm as measured by ellipsometry.
Immediately after deposition, the a-Si:H/c-Si samples were transferred into the UHV system
where photoelectron spectroscopic investigations and annealing processes have been carried out.
Gap state density analysis
The process of photoelectron emission can be divided into three steps which are assumed to be
independent of each other [7]. The optical excitation of hot electrons (1), the transfer towards
the surface (2) and the emission into the vacuum (3). The yield Y in dependence on the kinetic
electron energy EK and photon energy h is given by equations 1.
Y(E K , h )  C  D(E)  δ(E k  E)  T(E, h )  P 2 (h )  N occ (E  h )  N u (E)
(1)
where
T(E, h ) 
λ imfp (E)/ α 1 h 
1  λ imfp (E)/ α 1 h 
and
P2 ( hν)  c
h  3.5 eV
P2 hν   c  hν 5
h  3.5 eV
describes the probability of the electron transfer towards the surface depending on the generation
depth given by the inverse optical absorption coefficient -1(hv) and the inelastic mean free path
imfp(E). P2(h) is the energy dependent optical matrix element [9]. D(E) describes the energy
dependent emission probability into the vacuum, (EK-E) denotes the analyzer detection function
and C is an apparatus constant.
We want to determine the energetic distribution of the occupied gap states and upper valence
band states, DOS, of the thin a-Si:H layer. Equation 1 shows that this is possible if all quantities
are nearly constant except the occupied (initial) states Nocc(E-h). This precondition is fulfilled
for a-Si:H because the DOS of the conduction band Nu (E) (unoccupied final states) is constant
[8] and the optical matrix element P2(h) remains nearly constant up to h = 3.5 eV and
decreases above this value as shown in detail in [9]. The electron interaction with phonons,
electrons and plasmons is summarized by the energy dependent inelastic mean free path length
imfp(E) which determines the limit of the electron emission depth. For photon energies below 10
eV the emission depth increases strongly because the strong inelastic plasmon generation process
becomes energetically impossible. Considering these arguments we can measure the DOS of the
a-Si:H in the excitation energy range between the work function limit at 3.8 eV and about 7 eV.
The corresponding emission depth ranges from 10 to 7 nm [3].
The photoelectrons were excited by strong monochromatic UV-light (4 –7 eV) generated by
passing the light of a Xe lamp through a double-grating monochromator. In this low excitation
energy range we are able to determine the number of absorbed photons using calibrated silicon
diodes for the detection of the incoming and reflected photon beams. This procedure allows to
measure the absolute photoelectron quantum yield at each photon energy, in two different
modes. In the first mode, corresponding to standard UPS, we detected the kinetic energy
distribution of the photoelectrons during excitation with a fixed photon energy [3]. The achieved
energy resolution was 100 meV, using a SPECS EA10P energy analyzer. In the other measuring
mode, the energy analyzer was operated at a fixed energy (final state energy) while changing the
photon energy (4-7 eV), a technique called Constant Final State Yield Spectroscopy (CFSYS)
[10]. Additionally, we determined the total yield as a function of the photon energy, the so
called Total Yield Spectrum ,TYS(h). Here, all emitted electrons for each excitation energy are
counted independent of their kinetic energy. The derivative of the TYS spectrum with respect to
the photon energy h represents the DOS, as shown in [1,2].
RESULTS
Figure 1 shows the yield and the gap state distribution of a 20 nm thick a-Si:H(n) layer on c-Si
measured by UPS, by CFSYS at two different analyzer energies and the derivative of the total
yield (DTY).
0
1
-EB [eV]
2
3
1E-4
1E22
Yield
1E-7
1E-8
1E-9
work function
(sample)
CFSYS, Efin=0.1eV
CFSYS, Efin=1.0eV
DTY
UPS
1E-10
1E19
1E18
1E17
1E16
1E-11
1E-12
3.5
1E20
-1
1E21
UPS
1E-6
-3
EF
1E15
4.0
4.5
5.0
5.5
6.0
Ephot [eV]
6.5
7.0
7.5
N(E) [cm eV ]
1E-5
Figure 1.
Comparison of UPS results
(Yintvs. Ekin) obtained at h=6.5
eV excitation (top scale;
-EB = Ekin - h - qAnalyzer) with
dYtot/dh (DTY) and with
CFSYS results vs. the photon
energy (bottom scale). The
CFSYS were performed at 0.1
and 1 eV above the vacuum
energy.
Measurements were performed
on an a-Si:H(150 nm)/c-Si
structure.
All spectra are in relatively good agreement but exhibit different low yield detection limits
(dynamic ranges) as shown in figure1. The dynamic range of the CFSYS spectra decreases with
increasing final state energy, and for UPS it is about two orders of magnitude lower than for
DTY and CFSYS (0.1 eV final state energy).
The reason for the different dynamic ranges is not yet fully understood. The energies of the
photons exciting electrons in the vicinity of the Fermi edge are different in UPS (6.5 eV) and
CFSYS (about 4.5 eV). Therefore, the optical matrix element P2(h) should be more than one
order of magnitude lower in UPS compared to CFSYS (and DTY). In addition, the photon flux
emitted from the Xe lamp decreases with rising photon energy. Both facts lead to a lower
detection sensitivity for electrons from near the Fermi edge in UPS.
On the other hand, electrons emitted from the vicinity of the valence band edge are excited by
photons of comparable energies in all detection modes and result in nearly identical yield values.
Furthermore, scattering processes like electron phonon interaction might lead to a higher
population of the low kinetic energy levels. The DTY method ignores this fact as long as the
electrons are able to be emitted. The relatively good agreement between DTY and CFSYS (0.1
eV final state energy) clearly shows that this effect is not dominant.
Thus, the energy dependent optical matrix element and its influence on the supply function of the
photoelectrons has to be proved. If this will be successfully performed, the detection of the
photoelectrons as a function of excitation energy is of advantage over the selective measurement
of the kinetic energy as used in UPS. Both methods DTY and CFSYS use this advantage. Work
is in progress to solve this problem.
We determined the position of the Fermi level in reference to the valence band, the slope of the
valence band tail states (Urbach energy Eov) and the change in the gap state distribution in
dependence on the doping level. The subsequently presented results are based on UPS
measurements with an excitation energy of h = 6.5 eV.
Figure 2 shows the comparison of the DOS for a nominally undoped a-Si:H(i) and an a-Si:H(n)
layer doped with 104 ppm phosphine. Three features can be clearly seen. The gap state
distribution shows the exponential band tail states, the DOS in the gap region increases by
nearly one order of magnitude and the position of the Fermi level shifts towards the conduction
band edge for the phosphorous doped layer.
1E23
-1
1E22
1E21
-3
N(E) / cm eV
Figure 2.
Comparison of the DOS between an
undoped and a phosphorus doped 10
nm thick a-Si:H layer deposited on a
c-Si wafer (111).
PH3 addition
0 ppm
4
10 ppm
1E20
1E19
EC
EF E
F
EV
1E18
1E17
-0.5
0.0
0.5
1.0
(E-EV) / eV
1.5
2.0
From the sqrt Y(E) vs. E slope results the starting point of the parabolic density of states
distribution at the valence band edge and allows to determine the valence band edge EV. In [2,9]
it was shown that the transition from the localized gap states to the extended valence band states
takes place at a density of states of NOV  2 1021 cm-3 eV-1. This value was used to scale the
photoemission yield with the DOS distribution. Furthermore, it is possible to determine the
Urbach energy EOV of the exponentially distributed valence band tail states from the slope of a
log-linear plot of N(E) according to equation 2.
N(E  E V )  N OV  exp( (E  E V )/E OV )
(2)
The obtained value of EOV reflects the disorder broadening of the valence band. The value of EOV
increases with disorder. The Urbach energy is EOV = 65.2 meV for the undoped sample and
increases to EOV=101.3 meV for the doped sample. The defect density is another measure of the
disorder and also increases with the band tail slope as can be seen in Fig.2. These values are at or
slightly above the upper limit compared to the values of thick films [6] and suggests a stronger
disorder of such extremely thin films or non optimized preparation conditions. The influence of
surface contamination like O or C can be ruled out as a reason as proved by XPS measurements.
Figure 3 shows typical results of the influence of annealing at 350 °C whereby effusion of
hydrogen takes place.
1E23
-1
1E22
1E21
-3
N(E) / cm eV
Figure 3.
Comparison of the DOS between a sample
as prepared and after annealing for 10 min
at 350 °C (H-desorption). The a-Si:H(n)
layer on a c-Si (111) wafer was
phosphorus doped and 10 nm thick.
PH3 addition
4
10 ppm, as prepared
4
10 ppm, H-desorption
1E20
EF
1E19
1E18
1E17
-0.5
EF
EC
EV
0.0
0.5
1.0
1.5
2.0
(E-EV) / eV
Two features stand out in figure 3, the shift of the Fermi level by nearly 300 meV towards
midgap after annealing and a decrease of the Urbach energy from EOV = 101.3 meV to EOV =
89.6 meV. This indicates that both the doping efficiency and the degree of disorder decrease
during the annealing process (H-desorption (effusion)).
Figure 4.
Dependence of the Fermi level position on the
phosphine addition during layer growth.
Change of these values after annealing of the
10-15 nm thick samples at 350 ° for 10 min.
The thickness of the samples varied between
10 and 15 nm.
The Fermi level shift saturates at about 1.5 eV above the valence band edge at a phosphine
addition of 10000 ppm. This corresponds to a value of EF = 0.27 eV below the conduction band
edge, assuming a value of the gap width of 1.74 eV as determined for 100 nm thick films.
The saturation of the Fermi level is a self limiting process caused by the defect generation with
energy levels located in the gap. The shift of the Fermi energy after annealing towards midgap is
clearly seen in figure 4 . The Urbach energy shows a similar behavior cf. figure 5. This is not
unexpected because the disorder and defect generation represents the background for this
behavior in both cases. The assumption of a constant optical gap of the layers before and after
annealing could not be proved. This will be done by spectral dependent ellipsometric
measurements in the future.
Figure 5.
Dependence of Urbach energy on the
phosphine content of the silane during the
layer growth. Change of these values after
annealing at 350 °C for 10 min. The
thicknesses of the samples varied between
10 and 15 nm.
CONCLUSIONS
Our results lead to the conclusion that in such extremely thin layers of a-Si:H on c-Si the basic
features like the electronic structure, the exponential tail states, the gap state distribution and the
self limitation of the doping efficiency are preserved. Special attention has to be directed to the
comparison of different measurement modes. The applied electron spectroscopic methods allow
to determine the basic electronic properties like DOS and (EF-EV) of such extremely thin a-Si:H
layers. The influence of energy dependent optical excitation cross sections and electron
scattering processes on the estimated DOS and the change of band gap with thickness and/or post
annealing processes are in our focus for future investigations.
REFERENCES
[1]
[2]
[3]
[4]
B.von Roedern, L.Ley, M.Cardona, F.W.Smith, Phil. Mag. B, 40, 433 (1979).
K.Winer, L.Ley, Phys. Rev. B, 36, 6072 (1987).
M. Prutton, Introduction to Surface Physics p.24, Clarendon press, Oxford 1994, p. 23.
M.Tanaka, M.Taguchi, T.Matsuyama, T.Sawada, S.Tsuda, S.Nakano, H.Hanafusa,
Y. Kuwano, Jpn. J. Appl. Phys. 31, 3518 (1992).
[5] M.L.Scherff, A.Froitzheim, A.Uljaschin, M.Schmidt, W.R.Fahrner, W.Fuhs, Proceedings
European Photovoltaic Conference, Rome (2002) (in press).
[6] R.A. Street, Hydrogenated amorphous silicon, Cambridge university press, Cambridge 1991;
Emis data review series No.19, Amorphous silicon and its alloys, ed. by T. Searl, INSPEC,
London 1998.
[7] W.E. Spicer, Phys. Rev. 112, 114 (1958).
[8] W.B.Jackson, S.M.Kelso, C.C.Tsai, J.W.Allen, S.-J.Oh, Phys. Rev. B, 31, 5187 (1985).
[9] W.B.Jackson, S.-J.Oh, C.C.Tsai, J.W.Allen, Phys. Rev. Lett. 53, 1481 (1984).
[10] M.Sebastiani, L.Di Gaspare, G.Capellini, C.Bittencourt, F.Evangelisti, Phys. Rev. Lett.
75, 3352 (1995).