Proceedings Article | 7 October 2014
M. Bär, D. Starr, A. Lambertz, B. Holländer, J.-H. Alsmeier, L. Weinhardt, M. Blum, M. Gorgoi, W. Yang, R. Wilks, C. Heske
KEYWORDS: Silicon, Oxides, Microcrystalline materials, Oxygen, Crystals, Solar cells, Spectroscopy, Transparency, X-rays, Hydrogen
Hydrogenated microcrystalline silicon oxide (μc-SiOx:H) layers are one alternative approach to ensure sufficient
interlayer charge transport while maintaining high transparency and good passivation in Si-based solar cells. We have
used a combination of complementary x-ray and electron spectroscopies to study the chemical and electronic structure of
the (μc-SiOx:H) material system. With these techniques, we monitor the transition from a purely Si-based crystalline
bonding network to a silicon oxide dominated environment, coinciding with a significant decrease of the material’s
conductivity.
Most Si-based solar cell structures contain emitter/contact/passivation layers. Ideally, these layers fulfill their desired
task (i.e., induce a sufficiently high internal electric field, ensure a good electric contact, and passivate the interfaces of
the absorber) without absorbing light. Usually this leads to a trade-off in which a higher transparency can only be
realized at the expense of the layer’s ability to properly fulfill its task. One alternative approach is to use hydrogenated
microcrystalline silicon oxide (μc-SiOx:H), a mixture of microcrystalline silicon and amorphous silicon (sub)oxide. The
crystalline Si regions allow charge transport, while the oxide matrix maintains a high transparency. To date, it is still
unclear how in detail the oxygen content influences the electronic structure of the μc-SiOx:H mixed phase material.
To address this question, we have studied the chemical and electronic structure of the μc-SiOx:H (0 ≤ x = O/Si ≤1)
system with a combination of complementary x-ray and electron spectroscopies. The different surface sensitivities of the
employed techniques help to reduce the impact of surface oxides on the spectral interpretation. For all samples, we find
the valence band maximum to be located at a similar energy with respect to the Fermi energy. However, for x > 0.5, we
observe a pronounced decrease of Si 3s – Si 3p hybridization in favor of Si 3p – O 2p hybridization in the upper valence
band. This coincides with a significant increase of the material’s resistivity, possibly indicating the breakdown of the
conducting crystalline Si network.
Silicon oxide layers with a thickness of several hundred nanometres were deposited in a PECVD (plasma-enhanced
chemical vapor deposition) multi chamber system using an excitation frequency of 13.56 MHz with a plasma power
density of 0.3 W/cm2. Glass (Corning type Eagle) and mono-crystalline silicon wafer substrates were coated in the same
run at a substrate temperature of 185°C. The deposition pressure was 4 mbar and the substrate-electrode distance 20 mm.
Mixtures of silane (SiH4), 1% TMB (B(CH3)3) diluted in helium, hydrogen (H2), and carbon dioxide (CO2) gases were
used at flow rates of 1.25 - 0.18/0.32/500/0 – 1.07) sccm (standard cubic centimeters per minute) for the deposition of
μc-SiOx:H(B) layers. By changing the CO2/SiH4 gas flow rate ratio from 0 to 6, μc-SiOx:H(B) layers with a composition
of 0 ≤ x = O/Si ≤ 1 were prepared using a constant sum of SiH4 and CO2. The TMB flow and the H2 flow were kept
constant within the series. For more details see Ref. [1].
The oxygen content in the films was determined using Rutherford Backscattering Spectroscopy (RBS). With RBS, the
area-related atomic density of oxygen and silicon can be determined (± 2% [2]), and thus x can be calculated. This
quantity considers only the number of silicon / oxygen atoms and not the number of atoms of other elements, such as
hydrogen, which is also incorporated to a considerable extent: up to 20% in μc-SiOx:H (measured using the hydrogen
effusion method). To avoid charging effects, the measurements were performed on films deposited on a substrate of
mono-crystalline silicon wafers.
The electrical conductivity was measured in the planar direction of the film in a vacuum cryostat, using voltages from
- 100 V to + 100 V. For that two co-planar Ag contacts were evaporated on the film with a gap of 0.5 mm 5 mm.
In the present study, the optical band E04 is arbitrarily used as a measure for the optical band gap. E04 is defined by the
photon energy E for which an optical absorption coefficient of α of 104cm-1 is obtained. The absorption coefficient α(λ)
versus the wavelength λ of the films was determined by measuring the transmittance T(λ) and reflectance R(λ), using the
Beer-Lambert law, as suggested by Ref. [3]. The film thickness d was measured using the step profiler close to the
measurement spot of the spectrophotometer. It is important to measure the transmittance T(λ) and the reflectance R(λ) at
the same spot on the sample, to avoid inaccuracies in the calculated absorption spectra that arise from non-uniformity of
the film thickness and different positions of the reflectance and transmittance minima and maxima in the spectrum [4].
Hard X-ray photoelectron spectroscopy (HAXPES) experiments were conducted at the HiKE end-station [5] on the
KMC-1 beamline [6] of the BESSY-II electron storage ring. This end-station is equipped with a Scienta R4000 electron
energy analyzer capable of measuring photoelectron kinetic energies up to 10 keV. A pass energy of 200 eV was used
for all measurements. Spectra were recorded with a photon energy of 2003 eV using the first and fourth order supplied
by a Si(111) double crystal monochromator. The combined analyzer plus beamline resolution is approx. 0.25 eV for
spectra taken at both photon energies. The top surface of the sample was electrically grounded for all measurements. The
binding energy was calibrated by measuring the 4f spectrum of a grounded Au foil and setting the Au 4f7/2 binding
energy equal to 84.00 eV. In SiO2, the inelastic mean free path of electrons was estimated to be approx. 5 and 13-16 nm
for the core levels and valence band measurements performed with 2003 and 8012 eV [7].
