Accepted Manuscript Application of Raman spectroscopy for depth-dependent evaluation of the hydrogen concentration of amorphous silicon
C. Maurer, S. Haas, W. Beyer, F.C. Maier, U. Zastrow, M. Hülsbeck, U. Breuer, U. Rau PII: DOI: Reference:
S0040-6090(18)30132-9 doi:10.1016/j.tsf.2018.02.037 TSF 36499
To appear in:
Thin Solid Films
Received date: Revised date: Accepted date:
12 July 2017 16 February 2018 23 February 2018
Please cite this article as: C. Maurer, S. Haas, W. Beyer, F.C. Maier, U. Zastrow, M. Hülsbeck, U. Breuer, U. Rau , Application of Raman spectroscopy for depth-dependent evaluation of the hydrogen concentration of amorphous silicon. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Tsf(2017), doi:10.1016/j.tsf.2018.02.037
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Application of Raman Spectroscopy for Depth-Dependent Evaluation of the Hydrogen Concentration of Amorphous Silicon
C. Maurera*, S. Haasa, W. Beyera, F. C. Maiera, U. Zastrowa, M. Hülsbecka, U. Breuerb,
IEK-5 Photovoltaik, Forschungszentrum Jülich GmbH, Leo-Brandt-Straße, 52425 Jülich,
RI
a)
PT
U. Raua
b)
SC
Germany
ZEA-3 Analytik, Forschungszentrum Jülich GmbH, Leo-Brandt-Straße, 52425 Jülich,
NU
Germany
*
MA
Corresponding author: e-mail
[email protected], Phone: +492461619748, Fax:
D
+492461613735
PT E
Abstract
The hydrogen concentration in hydrogenated amorphous silicon (a-Si:H), produced by
CE
plasma-enhanced chemical vapor deposition, was measured by Raman and Fourier transform
AC
infrared spectroscopy before and after thermal annealing. The possibility of Raman spectroscopy to perform depth dependent analysis in adapting the analysis wavelength is used. The findings presented in this paper are supported by secondary ion mass Spectrometry depth profiles. Moreover, we show that Raman spectroscopy can be used to get depthdependent information about the microstructure and the influence of annealing on the microstructure of a-Si:H.
ACCEPTED MANUSCRIPT Keywords: Raman Spectroscopy; Amorphous Silicon; Hydrogen; Microstructure; Fourier transform infrared spectroscopy; Secondary ion mass spectroscopy; Solar Cells
Introduction
PT
Amorphous silicon (a-Si:H) has a long tradition in solar cell and thin film transistor
RI
fabrication [1-3]. It is widely used in the production of large flat panel displays. For thin-film
SC
solar cells it is used as both, an absorber material and for surface passivation of crystalline silicon (c-Si). The first a-Si:H solar cell was demonstrated in 1976. Solar cells based on a-
NU
Si:H have been steadily improved until the end of the 1990’s. In the early 2000’s a-Si:H found another use for solar cells as passivation for silicon heterojunction solar cells and as
MA
precursor material for thin-film crystals, e.g. liquid phase crystallization (LPC) [1, 4, 5]. Hydrogen passivation plays a key role in producing a material suited as absorber material in
along with the a-Si:H microstructure are key factors that
PT E
The hydrogen concentration
D
solar cells. Hydrogen is needed in amorphous silicon in order to passivate midgap defects [5].
determine the quality of a-Si:H for electronic and opto-electronic applications [5-7]. Without
CE
this passivation of the dangling bonds in the amorphous network, a-Si:H is not suited for the use in solar cells [5, 7]. Although a-Si:H has been investigated over several decades, the
AC
effect of hydrogen in a-Si:H is still not fully understood [5, 8]. As the electrical and optical properties depend on the hydrogen concentration [5], the knowledge of the hydrogen concentration in a-Si:H is important. For example, the hydrogen concentration has a direct impact on the mobility gap and therefore on the absorption edge of a-Si:H solar cells. In LPC processes used for thin-film crystalline solar cells, the hydrogen concentration of a-Si:H has a huge impact on the quality and size of the crystallites. In this case, a low hydrogen
ACCEPTED MANUSCRIPT concentration is favorable, as high hydrogen concentrations lead to a high density of nuclei and with that to small nanocrystalline crystallites.
In this work we aim to develop non-destructive and fast methods to improve the understanding of hydrogen in a-Si:H by improving the substrate independent measurement of
PT
the hydrogen concentration and enabling a depth-dependent analysis. Typically
is either
RI
analyzed by hydrogen effusion, Secondary Ion Mass Spectroscopy (SIMS) or Fourier
SC
Transform Infrared Spectroscopy (FTIR). In addition to SIMS and FTIR, we use different Raman excitation wavelengths to gain information about the depth profile of the hydrogen
NU
concentration.
We also use Raman spectroscopy to measure the microstructure parameter. This
MA
quantity is defined by the ratio of the Si-H stretching mode absorption near 2090 cm-1 over the total Si-H stretching mode absorption. The microstructure parameter is considered as a
D
figure of merit for the material quality of a-Si:H [9]. The stretching mode absorption centered
PT E
at 2090 cm-1 is related to surface bound hydrogen in voids whereas the 2000 cm-1 mode can be attributed to hydrogen incorporated in a dense material [10, 11]. The microstructure
CE
parameter is usually measured by infrared absorption [6, 8, 12-16]. Therefore, the use of infrared transparent substrates is required. In general, either FTIR or Raman was used to
AC
determine the average hydrogen concentration or to obtain information about the average microstructure [17-20]. Raman spectroscopy is a fast, straight forward and non-destructive analysis method offering the opportunity of depth-dependent analysis using the law of Lambert-Beer for electromagnetic waves. The Si-H vibrational modes centred at 640 cm-1 (wagging mode) and the vibrational modes between
cm-1 and 2090 cm-1 (stretching modes) are both infrared
and Raman active [18, 21]. Raman is already used for in-situ growth analysis of
ACCEPTED MANUSCRIPT microcrystalline silicon [18, 19]. The use of Raman spectroscopy to analyze a-Si:H will allow monitoring the hydrogen concentration during deposition. Recently, Volodin and Koshelev presented both Raman spectroscopy as well as FTIR absorption data on a wide series of asdeposited a-Si:H films [22]. They determined the concentration of bonds with stretching vibrations at 2000 cm-1 and 2090 cm-1 with both methods. Using the approach developed by ) and Raman
) with the hydrogen concentration published by Volodin. The comparison of both
RI
(
PT
Langford [23] we determined the microstructure parameter for FTIR (
SC
microstructure parameters demonstrates that both methods lead to similar results (Fig. 1)
NU
emphasizing the usefulness of further investigations to compare those methods.
In this current work a comparison of Raman and FTIR as methods for measuring the
MA
bulk hydrogen concentration and microstructure as well as the depth dependent hydrogen concentration and microstructure of furnace annealed a-Si:H thin films on c-Si substrates is
D
made. The results of the depth dependent hydrogen concentration obtained with Raman
PT E
spectroscopy are compared with SIMS measurements. For easier reading
is used for Raman spectroscopy.
CE
microstructure determined with FTIR, while
is used for the
Experimental details
AC
In this work,
silicon wafers using a
µm thick a-Si:H thin-films were deposited on 4 inch crystalline nm thick SiO2 interlayer to avoid blistering and lift-off of the
amorphous silicon layer [24]. The a-Si:H films were grown by plasma-enhanced chemical vapor deposition, a technique wildly used for photovoltaic materials [25-28], at a pressure of mbar, a gas flow of 10 sccm silane (SiH4), 10 sccm hydrogen and a radio-frequency (rf) power (13.56 MHz) of concentration of
W. As a result, an a-Si:H layer with a hydrogen
at.% is obtained. After deposition, the wafer was cut into
cm²
ACCEPTED MANUSCRIPT samples that were annealed in an ultra-high vacuum oven with annealing temperatures ranging from
°C
°C. Heating and cooling was performed at high rates of
K/min with an annealing time of five minutes at the dedicated temperature. In this work, annealing is referred to as a thermal treatment. Phase changes are not expected for
PT
°C [29].
RI
The FTIR measurements were carried out in transmission with a ThermoElectron
(1)
is the FTIR absorption coefficient and
the wavenumber. Here we
NU
for FTIR, where
SC
Nicolet 5700 apparatus. The hydrogen concentration is calculated via
MA
assume that the proportionality constant, the absorption strength A, is independent of hydrogen concentration. Instead, this proportionality constant is not constant but increases slightly with rising H concentration [12]. As this influence is small compared to the
PT E
D
measurement uncertainty, we assume A as constant in this work.
The FTIR absorption spectra in the range between
cm-1 and 2090 cm-1 is subject
CE
to interferences due to the SiO2 interlayer. As a consequence the background subtraction is
AC
crucial for the determination of the hydrogen concentration and small deviations lead to high scattering. The microstructure parameter however, is calculated from the relative signal intensity of two neighboring stretching modes at
cm-1 and 2090 cm-1. Therefore, the
setting of the baseline does not have the same impact on the calculation of calculation of
, as for the
. The FTIR 640 cm-1 wagging mode was found to be less affected by
interference than the stretching modes. Consequently, the 640 cm-1 mode was used to determine the hydrogen concentration with FTIR. Additionally the 2000 cm-1 and 2090 cm-1 stretching modes were evaluated to calculate the
. These vibrational modes centered at
ACCEPTED MANUSCRIPT 640 cm-1 (wagging modes) and the vibrational modes between
m-1 to 2090 cm-1
(stretching modes) are both infrared and Raman active [18, 21]. In a recent work, Volodin determined the Raman scattering cross-section for Si-H and Si-H2 by comparing the Raman intensity with FTIR data [22].
nm and
nm. In order to estimate the Raman information
RI
wavelengths
PT
Raman spectroscopy was performed in backscattering mode for both excitation
SC
depth, the absorption of the a-Si:H used in this work was measured. Taken into account that the information depth is only half of the absorption depth, we determined an information nm at
nm and
29 nm at
NU
depth of
a HeNe Laser, with an excitation wavelength of
nm for example, the information
233 nm. Consequently, a more detailed and exhaustive depth
MA
depth is calculated with
nm, respectively. By choosing
dependent hydrogen concentration mapping would be obtained, facilitating the comparison
D
with FTIR. The use of different excitation wavelengths with deeper penetration depths would
PT E
give additional information about the average hydrogen concentration within the analyzed depth. However, when we used a laser with an excitation wavelength of
nm, we
CE
encountered several difficulties. First, we found that the stretching modes centered at
AC
cm-1 and 2090 cm-1 wavenumbers were superposed with an interference. Consequently, the setting of the baseline is not straight forward. Indeed, the use of optical simulation software would be required for precise analysis. Next, the signal to noise ratio was too low for peak-fitting even for measuring times of 90 minutes. As this is in contradiction to fast and straight forward analysis, we did not use the excitation wavelength of 647 nm in this work.
The Raman intensity of the Si-Hn wagging mode centered at than the stretching modes centered at
640 cm-1 is weaker
cm-1 and 2090 cm-1. In addition it is difficult to
ACCEPTED MANUSCRIPT separate the longitudinal acoustic overtone (2 LA) of a-Si from the Si-Hn wagging mode 640 cm-1 [30-32]. Therefore the sum of the Raman intensity of the stretching
centered at
modes (2000 cm-1 and 2090 cm-1) was used to determine the hydrogen concentration with Raman spectroscopy. As an example, the Raman spectra obtained at an excitation wavelength of
532 nm are presented in Fig. 2. In this graph, it can be seen that the intensity of the
PT
Raman shift decreases with increasing annealing temperature. The hydrogen concentration is
RI
proportional to the sum of the stretching modes intensity centered at 2000 cm-1 and 2090 cm-1
is the wavenumber, and
(2)
is the intensity of the sum of the Raman peaks
NU
where
SC
according to
of 2000 cm-1 and 2090 cm-1. Two Gaussian functions with peak positions centered at the two
MA
individual stretching modes were fitted to match the Raman intensity. In addition, these stretching modes are also used to determine
. In order to eliminate changes in signal
D
strength due to changes in reflectivity etc., the Raman spectra have been normalized. The
PT E
Raman signal around 2000 cm-1 composes of the two individual stretching modes centered at 2000 cm-1 and 2090 cm-1. For normalization for the experiments with an excitation nm we used the maximum of the sum of the two Si-H stretching
CE
wavelength of
absorptions, imitating the method used for the FTIR peak centred at 640 cm-1. However, this
AC
was not applicable for the excitation with
nm, as the spectrum of our setup does not
include this peak. As our samples do not exhibit significant variations in the reflectivity, the N2 peak centred at
cm-1 was used for normalization.
The microstructure parameter is derived for both FTIR and Raman measurements in the same way. The Raman microstructure parameter is calculated from Raman
ACCEPTED MANUSCRIPT measurements, imitating the method generally used to determine
using FTIR [9, 12]. The
microstructure parameter is calculated with (3) where
are the intensities at wavenumbers 2000 cm-1 / 2090 cm-1 of the FTIR
/
PT
absorption or the Raman intensity, respectively.
RI
SIMS measurements were carried out using a Time of Flight (ToF) SIMS (ION-TOF). µm² and the
SC
Cesium was used as primary ion source. The sputtered area was sputter dose was
Ions/cm². For analysis, Bi3 was used with a dose of
NU
Ions/cm². For the hydrogen concentration measurement, the relative intensity of the
MA
mass 31 and mass 30 signals was used as a measure of hydrogen concentration (Si-H/Si). In order to obtain a depth scale, the sputter time for reaching the film-substrate interface was set
D
equal to the film thickness.
PT E
Results
Fig. 3 shows the dependence of the hydrogen concentration on the annealing for both FTIR and Raman analysis. As can be seen in Fig. 3, the determined
CE
temperature
AC
decreases with increasing annealing temperature
for both Raman excitation wavelengths
(stars) used in this work and FTIR (squares). For all samples, the as-deposited reference is plotted at the deposition substrate temperature
180 °C. The hydrogen concentration
was normalized to the as-deposited reference for all analysis methods. Please note, that there was no phase change involved due to the thermal treatment. Even though
has been
reduced considerably, there is no crystalline part in the Raman spectra and the thin-film is still amorphous. Thus, we expect only minor changes in the absorption for photon energies above the mobility gap.
ACCEPTED MANUSCRIPT
The hydrogen concentration derived from Raman spectra with an excitation wavelength of
488 nm is shown with blue stars, while those for the excitation with
532 nm are illustrated with green stars. The standard FTIR measurements show a strong decrease in the hydrogen concentration starting at temperatures
450 °C. In contrast, the
200 °C,
400 °C,
500 °C
RI
The SIMS depth profiles for annealing temperatures of
PT
Raman measurements exhibit an onset of hydrogen loss already at much lower temperatures.
SC
are presented in Fig. 4.
NU
According to former works, the surface as well as the a-Si/substrate interface should work as diffusion sinks [6, 33-36] with a uniform distribution of the hydrogen concentration
MA
throughout the a-Si:H layer for the as-deposited reference. However, the SIMS depth profiles show a significant increase of the hydrogen concentration towards the a-Si:H/SiO2 interface 200 °C Fig. 4. As a consequence, the measured
is higher
D
for the sample annealed with
PT E
for all annealing temperatures close to the interlayer interface than to the external surface. The depletion of hydrogen close to the surface is also detected when comparing the hydrogen
CE
concentration obtained by FTIR and Raman with different excitation wavelengths. As shown in Fig. 3, FTIR analysis results in the highest measured
(black squares), while the Raman .
AC
measurements with the shortest penetration depth (blue stars) exhibit the smallest
The SIMS results show (Fig. 4) that the as-deposited sample has an initial hydrogen enrichment towards the SiH/SiOx interface instead of the expected symmetrical concentration throughout the whole a-Si:H layer. When investigating the monitored deposition parameters we found pressure instabilities during the SiO2 deposition. This leads to the conclusion, that there is no stoichiometric SiO2 interlayer, but a SiOx interlayer, that influences the growth of
ACCEPTED MANUSCRIPT a-Si:H. However, this anomaly does neither influence the Raman and FTIR analysis, nor the possibility to compare those methods.
The SIMS hydrogen concentration, as listed in Table 1, is calculated for Raman considering the depth dependent signal strength with
PT
depth where the signal is obtained,
to the mass signal
NU
is the relation of the mass signals
SC
RI
and
where
(4)
the film thickness, and
(5) ,
is the calculated
is the absorption coefficient
MA
for the Raman wavelength. Due to the sputtering of the SIMS analysis all bondings are initially broken in the investigated area. After that molecules can form again close to the (mass 31 u) to the mass signal
(mass 30 u),
D
sample surface. Here, the relation of
PT E
was used to determine the relative hydrogen content. Similar methods are used for deuterated samples [14].
CE
Please note, that other isotopes and molecules than to the analyzed SIMS signals. For example, for the mass of
u,
do also contribute as well as
contribute to the measured signal. However, we found a linear relationship
AC
and
and
between the integrated value of
and the hydrogen concentration detected by
FTIR. Therefore, equation 4 can be used to calculate the hydrogen concentration. Hereby, the hydrogen concentration that is detected using Raman is calculated for the given depthdependent hydrogen concentration measured with SIMS (equations 4, 5). Therefore, the SiH counts were set into relation to the Si counts (Fig. 4).
ACCEPTED MANUSCRIPT On the other hand, the SIMS hydrogen concentration calculated for FTIR is the mean over the film thickness. After that,
was normalized to
for the measurement of the sample
annealed at 200 °C for all experiments. When comparing the obtained values for the hydrogen concentration for SIMS integration and Raman or FTIR, we find that the deviation
500 °C, and
200 °C,
400
RI
°C,
was calculated from the SIMS measurements for
575 °C. In order to obtain the relative change of the hydrogen
concentration, the sample annealed at
SC
In addition
532 nm.
PT
is rather small especially for FTIR and the Raman measurements at
200 °C was chosen as reference hydrogen
NU
concentration. The hydrogen concentration obtained for FTIR, as well as for Raman, is listed in Table 1. In order to analyze the depth-dependent hydrogen concentration, the absorption
MA
depth of 45 nm and 29 nm was taken into account for the excitation with 532 nm (green) and 488 nm (blue), respectively. The hydrogen concentration derived with SIMS retrieves similar
D
values as FTIR and Raman spectroscopy. In particular, FTIR and Raman spectroscopy with
PT E
an excitation wavelength of 532 nm show a good agreement and strong correlation with the SIMS results. The hydrogen concentration determined at 488 nm is overall smaller for Raman
CE
than for SIMS.
AC
The derived annealing temperature dependent microstructure parameters
and
, are illustrated in Fig. 5. All characterization methods show first a decrease of the microstructure parameter to a local minimum. This minimum of the microstructure parameter is followed by an increase, exceeding the initial microstructure parameter. After reaching the maximum, the microstructure parameter decreases again. The FTIR microstructure parameter is maximal centered at
500 °C. The maximum of the Raman microstructure
parameters is shifted to lower temperatures. The evolution is similar for FTIR and both
ACCEPTED MANUSCRIPT Raman excitation wavelengths. However, the calculated values for the microstructure parameter are shifted towards higher values for longer excitation wavelengths. Hence, the values derived with Raman and an excitation wavelength of
488 nm are smallest, while
FTIR yields the highest values. Regarding the relative change of the microstructure parameter from the as-deposited value to the maximum,
increases by 69 %,
RI
PT
by 42 %.
by 66 %, and
SC
In order to obtain information about the depth dependent distribution of the hydrogen in the amorphous silicon layer, SIMS measurements were performed for four different
NU
annealing temperatures. The distribution of hydrogen in the a-Si:H layer is shown in Fig. 4. Close to the surface, the hydrogen concentration decreases over one order of magnitude with
D
Discussion
MA
increasing annealing temperature.
PT E
As presented in the experimental part, the hydrogen concentration was determined by Raman spectroscopy is smaller than the one determined by FTIR absorption (Fig. 3). This
CE
finding is in agreement with the SIMS depth profile (Table 1), considering the wavelength dependent absorption of amorphous silicon. The as-deposited a-Si:H used for this work cm-1 for a wavelength of
AC
exhibits an absorption coefficient of 488 nm and of
cm-1 for
532 nm, respectively. Taken into
account that the information depth of Raman in backscattering mode is half of the optical penetration depth 1/, the information depth is
29 nm and
45 nm,
respectively. Under consideration of the plateau shaped nature of diffusion profiles, the hydrogen concentration decreases towards an interface that works as diffusion sink [13, 36].
ACCEPTED MANUSCRIPT Raman signals are more sensitive to the diffusion sink as the region close to the interface is more As mentioned above, the determined hydrogen concentration is in agreement for the spectroscopic methods Raman and FTIR, and SIMS. Considering the SIMS profile of the sample annealed at
200 °C (Fig. 4) two results can be noted. First, there is no measurable 200 °C, as the SiH/Si intensity does not decrease
PT
hydrogen diffusion for annealing at
200 °C samples can
RI
towards the external surface. Consequently, the measurements of the
SC
be also used as reference. Secondly is indicated that the Raman reference for the excitation wavelength of
200 °C is not detected in the SIMS profile (Fig. 3). A lower hydrogen concentration
NU
for
488 nm is estimated too high, as the drop of the hydrogen concentration
for the as deposited reference would shift all subsequent hydrogen concentrations to higher
MA
values. As a result, the hydrogen concentration retrieved with an excitation wavelength of 488 nm would approach the values obtained with SIMS.
D
In summary, the SIMS measurements proofed that Raman spectroscopy can be used
PT E
to perform a depth dependent analysis of the hydrogen concentration of amorphous silicon. However, the determination of the hydrogen concentration of a reference is crucial for Raman
CE
analysis. In Fig. 3 for example, the hydrogen concentration obtained with Raman spectroscopy for 532 nm is higher for samples annealed at
250 °C than for
200 °C.
AC
This is another indication for a poorly determined reference hydrogen concentration or a local deviation as an increase of the hydrogen concentration due to thermal treatment is not expected. For the shorter excitation wavelength of for
488 nm, we find a strong local drop of
250 °C. This might result either out of a deviation for the reference or an
irregularity of the
200 °C sample that leads to a smaller signal. Here it has to be taken
into account that the measurements for different annealing temperatures were not taken at the same sample position. As a consequence layer inhomogeneities can also lead to evaluation
ACCEPTED MANUSCRIPT uncertainties. For annealing temperatures below
300 °C, the diffusion length
is in
the order of several nm or smaller. Accordingly, the hydrogen concentration cannot change markedly for annealing temperatures below
300 °C.
The course of the microstructure parameter, depending on the annealing temperature
PT
is comparable for Raman and FTIR (Fig. 5). The increase of the microstructure parameter
RI
with increasing annealing temperature has been explained by an increase in void
SC
concentration due to an inhomogeneous densification (shrinkage) of the material [15] that is resulting of the hydrogen out-diffusion.
temperatures below
°C and above
NU
The nature of the decrease of the microstructure parameters for annealing °C is not yet explained. It has to be
MA
remarked that the microstructure parameter was first understood as the ratio between the sum of Si-H2 and Si-H3 bonds to the number of Si-H bonds [6, 8, 12-16]. As a consequence to this
D
historical definition, the stretching modes centered at 2000 cm-1 and 2090 cm-1 are still often
PT E
attributed to Si-H and Si-Hn (n = 2; 3) bondings, respectively. However, this does not affect the microstructure parameter as a figure of merit.
CE
As mentioned before, the calculated values of the microstructure parameter are smaller for Raman than for FTIR (
. This is in agreement with hydrogen
AC
depletion towards the film surface and with a direct dependence of the microstructure parameter on hydrogen concentration [15] for our annealed samples. This emphasizes the finding, that
changes within the layer and that the microstructure parameter directly
depends on the hydrogen concentration for our thermally annealed samples. Moreover, due to the growth properties of amorphous silicon a variation of the microstructure parameter throughout the thin film is reasonable [5, 37]. Hence additional information is obtained in using Raman with different optical penetration depths.
ACCEPTED MANUSCRIPT Nevertheless, it has to be taken into account that the intensity of the peaks decreases with decreasing
, leading to a reduced accuracy in the derived hydrogen concentration. The
intensity of the Raman signal decreases with decreasing
(Fig. 2). Consequently, the signal-
to-noise ratio is reduced and the fitting of the signal area is more difficult leading to a higher inaccuracy. With regard to using the microstructure parameter as a figure of merit for
PT
potentially good solar cell absorber material [7], the absolute values of the microstructure
RI
parameter derived with Raman are not competitive with those obtained with FTIR. However,
SC
the course of both methods is competitive. Accordingly, Raman can also be used to qualify if
NU
quantity of voids decreases or increases due to a post-deposition treatment.
Conclusion
MA
In this work we presented a comparison of Raman and FTIR spectroscopy for the evaluation of the hydrogen concentration and the microstructure parameter for hydrogenated
D
amorphous silicon thin films. It could be shown that both methods lead to comparable results
PT E
regarding the hydrogen concentration of amorphous silicon considering the analyzed depth. Furthermore, the influence of annealing on the microstructure can be qualified with both
CE
methods. Although the values of the microstructure parameter are not equal for annealed samples for both methods, the annealing temperature dependence is comparable. This is
AC
partially attributed to the limited penetration depth of Raman spectroscopy, depending on the excitation wavelength. In contrast, Raman spectroscopy can be used to make a depth selective analysis on
and
by adapting the excitation wavelength to the depth of
interest. Therefore, surface analysis is feasible by choosing a wavelength in the UV-region that will be absorbed within the first few nanometers. Further experiments should be performed to analyze whether UV-Raman spectroscopy with a very short penetration depth can be used to qualify the a-Si:H passivation of modern solar cell concepts with passivated
ACCEPTED MANUSCRIPT contacts. Raman spectroscopy can be used helping to understand and analyze the growth process of a-Si:H, as depth dependent changes and irregularities can be detected in a thin film at any production stage.
Acknowledgements
PT
The authors would like to thank A. Lambertz and S. Moll for sample preparation. Part of the
SC
RI
work was financed by BMU (project no. 0325446B).
References
AC
CE
PT E
D
MA
NU
[1] NREL, Best Research-Cell Efficiencies, in: Best Research-Cell Efficiencies, National Renewable Energy Laboratory, https://www.nrel.gov/pv/assets/images/efficiency-chart.png, 2017. [2] S. Muthmann, F. Köhler, R. Carius, A. Gordijn, Structural order on different length scales in amorphous silicon investigated by Raman spectroscopy, physica status solidi (a), 207 (2010) 544-547. [3] Y. Kuo, Thin Film Transistors: Amorphous silicon thin film transistors, Kluwer Academic Publishers, 2004. [4] T. Sawada, N. Terada, S. Tsuge, T. Baba, T. Takahama, K. Wakisaka, S. Tsuda, S. Nakano, High-efficiency a-Si/c-Si heterojunction solar cell, in: 1994 IEEE First World Conference on Photovoltaic Energy Conversion/Conference Record of the Twenty Fourth IEEE Photovoltaic Specialists Conference-1994, Vols I and Ii, 1994, pp. 1219-1226. [5] A. Shah, Basic Properties of Hydrogenated Amorphous Silicon (a-Si:H), in: A. Shah (Ed.) Thin-film silicon solar cells, EFPL Press, Lausanne, 2010, pp. 17-96. [6] W. Beyer, U. Breuer, R. Carius, W. Hilgers, D. Lennartz, F.C. Maier, N.H. Nickel, F. Pennartz, P. Prunici, U. Zastrow, Influence of hydrogen concentration on void-related microstructure in low hydrogen amorphous and crystalline silicon materials, Can J Phys, 92 (2014) 700-704. [7] M. Zeman, R.E.I. Schropp, Amorphous and microcrystalline silicon solar cells : modeling, materials and device technology., Kluwer Academic Publ., Boston, Mass., 1998. [8] N. Budini, P.A. Rinaldi, J.A. Schmidt, R.D. Arce, R.H. Buitrago, Influence of microstructure and hydrogen concentration on amorphous silicon crystallization, Thin Solid Films, 518 (2010) 5349-5354. [9] A.H. Mahan, P. Raboisson, D.L. Williamson, R. Tsu, Evidence for microstructure in glow-discharge hydrogenated amorphous SiC alloys, Solar Cells, 21 (1987) 117-126. [10] H. Wagner, W. Beyer, Reinterpretation of the Silicon-Hydrogen Stretch Frequencies in Amorphous-Silicon, Solid State Commun, 48 (1983) 585-587. [11] M. Cardona, Vibrational Spectra of Hydrogen in Silicon and Germanium, physica status solidi (b), 118 (1983) 463-481. [12] W. Beyer, M.S.A. Ghazala, Absorption Strengths of Si-H Vibrational Modes in Hydrogenated Silicon, MRS Proceedings, 507 (2011) 601-606.
ACCEPTED MANUSCRIPT
AC
CE
PT E
D
MA
NU
SC
RI
PT
[13] W. Beyer, Chapter 5 - Hydrogen Phenomena in Hydrogenated Amorphous Silicon, in: H.N. Norbert (Ed.) Semiconductors and Semimetals, Elsevier, 1999, pp. 165-239. [14] W. Beyer, J. Bergmann, U. Breuer, F. Finger, A. Lambertz, T. Merdzhanova, N.H. Nickel, F. Pennartz, T. Schmidt, U. Zastrow, Comparison of Laser and Oven Annealing Effects on Hydrogen and Microstructure in Thin Film Silicon, MRS Online Proceedings Library Archive, 1770 (2015) 1-6. doi:10.1557/opl.2015.431 [15] W. Beyer, W. Hilgers, D. Lennartz, F.C. Maier, N.H. Nickel, F. Pennartz, P. Prunici, Effect of Annealing on Microstructure in (Doped and Undoped) Hydrogenated Amorphous Silicon Films, in: doi:10.1557/opl.2014.667 (Ed.) MRS Proceedings, 2014. doi:10.1557/opl.2014.667 [16] W. Beyer, Microstructure characterization of plasma-grown a-Si : H and related materials by effusion of implanted helium, J. Non-Cryst. Solids, 338 (2004) 232-235. [17] M. Ivanda, Raman-scattering measurements and fraction interpretion of vibraitonal properties of amorphous silicon, Physical Review B, 46 (1992) 14893-14896. [18] M. Ivanda, O. Gamulin, K. Furic, D. Gracin, Raman-study of light-induced changes in silicon-hydrogen bonds stretching vibrations in a-Si:H, Journal of Molecular Structure, 267 (1992) 275-280. [19] T. Itoh, K. Yamamoto, K. Ushikoshi, S. Nonomura, S. Nitta, Characterization and role of hydrogen in nc-Si : H, J. Non-Cryst. Solids, 266 (2000) 201-205. [20] K.O. Bugaev, A.A. Zelenina, V.A. Volodin, Vibrational Spectroscopy of Chemical Species in Silicon and Silicon-Rich Nitride Thin Films, International Journal of Spectroscopy, 2012 (2012) 281851. [21] P.V. Santos, N.M. Johnson, R.A. Street, Light-enhanced hydrogen motion in a-Si:H, Phys Rev Lett, 67 (1991) 2686-2689. [22] V.A. Volodin, D.I. Koshelev, Quantitative analysis of hydrogen in amorphous silicon using Raman scattering spectroscopy, J. Raman Spectrosc., 44 (2013) 1760-1764. [23] A.A. Langford, M.L. Fleet, B.P. Nelson, W.A. Lanford, N. Maley, Infrared-Absorption Strength and Hydrogen Content of Hydrogenated Amorphous-Silicon, Physical Review B, 45 (1992) 13367-13377. [24] W. Beyer, F. Einsele, Hydrogen Effusion Experiments, in: D. Abou-Ras, T. Kirchartz, U. Rau (Eds.) Advanced characterization techniques for thin film solar cells 2, Wiley-VCH Verlag, Weinheim, 2016, pp. 569-595. [25] O. Gabriel, T. Frijnts, S. Calnan, S. Ring, S. Kirner, A. Opitz, I. Rothert, H. Rhein, M. Zelt, K. Bhatti, J.H. Zollondz, A. Heidelberg, J. Haschke, D. Amkreutz, S. Gall, F. Friedrich, B. Stannowski, B. Rech, R. Schlatmann, PECVD Intermediate and Absorber Layers Applied in Liquid-Phase Crystallized Silicon Solar Cells on Glass Substrates, IEEE Journal of photovoltaics, 4 (2014) 1343-1348. [26] Z. Zang, A. Nakamura, J. Temmyo, Nitrogen doping in cuprous oxide films synthesized by radical oxidation at low temperature, Materials Letters, 92 (2013) 188-191. [27] V.S.S. Paulo M. Gordo, Marco Duarte Naia, C. Lopes Gil, Adriano P. de Lima, G. Lavareda, C. Nunes de Carvalho, A. Amaral, Role of the RF Power on the Structure of Defects in a-Si:H Films Produced by PECVD, in: G.K.a.P.S. Werner Triftshäuser (Ed.) Positron Annihilation - ICPA-12, pp. 454-456. [28] Z.G. Zang, A. Nakamura, J. Temmyo, Single cuprous oxide films synthesized by radical oxidation at low temperature for PV application, Optics Express, 21 (2013) 11448-11456. [29] S. Koc, M. Závětová, J. Zemek, Physical properties of amorphous Si:H The role of annealing, Czech J Phys, 25 (1975) 83-90. [30] M.H. Brodsky, M. Cardona, J.J. Cuomo, Infrared and Raman-spectra of siliconhydrogen bonds in amorphous silicon prepared by glow-discharge and sputtering, Physical Review B, 16 (1977) 3556-3571.
ACCEPTED MANUSCRIPT
AC
CE
PT E
D
MA
NU
SC
RI
PT
[31] N. Liao, W. Li, Y. Kuang, Y. Jiang, S. Li, Z. Wu, K. Qi, Raman and ellipsometric characterization of hydrogenated amorphous silicon thin films, Sci. China Ser. E-Technol. Sci., 52 (2009) 339-343. [32] A. Shih, S.-C. Lee, C.-t. Chia, Evidence for coupling of Si–Si lattice vibration and Si–D wagging vibration in deuterated amorphous silicon, Applied Physics Letters, 74 (1999) 33473349. [33] W. Beyer, H. Wagner, The role of hydrogen in a-Si:H - results of evolution and annealing studies, J. Non-Cryst. Solids, 59/60 (1983) 161-168. [34] W. Beyer, H. Wagner, Determinataion of the hydrogen diffusion coefficient in hydrogenated amorphous silicon from hydrogen effusion experiments, J. Appl. Phys., 53 (1982) 8745-8750. [35] R.A. Street, C.C. Tsai, J. Kakalios, W.B. Jackson, Hydrogen diffusion in amorphous silicon, Philosophical Magazine Part B, 56 (1987) 305-320. [36] M. Reinelt, S. Kalbitzer, G. Moller, Proceedings of the Tenth International Conference on Amorphous and Liquid Semiconductors The diffusion of hydrogen in amorphous silicon, J. Non-Cryst. Solids, 59 (1983) 169-172. [37] S. Yamasaki, T. Umeda, J. Isoya, K. Tanaka, Existence of surface region with high dangling bond density during a-Si : H film growth, J. Non-Cryst. Solids, 227 (1998) 83-87.
ACCEPTED MANUSCRIPT Fig. 1 Calculated microstructure parameter for FTIR and Raman data obtained from the work of Volodin [22]. The linear fit of all data points is plotted with the light blue line. Fig. 2 Raman spectra of the stretching modes centered at 2000 cm-1 and 2090 cm-1 for the excitation with 532 nm. The decrease of the normalized Raman intensity with increasing annealing temperature is indicated with the black arrow.
PT
Fig. 3 Relative reduction of the hydrogen concentration with increasing annealing temperature. The hydrogen concentration is derived from the integrated peak area of the FTIR modes centered at 640 cm-1 (black squares).
RI
The corresponding hydrogen concentration obtained from the Raman modes centered at 2000 cm-1 and
SC
2090 cm-1 is shown in green stars for excitation with 532 nm and blue stars for excitation with 488 nm. Fig. 4 SIMS depth profiles of the relative Si-H bonding concentration for annealing temperatures 200 °C
NU
575 °C
MA
Table 1: Comparison of the determined hydrogen concentration
for SIMS and optical characterization
methods. Here, the hydrogen concentration obtained with Raman and FTIR is normalized to the hydrogen 200 °C.
D
concentration obtained for the annealing at
measurements (
PT E
Fig. 5 a) Calculated microstructure parameter derived from FTIR (
) black squares. b) refers to the Raman
). Blue symbols denote the Raman microstructure parameter derived from the
CE
measurements with an excitation wavelength of 488 nm while green symbols refer to an excitation wavelength of 532 nm. For all samples, the as-deposited reference is plotted at the deposition substrate temperature
AC
180 °C.
ACCEPTED MANUSCRIPT
575 °C
1
0.96
0.72
0.29
1 1
0.89 0.76
0.71 0.55
0.23 0.09
1
0.77
0.56
0.12
1
0.73
0.5
0.08
1
0.68
0.35
0.1
PT E
D
MA
NU
SC
RI
500 °C
CE
FTIR SIMS 532 Raman 532 SIMS 488 Raman 488
400 °C
AC
SIMS FTIR
200 °C
PT
Table 1
ACCEPTED MANUSCRIPT Highlights
AC
CE
PT E
D
MA
NU
SC
RI
PT
Analysis of hydrogenated amorphous silicon (a-Si:H) Depth dependent analysis of hydrogen concentration in a-Si:H with Raman Proof of the Raman results using SIMS Evaluation of a Raman microstructure parameter Comparison of the Raman microstructure with FTIR microstructure
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5