ADVANCED PECVD REACTOR FOR THIN FILM SOLAR APPLICATIONS
A. Taha, D. Chaudhary, M. Klindworth, F. Leu, J. Martin, A. Salabas,
W. Wieland, D. Zorzi, Ch. Ellert
OERLIKON Solar Ltd., Trübbach
Hauptstras 1a, CH – 9477 Trübbach, Switzerland
e-mail: abed.
ABSTRACT: Oerlikon Solar introduced in 2007 the micromorph technology [1] into the photovoltaic market.
Oerlikon Solar’s strategy adapted the PECVD KAI deposition systems, approved in mass production and developed originally for TFT display applications, to thin film silicon solar cells. Hereby the ba of Oerlikon’s thin film silicon solar device technology origins from the long-term rearch carried out at the Institute of Microtechnique of the University of Neuchatel (IMT-NE) over the last 25 years on amorphous and microcrystalline silicon. This is bad on the well-known very high frequency glow disc
harge (VHF-GD) deposition technique [2]. Oerlikon Solar is developing and providing mass fabrication equipment including process for entire thin film silicon manufacturing lines for module areas of 1.3m x 1.1m. This has led to successful production lines on various customer sites in Europe and Asia. In order to successfully implement the solar cell and module process developed within R&D Oerlikon Solar into a production line, the PECVD reactor is continuously adapted and improved such that the process parameters can be transferred straightforward into the mass production equipment. The reactor development plays a key role to enable this transfer. Relevant parameters for optimum performance of the PECVD equipment are the uniformity of the deposition over the entire surface of the substrate as well as the deposition rate.
In addition to optimum deposition quality the reactors need to be cleaned after each deposition step in order to provide for any deposition run identical starting conditions. This in turn is prerequisite for stable and reproducible run to run deposition results.
1 INTRODUCTION
Year after year the interest of using Plasma Enhanced
Chemical Vapour Deposition (PECVD) method for
silicon bad thin films solar cells is growing rapidly.
This was triggered by the introduction of the micromorph
cell design in 1996 by Meier et al. [1]. The combination
of a microcrystalline with an amorphous cell allows the
improvement of the cell efficiency to far beyond 11 % in
the so called micromorph tandem cell. It was also proved
that using (PECVD) methods is high productiv and
economically viable to produce micromorph tandem cell.
For this purpo high plasma excitation frequency was
移动怎么换套餐required. In addition to keep the productive time of a potential deposition tool as high as possible the lf cleaning method developed for PECVD tools in LCD industry was transferred to the solar applications. In this paper we prent the necessary steps which led to the prently available KAI s
ystem of Oerlikon Solar and the improvements that was done in the 2nd generation of Oerlikon advanced (PECVD) reactor.
车水马龙的近义词2 MOTIVATION
In order to achieve high deposition rate in the PECVD steps of the thin film silicon solar very high RF-frequency was found very advantageous [2-5], e Figure 1 as taken from [6]. Figure 1: Increa of the deposition rate under equivalent experimental conditions in a PECVD reactor.
Higher frequencies do not only allow more efficient dissociation of the precursor molecules like SiH4but as well yield higher rates while keeping the ion bombardment on the deposited layers low [7,8].
The most challenging point in using high frequencies RF plasma is the non uniform voltage distribution across the electrodes due to what is know as standing wave effect which becomes significant when the dimensions of the electrodes become comparable to a quarter of the vacuum wave length of the electromagnetic wave at the lected RF-frequency [9].
In order to overcome this effect Schmitt [10] suggested already for the application in LCD technology 手机死机
the introduction of a compensating dielectric layer into one of the two flat electrodes facing the large area plasma slab.
3 TECHNICAL DESCRIPTION
The predominant limiting factor to a flat distribution of the deposition rate is the above mentioned standing wave effect. At Oerlikon Solar a frequency of 40 MHz was lected since it reprents the best combination
of
layer quality, deposition rate increa and availability of RF-equipment. As can be en by the exam
ple shown in Figure 2 the thickness of a deposited amorphous single layer is more than a factor of 2 higher in the center than on the edge, which is due to the peak of the electric field in the center of the reactor, i.e. the standing wave.
Figure 2: Distribution of the thickness of an amorphous silicon layer in a commercial large area PECVD reactor deposited on a glass substrate of 1100mm x 1300mm at 40MHz.
By introducing a dielectric layer into one of the electrodes a local voltage distribution
according to the respective capacitances of this layer and the plasma gap is achieved. The key point is the proper lection of the distribution of the width of that dielectric layer. A very schematic sketch is shown in Figure 3. It is important to cover the vacuum gap which reprents the dielectric compen
sating layer by a non-conducting material which prevents the plasma to fill that gap, which would eliminate the necessary voltage distribution over the two gaps.
Figure 3: Schematics of the design of the RF-electrodes, which allows overcoming the standing wave effect.
4 DEPOSITION RESULTS
Using the mentioned concept Oerlikon solar designed the first PECVD reactor generation in 2007 providing good layer quality for thin film solar applications using two different reactor types (one for amorphous layers and one for microcrystalline) - since the plasma conditions for the deposition of the two materials are quite different both require a different shape of the electrodes – After over 2 years of industrial u of this two reactor types, the collected experience leads to optimization of the hardware to get higher performance efficiency and process flexibility.
The main goal was to improve the gas usage to get more gas in the plasma zone (over the glass), then to reduce the powder formation and powder trapping as good as possible. Then the last step was to redesign the electrical field distribution (using the electrode shape) to decrea the RF loss and to optimize the one reactor concept to provide both layers (amorphous and microcrystalline)
with suitable deposition uniformity and material quality for solar cell applications. Thus, doors are opened for further process optimization and to get higher productivity.
In Figure 4 and Table 1 the thickness distribution of an amorphous layer is shown on 15x13 matrix of measured points. The uniformity is described by U = (Max – Min)/ (Max + Min), which means that all thickness value lie within the interval ±U.
Power DR DR
U% U% Reactor [W] (15X13)
(13X11)
(15X13) (13X11)
St.d aSi 370 3.33 3.34 13.16 6.95 One Box 370 3.19 3.23 14.35 6.24 One Box 400 3.42 3.47 13.83 5.81 One Box
440 3.63
3.66
14.64 6.03
Table 1: Comparison between amorphous layer uniformity in standard a-Si reactor box and in the new advanced plasma box with 20 mm edge exclusion.
Figure 4: Distribution of the thickness of an amorphous silicon layer in 1100mm x 1300mm glass using the advanced plasma Box
The new plasma box was also optimized to get the best microcrystalline layer quality and to increa the gas usage efficiency for mc-Si layers as what is shown in Table 2 and Figure 6.
Table 2: Comparison in performance capitalist between amorphous layers uniformity in standard aSi box and in the new advanced plasma box with 20 mm edge exclusion
Figure 5: Thickness distribution of microcrystalline layer deposited at about 5.5 A/s, with a uniformity U(15X13) = 8.9%.
However, in order to judge about the uniformity of a deposited microcrystalline layer not only the thickness but as well the Raman crystallinity (RC) is relevant, since the quality of the absorbing microcrystalline layer is critically determined by RC. This parameter is defined by
the ratio of the crystalline fraction (peaks at 520cm -1
东海县实验小学and
505cm -1
) with respect to the total respon including the
amorphous peak at 480cm -1
. The spectra are measured in a common Raman spectrometer. The distribution of the RC value over the glass substrate with an edge exclusion of 20 mm is shown in Figure 5.
Figure 5: Distribution of the Raman crystallinity of a microcrystalline layer. This distribution can be characterized by an interval of ±11% at low crystallinity level (0.56).
Figure 6 is a direct comparison between mc-Si layers done in the 1st and 2nd mc-Si reactor generations with the same Raman crystallinity. (around 0.6). It shows that less gas is needed to get the same deposition rate. In the same time less RF power needed to be delivered for the same mc-Si layer crystallinity. The two advantages allow the ur either to u the same deposition rate with less production costs or to increa the deposition rate to the new limits.
Figure 6: mc-Si deposition rate in PECVD reactor (V1.0 & V2.0 generation) with same Raman Crysta
llinity and SiH4 gas concentration
5 SUMMARY
In summary, the OC Oerlikon Solar PECVD reactor was described in its basic principles. The standing wave effect prent at 40MHz in conventional reactors was overcome, thus a flat thickness and crystallinity distribution was achieved for amorphous and microcrystalline layers.
那年盛夏
The upcoming 2nd generation of Oerlikon’s microcrystalline PECVD reactor will offer improved layer uniformity in thickness and crystallinity distribution as well as significantly better resource usage efficiency in terms of gas and RF power in the near future. As an addition feature it may as well replace the 1st generation of a-Si deposition reactors with its comparable growth rates and layer homogeneity.
[1] J. Meier et al., Proceedings of the 1st WCPEC,
Hawaii, USA, (1994) 409
[2] H. Curtins et al., Plasma Chem. Plasma Process. 7
排队机
(1987) 267
[3] R. Zedlitz, M. Heintze, G.H. Bauer, Proc. Mater. Res.
Soc. 258 (1992) 147
剃头师傅打一成语[4] A.A. Howling et al., J. Vac. Sci. Tech. A10 (1992)
1080
[5] W.G. Van Sark, J. Bezemer, W.F. van der Weg, Surf.
Coatings Tech. 74/75 (1995) 63
[6] F.Meillaud et al., Phil.Magazine, (2009) 1-22
[7] J. Dutta et al., J. Appl. Phys. 72, (1992) 3220
道理的英文[8] E.Amanatides, D.Mataras, D.E. Rapakoulias,
J.Vac.Sci.Techn A20(1), Jan/Feb 2002
[9] L Sansonnens et al. Plasma Sources Sci. Technol. 6,
(1997) 170-178
[10] J. Schmitt, Patent, US 6,228,438 B1, May 8, 2001
