Project C1: Magnetic flow control in growth and casting of photovoltaic siliconPI: Gunter Gerbeth (HZDR)
Partners: HZDR, TU-BAF
1. Scientific case for the project
1.1 BackgroundThe mass production of photovoltaic silicon (pv-Si) is today based on directional solidification of multi-crystalline ingots or on the Czochralski (Cz) growth of mono-crystalline (mc) ingots . The yield of usable wafers as well as the efficiency of the solar cells produced from the ingots are detrimentally affected by system immanent impurities, mainly carbon, nitrogen, and oxygen. SiC precipitates or filaments, for instance, have been identified to cause ohmic shunts in solar cells based on multi-crystalline silicon , whereas the phenomenon of light-induced degradation appearing mainly in Cz-based solar cells is ascribed to oxygen-related complexes . Therefore, defect engineering to control and reduce the contamination level in melt and crystal is one of the major challenges crystal growers are faced with in optimisation of pv-Si growth.
In conventional growth systems, the impurities mentioned above are always present due to the use of graphite heaters and SiO2-crucibles/SixNy(O)-crucible coatings. Their concentration and spatial distribution result from defect reactions at the melt – crucible interface as well as at the melt surface in contact to the atmosphere, and from the level of melt mixing induced by a convective flow. At present, most industrial furnaces for directional solidification of multi-crystalline silicon are designed just in such a way that the heating is done by resistance heaters with a low melt mixing leading to an unfavourable agglomeration and precipitation of impurities in certain regions of the melt and crystal . Several strategies of electromagnetic or mechanical flow control are currently under consideration to overcome this problem. Among the approaches proposed, melt stirring by the magnetic field of a specially designed internal heating system appears to be the most promising one, at least with respect to possible applications in industrial growth . However, systematic investigations of the impact of an additional melt stirring on the defects in multi-crystalline silicon have not yet been published.
The further development of the Cz-growth technique, either for photovoltaic or microelectronic purposes, depends strongly on a better understanding and control of the melt flow in the crucible and the related oxygen transport. For that various types of magnetic fields have been considered during last years , but only some of them have received industrial realization. Today either sophisticated combined AC-DC fields or just a strong transverse DC field are used in industry. Surprisingly, the simple version of a rotating magnetic field (RMF) has never been tested on real scale, though it may provide a very efficient reduction of turbulent temperature fluctuations and a considerable stabilization of the flow in the core of the melt.
1.2 Most important goals of the planned workThe general goal of the proposed project is the increase of the crystal quality and the yield of usable wafers in directional solidification as well as Cz growth of pv-Si. The main objectives of the work on directional solidification of multi-crystalline silicon are the control and reduction of the contamination level in melt and crystal on the basis of an extensive, numerical and experimental study of the defect interaction in the growth system, which is still little understood. Numerical simulations including thermodynamics and species transport as well as growth experiments in a laboratory scale furnace under systematically varied conditions (melt flow, composition of the atmosphere and crucible coating) are planned to identify and analyse the relevant impurity reactions and their impact on the parameters of the grown crystals with focus on
- the resolution and/or release of impurities at the melt – crucible and melt – atmosphere interfaces,
- the formation and agglomeration of inclusions in melt and crystal, and
- the axial and radial segregation of impurities in the crystal.
In case of Cz-growth the recent studies at HZDR lead to the clear conclusion that the use of an RMF of rather low strength should give the most promising results. This has up to now never been tested on industrial scale. To prepare this step, the project will realize experiments at a Cz-model of almost realistic scale available at HZDR. It involves crucible and crystal rotation as well as temperature gradients and heat fields close to Cz reality. Those experiments will be done at the MULTIMAG facility of HZDR allowing the independent superposition of all relevant magnetic fields (rotating, traveling, pulsating, DC-unifom, DC-cusp). The flow fields in those model experiments will be measured using UDV and local electromagnetic velocity probes.
2. Existing competencies and infrastructureThe present project is a joint research activity by HZDR and TUBAF. The crystal grower group of TUBAF has been working on melt growth of semiconducting crystals for more than 10 years with the development of novel or improved growth processes and technologies being one of the major research topics. Special competencies and experiences exist in the field of vertical gradient freeze growth of Ge- and GaAs-crystals under the influence of external magnetic fields [7,8] and in the field of vertical Bridgman growth of multi-crystalline silicon from an inductively heated melt . Furthermore, there are also competencies on global thermal simulation and thermodynamic modelling of crystal growth, whereas the flow simulation including species transport can benefit from the close and successful collaboration between TUBAF and HZDR in this field (see, e.g., [7,8]).
In the crystal growth laboratory of TUBAF, a high-vacuum induction furnace for the directional solidification of multi-crystalline Si ingots with diameters up to 110 mm is available. It is equipped with a translation/rotation stage for growth by a modified vertical Bridgman technique under mechanical rotation, and with an external gas flow unit for controlling the composition of the atmosphere during growth. Electromagnetic stirring of the melt, which is induced by the internal induction coil itself, can be controlled by changing the growth setup. Sessile drop measurements can be performed in a 3-zone resistance furnace, which is also equipped with an external gas flow unit. For the thermal and thermodynamic simulation the commercial software tools CRYSMAS and FACTSAGE are available at TUBAF.
A decade ago, HZDR was already actively involved in the first Cz-growth of 300 mm single crystals which was a prerequisite for the transition from 200 to 300 mm wafer technology . During last years, several studies have been performed on the influence of various magnetic fields on Cz and VGF crystal growth processes [11-13]. Recently, major interest was focussed on the effect of a simple RMF [14,15]. For experimental modelling of Cz-related flow fields, a Cz model of almost realistic scale has been installed inside the home-made magnetic system MULTIMAG. It involves crucible and crystal rotation as well as temperature gradients and heat fields close to Cz reality .
3. Resource planning and Budget JustificationThe described works shall be realized by two PhD students, one at HZDR and one at TUBAF.
Links: Close relations exist to projects A4, C2 and YIG.
References 6th Report on Solar Generation: Solar Photovoltaic Electricity Empowering the World, EPIA Publications 2011, www.epia.org
 O. Breitenstein, J. Bauer, P.P. Altermatt, K. Ramspeck, 2010, Influence of Defects on Solar Cell Characteristics. Solid State Phenomena, Vols. 156-158, 1-10.
 J. Schmidt, 2004, Light-induced degradation in crystalline silicon solar cells. Solid State Phenomena, Vols. 95-96, 223-228.
 C. Reimann, M. Trempa, J. Friedrich, G. Müller, 2010, About the formation and avoidance of C and N related precipitates during directional solidification of multi crystalline silicon. J. Cryst. Growth, Vol. 312, 1510-1516.
 P. Rudolph, 2008, Travelling magnetic fields applied to bulk crystal growth from the melt: The step from basic research to industrial scale. J. Cryst. Growth, Vol. 310, 1298-1306.
 A. Muiznieks, A. Krauze, B. Nacke, 2007, Convective phenomena in large melts including magnetic fields. J. Cryst. Growth, Vol. 303, 211-220.
 O. Pätzold, I. Grants, U. Wunderwald, K. Jenkner, A. Cröll, G. Gerbeth, 2002, Vertical Gradient Freeze growth of GaAs with a rotating magnetic field. J. Cryst. Growth, Vol. 245, 237-246.
 R. Lantzsch, V. Galindo, O. Pätzold, M. Stelter, G. Gerbeth, 2008, Vertical Gradient Freeze growth with external magnetic fields. J. Cryst. Growth, Vol. 310, 1518-1522.
 L. Raabe, O. Pätzold, I. Kupka, J. Ehrig, S. Würzner, M. Stelter, 2011, The effect of graphite components and crucible coating on the behaviour of carbon and oxygen in multi-crystalline silicon, J. Cryst. Growth, Vol. 318, 234-238.
 V. Galindo, G. Gerbeth, W. von Ammon, E. Tomzig, J. Virbulis, 2002, Crystal growth melt flow control by means of magnetic fields. Energy Conversion and Management, Vol. 43, 309-316.
 R. Lantzsch, I. Grants, O. Pätzold, M. Stelter, G. Gerbeth, 2008, Vertical Gradient Freeze growth with external magnetic fields. J. Cryst. Growth, Vol. 310, 1518-1522.
 I. Grants, G. Gerbeth, 2008, Use of a traveling magnetic field in VGF growth: flow reversal and resulting dopant distribution. J. Crystal Growth, Vol. 310, 3699-3705.
 I. Grants, G. Gerbeth: Linear and nonlinear stability of a thermally stratified magnetically driven rotating flow in a cylinder. Phys. Rev. E, Vol. 82, 016314, 2010.
 I. Grants, G. Gerbeth, 2007, The suppression of temperature fluctuations by a rotating magnetic field in a high aspect ratio Czochralski configuration. J. Crystal Growth, Vol. 308, 290-296.
 I. Grants, G. Gerbeth, 2012, Transition between turbulent magnetically driven flow states in a Rayleigh-Benard cell. Phys. Fluids, Vol. 24, 024103.
 A. Cramer, M. Röder, J. Pal, G. Gerbeth, 2010, A physical model for electromagnetic control of local temperature gradients in a Czochralski system. Magnetohydrodynamics, Vol. 46, 353-361.