3rd Intenational Workshop on Crystalline Silicon Solar Cells SINTEF/NTNU, Trondheim NORWAY 3-5 June 2009 DOPANT SPECIFICATION OF COMPENSATED SILICON FOR SOLAR CELLS OF EQUAL EFFICIENCY AND YIELD AS STANDARD SOLAR CELLS Erik ENEBAKK, Anne K. SØILAND, John T. HÅKEDAL, Ragnar TRONSTAD Elkem Solar AS, 4623 Kristiansand, NORWAY * E-mail: erik.enebakk@elkem.no ABSTRACT Pilot production of solar grade silicon has taken place at Elkem Solar in Kristiansand, Norway, the last year. Ingot, wafer, and solar cells have been made at institutes and in industrial scale with good results. Based on this experience and calculations it is concluded that compensated SoG-Si with less than 0.45 ppmw B can be used for solar cells with efficiencies and yield equal to standard solar cells based on poly silicon. DOPANT LEVELS IN SOLAR GRADE SILICON One of the focus areas in the development of ESS from metallurgical route has been to produce a high quality silicon feedstock with sufficiently low boron and phosphorus content for high quality solar cells. When the boron content in the product is too high, the lifetime in the wafer and the effciency of the solar cells will decrease. A high boron content can also lead to increased light induced degradation related to the B-O complex formation, and reduced efficiency. Therefore, to avoid these efficiency losses the boron concentration is fundamental and the most important criteria for a SoG-Si feedstock. Despite the large number of potential new producers of SoG-Si from metallurgical route, there exist a limited amount of publications and no well established standard for the requirements for dopants and impurities in compensated silicon for solar cells. Here, general dopant requirements for compensated SoG-Si are described, and general specification limits for dopants in compensated SoG-Si is suggested. SoG-Si from metallurgical route will always contain a remaining amount of phosphorus in addition to boron, i.e. the feedstock is compensated. Phosphorus will, as boron, also contribute to the ingot resistivity but more importnant is it’s influence on the ingot yield due to the type-conversion. Because of different segregation behaviour of the dopants, a type conversion will be present in the ingot, which can reduce the overall ingot yield if there are too much phosphorus in the feedstock. The boron specification for standard p-type noncompensated multi crystalline silicon solar cells are typically 0.05-0.15 ppmw B, corresponding to ingot and wafer resistivity in range 0.5 – 2.0 ohm cm to satisfy acceptable efficiency targets. One of the challenges with SoG-Si from metallurgical route are to control the balance between boron and phosphorus and the distribution of the dopants in the following ingot growth. Not only is it challenging by metallurgical refining to obtain sufficiently low boron and phosphorus levels, but there is also a need for precice and accurate analytical methods. Keywords: Silicon solar cells, feedstock, SoG-Si INTRODUCTION Elkem AS has a long history as one of the leading silicon producers in the world. Together with Sintef/NTNU in Trondheim, Norway, a high technological level on process metallurgy and silicon product knowledge have been developed, and this knowledge base has been a needed platform for development of a metallurgical route for solar grade silicon. Since the first project in the late 70-ties until the start-up of Elkem Solar in 2001, several different technologies have been tested. The present process consists of 3 purification steps; slag treatment, acid leaching and directional solidification. The product is cut to size in a post treatment plant and surface impurities are removed by etching and washing. The product has been tested and qualified at institutes and in the industry. Industrial tests in which more than 150.000 wafers were produced showed efficiencies of Elkem Solar Silicon® based cells equal to poly-Si based references. Ingot yield was according to the p/n transition at the top part of the ingot, due to the typeconversion resulting from the B and P concentration[1]. The philosophy of Elkem Solar is to develop a product which has equal properties in ingot and solar cells as poly based ingots and solar cells. A property which is focused is the resistivity in ingots made from ESS. All produced multi crystalline ingots should be within the standard range of 0.5 to 2.0 ohm cm. Therefore, the needed dopant concentration in feedstock is determined by requirements relating to solar cell performance, ingot resistivity and ingot yield. 1 Extended abstract When SoG-Si from metallurgical route is used in ingot production, there will exist different opinions and ways of using the feedstock depending on the specifications at ingot- and solar cell producers. The feedstock may be mixed with poly silicon, various off-cuts, scrap or other dopants to adjust resistivity and yield, or it can be processed directly to solar cell ingots without blending. A reliable control scheme of dopant levels and their limits in the feedstock are needed in any case to support a stable ingot and solar cell production. Therefore, this paper is focusing on dopant control in compensated SoG-Si and how this can influence on ingot yield and resistivity. As a simplification the mobility can be assumed constant across ingot heigth, or the dependence of NA and ND can be modelled by simpler or experimentally determined relations. The resistivity calculations in this paper are based on carrier mobilities calculated by Equation (1.4). Table 1: Parameters for mobility calculation B (p-type) P (n-type) 44.9 68.5 µmin 470.5 1414 µmax n0 2.23x1017 9.20x1016 0.719 0.711 α In addition to the base resistivity in solar cell ingots, the ingot yield is important to focus on. Ingot yield can be defined in two ways: (1) the position of type changeover, or (2) the cut-off position where the base resistivity exceed a given value depending on the criteria from the different solar cell producers. The latter definition relates directly to the amount of usable material for further solar cell processing from the ingot, and a typical cut-off value would be 2 ohm cm. The ingot yield in terms type-changeover is calculated by Equation (1.5). Resistivity across the ingot height is calculated by simple formulas and assumptions. The general expression for resistvity in a silicon semiconductor is ρ= 1 (1.1) p ⋅ µ ⋅e where p is the concentration (cm-3) of majority carriers, µ is the corresponding carrier mobility (cm2/Vs), and e is the charge (C) of an electron. In compensated p-type Si for solar cells, the carrier concentration can be assumed equal to the net dopant concentration as given in Equation (1.2), p = NA − ND 1 N A0 ⋅ k A Type changeover = 1 − 0 N D ⋅ kD k D −k A (1.5) (1.2) Examples of a resistivity calculation are shown in Figure 1. Here, the feedstock melt contains 0.45 ppmw B and 1.00 ppmw P, and the resulting resistivity is above 0.5 ohm cm and the ingot yield is close to 90% of the total ingot. In the lower example the same feedstock has been mixed with 50% poly and some added boron. It results in a solar cell ingot with a higher resistivity and approximately the same yield. where NA and ND is the concentrations (atoms/cm3) of acceptors and donors, respectively. The Scheil equation is used to find NA and ND across the ingot height N i = N i0 ⋅ k i ⋅ (1 − g ) ki −1 (1.3) where N0 is initial melt concentration (atoms/cm3), i is a dopant element, e.g. B or P, and ki is the segregation coefficient for element i in silicon, and g represents the ingot height given by the fraction solidified. Segregation coefficients used in calculations in this paper are kB =0.80 and kP =0.35. The mobility values that goes into Equation (1.1) for both for p-type and ntype silicon is calculated by 3.0 0.45 ppmw B Resistivity (ohm cm) 2.5 1.00 ppmw P 2.0 1.5 1.0 0.5 µ max − µ min N +N 1 + A D n 0 α 0.0 (1.4) 3.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.241 ppmw B 2.5 Resistivity (ohm cm) µ = µ min + where the respective parameters are given in Table 1. Equation (1.4) accounts to some degree for reduced mobility in compensated silicon, which has been reported by others[2]. There are also reported that mobility in mc-Si ingots are lower due to grain boundaries than values obtained by Equation (1.4)[3]. However, the dependence between dopants and mobility in compensated and multicrysalline silicon is not clear. 0.500 ppmw P 2.0 1.5 1.0 0.5 0.0 0.0 0.1 0.2 0.3 Fraction solidified Figure 1: Examples on resistivity profile and typeconversion calculated from feedstock values. 2 Extended abstract Resistivity calculations can be used as a tool to visualize and model relations between chemical composition of the SoG-Si-feedstock and the expected resistivity and yield in ingot. Figure 2 shows how ingot resistivity and ingot yield can be found directly from the initial B and P concentration in the melt feedstock. For a SoG-Si feedstock containing 0.45 ppmw B and 1.00 ppmw P, the calculated resistivity and ingot yield can be found directly by interpolation between pairs of iso-resistivity and iso-yield lines. In this example, the corresponding ingot resistivity close to the ingot bottom is slightly below 0.6 ohm cm, and the type-changeover position is at >90% of ingot height. The diagram can also be used the other way around in order to determine the actual initial melt concentration from observed ingot resistivity and changeover position. 1.6 Red: Iso-resistivity lines (@5% above ingot bottom) Blue: Iso-yield lines (yield definition: pn changeover) 75 % % 80 85 % 90 4.0 3.5 ppmw P in initial melt 3.0 oh m 0. 5 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 ppmw P in initial melt cm SoG-Si Can be used directly >90 e -typ %p UMG-Si High boron Low resistivity Blending or refining needed 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 CONCLUSION Based on calculations and supported by experience, it can be concluded that typical maximum limits for dopants in SoG-Si are <0.45 ppmw B and <1 ppmw P. Even though the requirements for B and P are clear, attention will still be needed towards precise analytical methods and control mechanisms for product quality and stability to ensure an efficienct ingot and solar cell production. % cm cm oh m 1. 0 0.25 m oh Figure 3: The hatched area illustrates the required B and P range for SoG-Si feedstock which will result in an ingot with resistivity above 0.5 ohm cm and ingot yield above 90%. 0.4 0.2 1.5 5 ppmw B in initial melt 0.6 0.15 2.0 0.1 0.8 0.1 . >0 0.0 1.0 0.0 High phosphorus and boron Low resistivity and ingot yield Blending or refining needed 0.5 95 % 0.2 UMG-Si High phosphorus Low ingot yield Blending or refining needed 2.5 1.0 1.4 1.2 UMG-Si 0.7 ppmw B in initial melt Figure 2: Example of a dopant – resistivity digram. Resistivity lines (red lines) indicates ingot resistivity at 5% of ingot heigth, i.e. at bottom of ingot. The ingot yield (blue lines) is here defined as type-changeover position. REFERENCES [1] V. Hoffmann, 23rd European Photovoltaic Solar Energy Conference, Valencia, 2008 [2] K. Peter, 23rd European Photovoltaic Solar Energy Conference, Valencia, 2008 To establish a general dopant specification for SoG-Si there are two targets that must be fulfilled. First, the base resistvity in the finished solar cell ingot should be in the range 0.5-2 ohm cm. This represents the limits where industrial and high efficiency cells often are made. Some producers may accept higher values, but few would accept lower. Secondly, the ingot yield – or amount of usable ingot – should be above 90%. The definition of yield can vary among producers and the impact on ingot yield for different technologies can also vary. The area in Figure 3 defines where these resistvity and yield targets are fulfilled at the same time, thereby defining the absolute dopant requirements for compensated SoG-Si feedstock. [3] A. Cuevas, Conference on Optoelectronics and Microelectronic Materials and Devices (COMMAD ’08), Sydney, July 2008 Utilizing boron and phosphorus compensated SoG-Si material require adequate control of both level and ratio of the two dopants, as can be concluded from Figure 2 and 3. When the requirements for resistivity and yield are set, the B and P concentration must be within certain limits. Higher concentrations of the two dopants lead to narrower uncertainty limits for the feedstock concentration. Precise and accurate analytical methods are needed to meet theese requirements for administration of dopants in order to fulfill the expectations for resistivity and yield. 3