http://www.guangweiquartz.com/Effects of melting parameters and quartz purity on silica glass crucible produced by arc method

We have investigated the effect of hydroxyl (OH) content in fused silica crucible on the scintillation and optical properties of the CsI single crystal, but not limited to, grown by Bridgman technique. For the purpose, 0.1 mol% Tl doped CsI single crystals were grown in crucibles made from fused silica of different grades with OH content varying from 20 ppm to 200 ppm. Silica glass of crucibles was characterized by FTIR and UV–VIS-NIR spectroscopy for the estimation of OH content. Grown crystals were tested for their scintillation performance and a correlation between OH content in silica glass and crystal quality is established. The possibility of ‘OH’ out-diffusion from silica crucible into the melt at higher temperature was further established by temperature dependent study of outgassing from silica crucible by residual gas analyzer (RGA). Further, an optimized process for silica crucible annealing to remove OH (<20 ppm) is proposed to achieve excellent crystal quality of a 5.6% energy resolution at 662 keV without any co-doping in Tl doped CsI.

In photovoltaic industry, silica crucible has an important influence on the quality of single crystal silicon. To obtain a silica glass crucible with large diameter, high uniformity, and low bubble content, two series of crucibles were prepared by the arc melting method, one with various melting parameters (initial power, melting power, and melting time) and crucible sizes, and the other with various high purity quartz crucible. The bubbles inside the crucible wall and pores on the inner surface were all measured using a polarised optical microscope and a portable microscope. The results show that all crucibles have a bubble aggregation area in their inner surface (0–0.4 mm), in which the density and size of bubbles are affected by melting time, melting power, and the distance between the crucible and the graphite electrode. The uniformity of the crucible decreases as the crucible diameter increases (16–28 inches), and the crucible is relatively stable when the initial power is below 400 kW. In final, a silica crucible with large size (diameter of 28 inches) and low bubble content on inner surface (∼50% reduction than that of traditional crucibles) was successfully prepared, which is of great value to the photovoltaic industry.

Currently, the primary materials for fabrication of solar cells are polycrystalline silicon and monocrystalline silicon, with a market share greater than 85% [1]. Solar cells with higher efficiency can be fabricated from monocrystalline silicon, which is usually obtained using the Czochralski (Cz) method [2, 3]. The silica crucibles used in the Cz method are typically made from high-purity amorphous silica. In general, these crucibles consist of two different layers: a transparent layer (Almost bubble free) and a bubble-containing layer (BC layer) [4]. In the outer BC layer, the material contains many bubbles, which decrease transparency. The composition of the gas inside the bubbles remains a matter of debate. It is most likely air, perhaps containing traces of carbon, or water vapour [5]. The inner transparent layer is almost completely transparent, and because this layer is in direct contact with the silicon melt, it is important to keep it free from bubbles throughout the Cz process to ensure that fewer bubbles are released into the melt and subsequently into the silicon ingot.

The silica crucible, which is in direct contact with liquid silicon, has an important impact on the quality of monocrystalline silicon, and silicon wafers with pinholes or dislocations cannot be used for solar-cell fabrication [6]. The industry has therefore devoted extensive efforts to preventing bubbles from entering the melt during the phase of crystal growth [4]. The high-purity quartz sand used to prepare glass crucibles contains various amounts of mineral inclusions (mica, feldspar, etc) and fluid inclusions [7, 8], which can form bubbles at high temperatures. In addition, other gases can affect the quality of monocrystalline silicon. Gas bubbles of SiO and CO can be produced in the melt–crucible interface [9, 10], forming small gas bubbles in the crystal or leading to the generation of dislocations inside the crystal [11]. Argon may also enter the silicon ingot as a protective gas [12]. Reducing the release of bubbles from the transparent layer into the melt during the Cz process can reduce the number of defects in the structure of monocrystalline silicon [13]. In fact, many experiments have been carried out to improve the properties of polycrystalline silicon by improving the purity of fused silica crucibles, but few studies have been reported in the field of single-crystal silicon [14–16]. Therefore, one of the purposes of this experiment is to reduce the bubble content in the transparent layer of crucibles by reducing the impurity element content of quartz.

Another research hotspot in the field of monocrystalline silicon involves increasing the size of the monocrystalline silicon rod, which first requires an increase in the size of the silica crucible [17, 18]. The increasing crystal-preparation time also results in stricter requirements for the quality and service life of the crucible. The preparation parameters must be adjusted to ensure the stability and uniformity of the larger-size crucible, which is urgently required in the industry and is another research purpose of this experiment.

To obtain high-quality monocrystalline silicon, many researchers have attempted to simulate the temperature distribution in the furnace and melt, to optimise the melting parameters of the Cz process [19–24]. The effects of the argon flow rate and silica crucible rotation speed on the concentration of silicon oxygen in single crystals have also been investigated [25–27]. To date, a great amount of research has focused on the use of crucibles (Cz process), while there have been only a few reports on the crucible preparation process, which is the main objective of this paper.

The main goal of this study is to obtain a large crucible with a uniform structure by adjusting the preparation parameters, and with a low bubble content in the transparent layer using high-purity quartz. At the same time, the influence of various melting parameters and impurity element contents on the formation of bubbles in the process of crucible preparation are summarized. To achieve these goals, we prepared two series of fused quartz crucible by graphite arc furnace, and applied them to prepare monocrystalline silicon through CZ method. Then, we examine the structure of silica crucibles prepared with varying initial power, melting power, melting times, and purities of raw materials. The bubbles inside the crucible wall and the pores on the inner surface of the crucibles were observed and measured using a polarised optical microscope and a portable microscope.

2. Materials and methods

2.1. Raw materials

As shown in table 1, the raw material used for the preparation of the silica crucible was quartz sand. The quartz ampoule bottle in this experiment was divided into two categories according to its purity: high purity (HP) and standard purity (SP), all produced by Covia (formerly Unimin, USA), and their product numbers are IOTA-6 and IOTA-CG, respectively. The chemical compositions of two kinds of quartz sand were measured using an inductively coupled plasma mass spectrometer (ICP-MS, Agilent 7500 Ce, Agilent, USA). The total content of impurity elements in HP quartz sand is 6.1 (μg·g−1), less than that of SP quartz sand. Other possible elements (P, Ni, Ba, Mg, Cr, Mn, and Cu) were also tested; they are not shown in table 1 because their contents were less than 0.01 μg·g−1. The block polysilicon used in Cz process, with bulk purity of 99.999999%, was purchased from Wacker Chemie AG, Germany.

The particle size distribution of quartz sand was evaluated using a laser particle size analyser (MASTERSIZER 2000, Malvern UK), as shown in figure 1. The refractive indices of the particles and dispersant are 1.544 and 1.000, respectively. SP quartz sand was graded into two categories: SP fine quartz sand and SP coarse quartz sand, which were used to prepare the transparent layer and BC layer of the silica crucible, respectively. HP quartz sand has only one granularity (HP fine quartz sand), which was used to prepare the transparent layer of the silica crucible. In terms of raw materials, the main variable of this experiment is the content of impurities. To eliminate the influence of the particle size of different grades of quartz sand on bubble formation, the particle size of quartz sand was strictly controlled using a mesh screen, so that the particle size distributions of HP fine quartz sand and SP fine high permeability quartz square cylinder. The D50 (50% passing size) of HP fine quartz sand, SP fine quartz sand, and SP coarse quartz sand are approximately 169, 155, and 242 μm, respectively.

The silica crucible manufacturing process is shown in figure 2(a). First, a coarse fused quartz crucible sand was poured into a rotating steel mould. The quartz sand adheres tightly to the mould walls under centrifugal force, which ensures an even distribution and thickness of quartz sand. The external diameter of the crucible is 16–28 inches, and the thickness of the crucible wall is 8–13 mm. A second layer of finer quartz sand was introduced on top of the coarse quartz. Air was then extracted from the hole in the steel mould for 2.0 min to keep the vacuum at 0.04 MPa, during which the power of the graphite electrode is called the initial power (350–450 kW). The crucible is heated for 8.0–23.0 min (melting time) through the arc between the graphite electrode and the steel mould to ensure that the quartz sand melts into fused quartz, during which the power of the graphite electrode is called the melting power (450–1200 kW). The layer of finer quartz fuses into a transparent layer with fewer bubbles, and the coarse quartz layer fuses a second time, forming an outside layer (BC) with a higher number of bubbles. Finally, a thin layer of unmelted sand was left between the finished crucible and the steel mould to simplify the removal of the crucible at the end.

The main process involved in the Cz method is shown in figure 3(a). The silica crucible was placed into a steel mould, and argon was used as a protective gas. The silicon in the silica crucible is heated to approximately 1500 °C to ensure melting. After the melt stabilisation, a crystal seed was dipped into the melt and then pulled slowly upwards while being rotated in the direction opposite to the crucible. Finally, the liquid silicon adheres to the crystal seed and becomes a monocrystalline silicon rod. The duration of the Cz method is approximately 110 h. As shown in figure 3(b), samples from the top, corner, and bottom of the crucible were selected for testing in this experiment; these samples were cut to 40 mm in length and 10 mm in width, with a thickness identical to that of the crucible wall.

This study performed the following two series of experiments. As shown in table 2, in the first series, four crucibles with different sizes were prepared according to different preparation parameters, formed using SP fused silica. These four crucibles were tested before the Cz process. The samples were designated S16, S18, S20, and S28, sequentially indicating the external diameter of the crucibles, that is, 16, 18, 20, and 28 inches, respectively. In a second series of experiments, two additional types of crucibles, which had undergone the Cz method, were examined: (i) two reference crucibles (R1 and R2) prepared using SP quartz sand, and (ii) two composite crucibles (H1 and H2), in which the BC layers were prepared using SP quartz and the transparent layers were prepared using HP quartz sand.