Article 1

Introduction

Like the majority of industrial silicon ingot manufacturers, Siemens Solar Industries (SSI) was dependent on the standard Czochralski (CZ) batch process. This method has been used commercially for decades, and is well defined. As with many single batch processes, there are several manufacturing limitations:

· Production output is limited by the initial charge capacity.
· Throughput is hindered by non-productive setup, heat-up, and cool down time.
· A relatively high percentage of the charge (7.5%) can not be grown from the bottom of the crucible, resulting in yield losses.
· Expensive quartz crucibles can not be reused, contributing to high manufacturing costs.
· Hotzone components deteriorate due to thermal cycling and excessive handling.
· Since Silicon charge materials must have sufficient mass to achieve reasonable meltdown times and prevent bridging, alternative materials such as granular silicon and fines can not be used in high volumes. This limits the potential for cost and energy savings.

In an effort to overcome these limitations, SSI, in cooperation with the Northwest Energy Efficiency Alliance (NEEA), and Jax Industries, developed a system for multiple batch recharging. This process enables crucibles to routinely be refilled seven to eight times before powering down the furnace. The realized benefits from the recharge project include:

· Lower manufacturing costs (21% annual savings).
· Improved yields (10%).
· Increased throughput (15%).
· Reduced energy "payback" of finished solar modules from 3.3 y to 2 y.
· Reduced material handling and processing by 83%.
· Eliminated 86% of scrap silicon and 87.5% of scrap quartz.

While the concept of multiple batch recharging is not new, most industrial manufacturers have not adopted the technology. Successful recharging requires high purity granular feedstock, long-lasting crucibles, and affordable recharge systems with high reliability. Working in conjunction with Siemens engineers, Jax Industries has continuously refined their recharge system into the simple, reliable and robust system in use today.
MEMC is credited with many of the recent advancements. In a joint venture with their crucible manufacturer, they developed a crucible coating which extends the potential life of the crucible to over 222 h. They also significantly improved the quality of the granular silicon being produced in their Pasadena, Texas facility. During the early stages of SSI's project, MEMC provided valuable insight into recharge techniques, and supplied on-sight technical assistance.
Some recharging systems are commercially available, but are expensive, complicated, and have not been proven in a high-volume manufacturing environment. In the effort to develop an alternative system using readily available materials to keep costs to a minimum, Jax Industries also designed the feeder system to be robust, simple to operate and maintain, and adaptable to the various furnace models.

Project History

In April of 1998, SSI entered into a jointly funded project with the NEEA to develop a more energy efficient approach to crystal growing. The established project objectives were very ambitious:

· Reduce power consumption by 40%.
· Reduce Argon usage by 50%.
· Increase productivity by 15%.
· Improve or maintain the quality of the grown ingots.
· Encourage the Pacific Northwest semiconductor manufacturers to adopt the energy efficient technology developed at Siemens.

In preparation for marketing the technology to Northwest semiconductor manufacturers, a market analysis was performed. During this analysis, engineers and managers from several local ingot growers were interviewed. It was surprising to learn that much of the interest in the project focused around recharge systems. Some of the participants were already working to develop recharge, and along with the lack of a reliable recharge equipment, they were experiencing difficulties with crucible lifetime issues. It was clear that developing a capable recharge process and equipment would dramatically increase marketing potential for the overall project. With that in mind, SSI shifted it efforts from topping off the initial charge to refining multiple batch recharge.

Results and discussion

Once the decision to resume multiple batch recharge development was made, progress moved rapidly. Having past experience and hardware already in place proved to be a big advantage. One major advantage over our previous work was the improved thermal characteristics of the energy efficient hotzone (EEH). The additional insulation and shielding lowered the overall temperature of the crucible walls and minimized fluctuations. These factors play an important role in crucible lifetime. Crucible lifetime can also be extended with the use of coatings. Effective coatings were not available until recently. Arrangements were made to procure coated crucibles, and a number of standard runs were made so that performance could be compared. Once verifying that the coated crucibles performed as well as our standard crucibles, continuous recharge experimentation began using the Jax Industries recharge system.
The first two experimental runs demonstrated conservative batch recharging, with each run lasting 3 days and achieving two successful recharge cycles. The third run surpassed all expectations. This run continued for 9 days, producing a dozen full term ingot sections. This experimental run produced more silicon product at higher yield and throughput than any run previously produced at SSI. Since then, numerous optimization experiments were conducted. While run cycles as long as 12 days were demonstrated, the most consistent results were obtained by performing seven recharge cycles and averaging 7 days at temperature.  During this testing phase, Jax Industries continually strived to refine the recharge equipment for lighter weight, high purity and greater reliability and safety.
The yield and throughput benefits of multiple batch recharge are indicated in the accompanying chart. Yield is improved because the residual melt (potscrap) becomes a smaller portion of the total charge with each successive recharge cycle. Product yield is determined by dividing the length of usable product grown at target diameter by the theoretical maximum length for a given charge weight and target diameter. Productivity is increased because non-productive set-up, heat-up, and cool-down times are eliminated. Productivity is calculated by dividing the total length (mm) grown by the total cycle time. Increased throughput greatly reduces the amount of energy and argon required for each run.
Improvements in yield and throughput lower manufacturing costs dramatically. Additional savings are realized by reducing the consumption of quartz crucibles, and by reducing the preparatory handling of the feedstock. Most low-cost silicon feedstock requires resistively testing and acid etching.

Granular silicon does not need these labor-intensive activities. The total savings achieved using seven recharging cycles represents annual financial savings of over 21%.
SSI is dedicated to providing customers with economically and environmentally benign solutions to their energy needs. NEEA's main focus is on permanently increasing market share of energy efficient products and services. This project fulfills the mission objectives for both organizations. In addition to financial savings, recharging drastically lowers the energy requirements for producing a silicon ingot. In fact, ingots manufactured using Jax Industries recharge system require less than half the energy used in standard CZ growth. The main reason for this is the efficiency of the fluidized bed reactor (FBR) process that is used to produce granular silicon. The energy content of semiconductor-grade silicon produced by the standard (Siemens) process is over 600 kWh(e)/kg, while the energy embodied in granular silicon produced by the FBR process is only 240kWh(e)/kg.

Conclusion

This project has resulted in radical reductions to CZ manufacturing costs for industrial silicon ingots. Multiple batch recharging has the potential to deliver a quantum leap in silicon ingot manufacturing efficiency. While the financial impact is astounding, the environmental impact is even more important. The combined annual energy savings of SSI's multiple batch recharging using Jax Industries systems and EEH have accomplished the equivalent of building a 370 kW average power plant. According to Jeff Harris, Senior Conservationist to the Northwest Power Planning Council, "…if they (Northwest Silicon Manufacturers) all adopted this technology they will have built (the equivalent of) a power plant of just under 15 average megawatts or enough power to serve a city of 13,000 people!" When the energy requirements of manufacturing the polysilicon starting materials are also considered, the environmental impact is even greater.

By Bryan Fickett and Greg Mihalik