Industrial Ph.D.: Equilibrium and Kinetic Considerations in Refining of Silicon

One of the main findings in Jesse White’s Ph.D. project shows that previous kinetic models are far too simplistic and grossly underestimates the rate of boron removal in the refining of silicon.

On November 29 Jesse White defended and passed his thesis Equilibrium and Kinetic Considerations in Refining of Silicon

Jesse White has written a Ph.D. thesis in collaboration with Elkem. The Ph.D. thesis titled Equilibrium and Kinetic Considerations in Refining of Silicon is an important contribution to fundamental research in the thermodynamics and kinetics of slag refining. Jesse, who works as a research engineer for Elkem Technology chose to do his research at the Royal Institute of Technology (KTH) in Sweden. At KTH he also teaches the undergraduate course Materials Thermodynamics, for which he was awarded the prize for “Teacher of the Year” for 2013 in the Materials Science and Engineering department. Jesse defended and passed his thesis on November 29, 2013.

Can you describe why this type of fundamental research was necessary?

There are several areas where fundamental knowledge is lacking when it comes to refining silicon for photovoltaic applications with the Elkem Solar process, and this study attempted to address some of the key areas. The focus areas in this study were slag refining of silicon, and silicon melting in general.

Why did you choose to do the research at KTH in Sweden?

It was very advantageous to conduct this type of study at KTH in Sweden, since they have a long tradition of research in slag-metal reactions, mainly in iron and steel-making. As a result they have well-developed experimental techniques and I was saved a lot of time and trouble. I had the guidance of my advisor Prof. Sichen Du who is a great experimentalist with good fundamental knowledge.

What were the results, and what methods did you use?

For one, the thermodynamics of boron extraction from silicon using slag treatment was re-examined. Here we are talking about the determination of what factors affect boron removal, and to what degree we can remove boron from silicon. The motivation here was that published data in the past are in wide disagreement and show far inferior results from what we actually achieve in our process. In this study, a silicon-slag-gas equilibrium technique was employed. The evidence shows that nitrogen gas actually helps the removal of boron; this was totally unknown before. The composition of the slag is also a factor, but not at all to the degree that was shown in previous studies.

A second key area was the investigation of the kinetics of boron extraction, which is to understand how fast boron can be removed from silicon. This is critical knowledge if we are to go forward with improving the slag refining technology, i.e. improving reactor design and the overall efficiency of the slag refining process. A unique impeller mixing system was constructed for this study. The results were surprising: it was discovered that the transfer of calcium leads to low temporary interfacial tension and a very high interfacial area. This means that previous kinetic models were far too simplistic and grossly underestimated the rate of boron removal.

Finally, fundamental studies of silicon melting in contact with graphite generated a few surprises. A sessile drop system was constructed for this work. It was found that silicon can react with graphite well under the melting temperature of silicon and form silicon carbide via the gas phase, which affects how silicon wets the graphite. What is more, a carbon monoxide concentration at a certain threshold in the gas phase can directly react with silicon and form a carbide shell that is impossible to melt.

Why does it matter?

In the short term, this new insight into the thermodynamics and kinetics of slag refining may give the Elkem Solar process better flexibility in how the slag refining process is operated. The slag compositional range can be more flexible for example. If nitrogen plays a more active role in the chemistry of the process this is important to know! The kinetics of boron oxidation is far faster than anticipated, and mechanical mixing is definitely an effective means to improve reaction rates. Calcium transfer may have a major impact on the rate of boron removal, and it may be possible that we can even exploit this to further improve the kinetics. In the long term, it would be ideal to develop a continuous, counter-current process for slag refining, and this information can help lead to a viable reactor design for such an endeavor.

The findings from the silicon melting studies have wider implications. Silicon is melted in many different unit operations at Elkem. How to better control the furnace atmosphere to avoid melting problems such as accretion formation could lead to significant process improvements.

What are the next steps?

I feel that these results go to show that fundamental research and practical value are not mutually exclusive. Both short-term and longer-term practical gains can be made by such work. But, as it often goes with fundamental research, by answering some questions, many more new and interesting questions arise! During the course of this research study, it became increasingly clear that there are still many gaps in our knowledge base where very critical data are sparse, and there is plenty of room for future Ph.D. studies in silicon processing.

For example, thermodynamic data are sparse for alloys of calcium and silicon, which is an important system in silicon refining. The activity of Ca in liquid silicon should be measured more carefully, especially in the infinitely dilute range. Similarly, reliable data for the integral molar enthalpy of mixing for Ca in Si could improve process control of alloying.

Another example: Carbon and nitrogen are omnipresent in the melting and slag refining of silicon, and affect inclusion and accretion formation. A more accurate picture of how carbon and nitrogen behave in slags would be very beneficial and could improve many areas of the refining process and lead to a cleaner silicon product.

Finally, the development of kinetic models for silicon refining would be a big step forward. This experimental effort has revealed that mass transfer in silicon refining is far more complex than previously known. A promising approach for reactor design could be to develop a computer mass-transfer “micro-model” that employs finite element methods to incorporate complex fluid flow caused by forced convection and mass transfer under transient interfacial conditions. Nowadays we have the tools such as COMSOL and raw computing power that puts this in reach.

Changes in characteristics of slag-metal interface with increasing reaction time at 100 rpm mixing speed, slag phase doped with boron


Last updated: