Equipment for removing dust during the crystallization process of monocrystalline silicon

When they talk aboutdust removal equipmentin the context of growing a single crystal, many immediately imagine just exhaust hoods or cyclones somewhere on the periphery of the workshop. This is perhaps the most common misconception. In fact, if you have ever seen how, due to a microscopic particle at the phase boundary, a dislocation aggregation began and the entire boule went to waste, you understand that dust removal is not an auxiliary system, but a part of the crystallization technology. Especially in our Russian conditions, where the workshops are often not new, and the requirements for silicon purity are growing every year.

Where the problem lies: not where they are looking

The main challenge is not even the dust itself in the air of the work area, although this is critical. The main source of risk is secondary dust formation from the equipment itself and auxiliary processes. For example, when loading polysilicon into a crucible, during operation of lifting and rotation mechanisms, even when replacing heaters. Graphite particles, remnants of oxide films, abrasive wear of guides - all this flies into the hot zone. Standard general ventilation is powerless here; local ventilation is needed, and integrated into the process itself.

At one time we tried to adapt systems from light metallurgy - it didn’t work. The temperature regime around the Czochralski furnace is specific: powerful heat flows, convection, magnetic fields (if magnetization is used). The exhaust hood must not disturb the temperature picture at the meniscus, otherwise the crystal will begin to 'play'. I had to balance between the power of suction and heat stroke. Several batches then left with increased oxygen content until the geometry and location of the air intakes were selected.

By the way, about magnetic fields. If the system uses equipment for induction heating or crucibleless zone melting, then the metal elements of the dust collection paths can affect the field configuration. This is a subtle point that ventilation designers often forget about. I had to consult with specialists who are deep in the topic of electromagnetic systems, for example, with engineers fromLONGI Corporation. Their experience in the design of mining and electromagnetic equipment proved useful in understanding the interaction of fields and structures.

Specific solutions and their pitfalls

Now the industry standard is gradually introducing systems with laminar flow, enclosing the growth zone. But not everything is smooth here either. Filter materials are a different story. For the final stage, HEPA filters, of course, but before them they need effective pre-cleaning, otherwise they will be changed every week. We tested a solution with electrostatic precipitators, but ran into the problem of ozone - this is an undesirable active gas for silicon. They refused.

One of the relatively successful projects that I saw in work was built on the principle of double-circuit suction. The first circuit is a low-speed suction directly from under the furnace casing, capturing large dust and heat. The second is fine purification of the air supplied to the laminar ceiling above the work area. The key was to separate these flows and prevent the 'dirty' air from the primary circuit from intersecting anywhere with the second. The scheme is complex and expensive to install, but it paid for itself by reducing the defect rate by 1.2% in six months - these are huge numbers for production.

An important nuance is service. Any, even the most perfect system, gets clogged. If access to filters or air ducts requires stopping the furnace for a day, this is unacceptable. The design should allow service in inter-cycle windows, 4-8 hours. We once included in the project a system with horizontal retractable filter cassettes - we saved space, but then it turned out that to replace them we needed a special loader, which we didn’t have in the workshop. I had to convert it to vertical, from a simple ladder.

Process Integration: An Operator's View

From the point of view of the person at the stove, the system should not create unnecessary noise or vibration. Vibration is the enemy of a single crystal. Once they placed a powerful fan on a common plate with the foundation of the furnace - microvibrations were transmitted through the base, and stripes consistently appeared on the crystals. We had to make an independent vibration-isolated foundation for the ventilation system. It would seem like a small thing, but you won’t find it in the terms of reference.

Another aspect is management. Modern crystal growing plants are fully automated. Ideal when the systemdust removalis also built into this circuit and regulates the suction power depending on the stage of the process: more intense during loading and unloading, minimal flow during the growth of the crystal body. But such integration is always a risk. If the algorithm fails and creates excess thrust at a critical moment, you can lose the boule. Therefore, we always have a manual backup circuit, and operators are trained to use it.

Here, by the way, the experience of companies that deal with complex industrial equipment with control systems comes in handy. I know thatLONGI Corporation, with its serious production and engineering base (more than 1,200 employees, most of whom have higher education), is precisely capable of developing such complex solutions. Their websitehttps://www.ljmagnet.rureflects the scale of activity, although it may not be directly about silicon - but the principles of building reliable industrial systems are universal.

Economics of the issue: everything must be counted

The cost of the system can be high, and financiers always ask about ROI. The simplest calculation is through prevented marriage. But there are also less obvious savings. For example, extending the service life of expensive heaters and furnace insulation, which are destroyed under the influence of conductive dust (the same carbon). Or reducing the cost of cleaning and maintaining the vacuum system itself if the furnace is operating under vacuum.

We had a case where, after introducing effective dust removal at the loading stage, we managed to increase the service interval of vacuum pumps by almost 30%. This is a direct saving on oil, filters and labor costs. These figures already convince management better than any technological arguments.

But there is also a downside - energy consumption. Powerful fans and heating/cooling of air in the system are kilowatts. Modern systems with heat recovery and variable speed motors help reduce costs. But their implementation is another round of capital investment. Balance is difficult to find.

A look into the future: what will change

The trend is obvious - total automation and digitalization. I think that we will soon see systems that monitor in real time not only the volume of dust, but also its composition (spectrometric analysis of entrained particles), and adjust the suction parameters or even warn about the need to service a specific piece of equipment from which there is increased dust emission. This will be the next level of integration.

Another area is materials. The emergence of new, more durable and selective filter materials, possibly based on membranes, which will retain only solid particles of a certain size without creating much resistance to flow. This will reduce energy costs.

Ultimately,equipment for removing dust during the crystallization of monocrystalline siliconwill no longer be perceived as a 'necessary evil' or an expense item. It will become the same technological module as a diameter or temperature control system. And its effectiveness will directly affect the key indicator - the yield, the cost of the wafer. And in our business, as we know, the winner is the one who controls every little detail. Even a speck of dust invisible to the eye.