The accident at OCIM’s polysilicon plant shouts for more safety

The severe explosion at the polysilicon plant of OCI Malaysia (OCIM) in August 2024, which cost two workers their lives, has revealed serious safety gaps. What an expert recommends for process safety.

By Alan Crawford, Methylchlorosilanes and Trichlorosilane Process Consultant

There is no known organization that promotes routine sharing of process safety information across the global solar-grade polysilicon industry. This is a major shortcoming of this industry. Shared experiences (good and bad) have the real chance to improve the overall process safety of the entire industry. A major accident experienced by one producer, such as the incidents at OCI Malaysia (OCIM) in August 2024 or at Xinjiang GCL in July 2020, can give the entire industry a negative reputation.

OCIM experienced a major process safety upset on one of their hydrochlorination (HC) units for the synthesis of trichlorosilane (HSiCl₃ or briefly TCS) at the Bintulu site in Sarawak, Malaysia on August 14, 2024. Ten people were injured and two of these died in the weeks following the accident. OCIM has not released any information about the accident as of early January 2025.

The accident resulted in what is clearly a de-inventory/de-pressure of a HC synthesis reactor and associated process piping given the massive cloud of hydrogen chloride (HCl), steam and fine solids that was generated. The accident appears to have occurred with one of the four original Tokuyama Phase-2 HC units and specifically in the bottom area of one HC reactor. The site was originally built by Tokuyama, which had significant problems with commissioning and startup and sold the site to OCI (South Korea) in 2017.

Pictures and a video of the accident, combined with Google Earth and published site images, were analyzed to determine the general location within the overall OCIM site of where the accident occurred. From this general location, a list of potential failure modes was generated and analyzed. Unconfirmed rumors about the cause of the accident were ignored.

An assumption is made that the main gas feed pipe was constructed of Incoloy 800H (UNS N08810). Use of Incoloy 800H is well known by all industrial HC operators and is recognized as the only proven material of construction for use in specific areas of the HC unit. Some different materials are known to have been recently used by some producers, but there is no long-term safe history of these materials. If OCIM were to have used something other than Incoloy 800H in the area of the fatal accident, this would then become a strong root cause suspect.

A summary of potential failure modes of the OCIM HC reactor is provided in the table above. The five identified failure modes represent possible causes of the OCIM accident in August 2024.

Failure hypotheses

1. OCIM could have failed to implement and execute routine mechanical integrity checks of process lines including visual internal inspections of the main gas feed pipe. Routine internal corrosion was not recognized.
2. A change could have been made to the particle size distribution of ground metallurgical-grade silicon (MGS), which resulted in an elevated amount of MGS ‘sifting’ through the grid holes at the bottom of the HC reactor and into the area under the grid and into the main gas feed pipe. Localized erosion of the main gas feed pipe was not recognized.
3. OCIM could have poor return-to-service procedures with regards to ensuring that process piping, such as the main gas feed pipe, is properly cleaned of internal solids and properly dried before returning to service. Failure to dry and clean can result in accelerated HCl-related corrosion.

Recommendations

Regardless of the validity of the information received about the OCIM fatal accident and the analysis presented here, the following recommendations are made. These recommendations must be carefully evaluated by subject matter experts (such as those with Incoloy 800H piping design experience) prior to implementation.

Any industrial operator of a hydrochlorination crude TCS synthesis unit, whether for TCS-based or monosilane-based polysilicon production, or for merchant silane gas production, is strongly advised to heed this recommendations as soon as possible. This statement applies to both long-term legacy HC operators, newer entrants (China, Malaysia, South Korea) and Indian projects that are currently in the design/engineering phase.

1. Any current operator of an industrial HC TCS synthesis unit, regardless of experience or plant status (from early design phase through multi-year industrial operation), is recommended to consider conducting an independent review of all aspects of process safety with regards to HC unit process design, operations and maintenance strategy.
2. Examine the following questions: Has the specific site adopted the USA OSHA 1910-119 standards with regards to plant design, operation and maintenance? If not, why? Is the site using a similar system instead?
3. Review the site mechanical integrity inspection program (routine external pipe thickness measurements, for example, and non-routine visual internal inspections) and determine if the current program would identify localized erosion/corrosion of the main HC FBR gas feed pipe prior to catastrophic failure.
4. Review the site standard operating procedures including “return to service.” Do these procedures address visual inspections of process equipment and piping to ensure that hygroscopic solids (easily “taking up water”) have been removed prior to restart?
5. Check the following item: Does the site have an adequate management of change (MOC) program that would identify a process change, such as new particle size distribution (PSD) of ground metallurgical grade silicon (MGS), as placing a risk to safe operations?
6. Review current quality control plan for purchased ground MGS or MGS that is ground on-site to ensure that the current ground MGS powder PSD specification is always achieved.
7. Any industrial HC operator that has not implemented a formal mechanical integrity checking program nor has routinely used a mechanical integrity program (both routine external measurements and visual internal inspections during all planned maintenance outages) is recommended to carefully inspect all main gas feed piping between final superheaters and the inlet to the HC FBR as soon as possible. If the history of the piping is not known in terms of original fabrication and designs to accommodate thermal stresses, the best action would be to replace all main gas feed piping.
8. A review of known HC FBR Piping & Instrumentation Diagrams (P&IDs) shows that none of these designs include emergency isolation valves on the main gas feed line and other gas feed lines to the FBR vessel. Why are these valves not included? Known industrial reactors for methylchlorosilane synthesis and direct chlorination TCS synthesis have such valves. These valves are designed to prevent exactly what happened at OCIM should a feed line fail – catastrophic de-inventory of the HC fluid bed reactor (FBR).
9. New corrosion studies are recommended to be conducted for applicable materials of construction such as Incoloy 800H for the following systems:
   a. System 1: Hydrogen (H₂) – Silicon Tetrachloride (SiCl₄ or STC)
   b. System 2: H₂ – STC – HCl
   c. System 3: H₂ – STC – HCl – Metallurgical-grade silicon
   Results of these studies must be published in the public domain, preferably in a peer-reviewed journal. The technical procedures followed by Union Carbide, published in the only public domain corrosion study that includes Incoloy 800H (UNS N08810) for the specific system H₂ – STC – Metallurgical-grade silicon, should be followed for new corrosion studies.

This article is the summary of a 90-page in-depth analysis of the accident at OCI Malaysia by Alan Crawford. Download his full report (5.4 MB).