JASRI’s Technical Investigation Commission for the Burst of a Cryostat submitted an interim report. Its English version, which is not a rigorous translation of the Japanese version, is attached below. This burst was initially reported misleadingly as “explosion at SPring-8”; however, what really happened was the burst of a cryostat in an experimental hutch after the completion of experiments.

Akira Kira, Director General, JASRI/SPring-8

Report on the Burst of a Cryostat at BL12XU

Technical Investigation Commission
for the Burst of a Cryostat

 
1) The Accident
   At BL12XU, Taiwan BL, three scientists were conducting their experiment on ice at extreme conditions since June 21 2005. The aim was to study the electronic structure of "Ice" under a pressure of 2.6 GPa and a temperature of 4 K using an inelastic x-ray scattering (IXS) technique. The sample of "Ice" was contained in a diamond-anvil-cell (DAC), and the sample size was 0.4 mm in diameter and 0.15 mm in thickness (see Fig.1(a)).
   They completed the series of IXS measurements, began on June 21, and finished taking the calibration data and then started to heat the sample back to room temperature at 2:00 pm on July 2. The accident - the destruction of a Be sample-container and Be windows- happened at 2:25 pm on July 2, 2005

2) The Cause of the Accident
   The Commission investigated the damaged cryostat (see Fig.1(b)) on July 4, and they found that both the Be windows, which were fixed to the lower part of the cryostat, and the Be sample-container, which was placed in the cryostat, were totally broken into hundreds of fragments scattered around the IXS spectrometer, except for the parts of their upper flanges and bottoms (see Fig.2). This implied that the burst of the Be sample-container brought about the destruction of the Be windows.

   The Be sample-container is a closed box that is filled with He gas for heat exchange in order to cool the sample, contained in the DAC, down to 4K. the cryostat is equipped with a pair of inlet and outlet valves for the He gas.
   The cryostat is also equipped with a relief valve (Fig.3) in order to prevent abnormal increase of the gas pressure. The relief valve is designed to release the inside gas when its pressure becomes abnormal, while it works to maintain the inside vacuum when it is normal.

   The Commission investigated the relief valve of the troubled cryostat after the accident, and they found that a substantial amount of air was leaking through the relief valve under a normal pressure. Even after several open-close operations, the relief valve stayed in trouble, i.e. lost its function to maintain the inside vacuum.
Presumably the reason is an improper setting for the cracking pressure.

A possible scenario for the accident is as follow:
(1) A substantial amount of air is leaking into the Be sample-container through the troubled relief valve, since the inside of the container becomes negative pressure at low temperatures.

(2) The air in the container condenses and then solidifies during the IXS measurements at temperatures below the liquefaction temperature of air.

(3) The moisture in the air condenses and then solidifies to become ice, and the ice grows during the experiments to finally block the connection SUS tubes placed between the Be sample-container and the relief valve.

(4) On heating the sample after the experiment, the liquid or solid air evaporates while the connection tube is still blocked by the ice. As a consequence, the inside pressure increases unexpectedly in temperatures between ~100 K and ~273 K.
The liquid or solid air becomes gas at ~80K, while the ice stays in solid up to 273 K.

(5) Finally, the Be sample-container was broken when the pressure inside the Be sample-container reaches a critical point.

  In summary, the consensus of the Commission is that the burst of the Be sample-container, followed by the destruction of the Be windows, is caused by a failure in the relief valve.