I read your referenced article, but all it said was: There are significant temperature effects on these properties, and cavitation itself will be dramatically affected with increasing temperature. There are no context or data associated with the statement, and therefore, from my perspective, there is no proof for the statement. I do not consider the article a credible reference.
As far as the smallest particle, it’s a complex question. The basic surface roughness of the record is reported to be between 0.01 and 0.005 microns. The smallest amplitude the stylus is reported to be able to reproduce is 1-mciron. So theoretical cleanliness would lie someplace in-between such as 0.1 micron. However, this is not achievable in a residential environment - you would need a clean room. So, practicality dictates something more reasonable. However, you have no method to verify the cleanliness without use of laboratory methods. Consequently, the cleaning ability of 40-kHz UT tank should be sufficient; preferably filtered to remove small particulate. Otherwise, you end up cleaning a record in dirty water - even if it looks clear. But absent a powerful UT tank, low powered UT tanks can benefit from a very small amount of nonionic surfactant to just reduce the surface tension to improve cleaning efficiency. But there can be a big difference between record general cleanliness and record best achievable cleanliness.
Take care,
Neil

