Potential issues with the experimental hydration free energies for mobley_4850657 (methanesulfonyl chloride) and mobley_9201263 (methyl methanesulfonate)?

[xiaowei-xie2]

Hi,

I would like to inquire about the experimental hydration free energies for two molecules listed in the FreeSolv database: mobley_4850657 (methanesulfonyl chloride) and mobley_9201263 (methyl methanesulfonate).

Both molecules are reported in the database to have an experimental hydration free energy of -4.87 kcal/mol, sourced from the same reference (10.1139/v98-084). I looked at the original paper, and it appears to me that the values should be -2.97 kcal/mol for mobley_4850657 and -5.27 kcal/mol for mobley_9201263, as indicated in Table 4.

Could you clarify how the experimental value of -4.87 kcal/mol was derived? Is there something I might have overlooked?

Thank you,
Xiaowei

[xiaowei-xie2]

Also for methanesulfonyl chloride, the experimental hydration free energy reported in the source paper appears to be an estimation rather than a direct measurement. The authors noted that direct measurement was not possible due to its rapid reaction with water. Instead, they estimated the value from group additivity based on the hydration free energy of another molecule, propanesulfonyl chloride.

Below is the quote from the original paper.

"We hoped to determine the solubility of methanesulfonyl
chloride, despite its rapid hydrolysis, by determining the
pseudo-zero order rate of reaction for a saturated solution (20),
but the process proved to be too rapid to follow titrimetrically.

Instead we determined the free energy of transfer of propane-
sulfonyl chloride and estimated that of methanesulfonyl chlo-
ride by group additivity. For propanesulfonyl chloride the zero
order rate of hydrolysis, at a pH of approximately 2.7, was
2.57 × 10–5 M s–1, while the rate constant for uncatalyzed hy-
drolysis is 3.93 × 10–4 s–1 (21). These values lead to a solubility
of 0.065 M. From vapor-pressure data in Beilstein we could
calculate (6) the heat of vaporization as 12.88 ± 0.09 kcal/mol.
The heat capacity of vaporization was estimated from that of
dimethyl sulfone by atomic additivity (12), as –30.67 cal deg–1
mol–1. The vapor pressure can then be estimated (6) as 0.67
Torr. From the solubility and vapor pressure we obtain ∆Gt for
propanesulfonyl chloride as –2.55 ± 0.23.
The free energy of transfer for methanesulfonyl chloride
could be estimated from that for propanesulfonyl chloride by
means of the group contribution for a CH2 group (22); in this
way we obtained ∆Gt = –2.97 ± 0.27."

[davidlmobley]

Thanks. I haven't dug into this yet, but just to help me out -- what's the historical origin of these, if you know? e.g. were they used in a blind challenge?

The short version of my answer is "I don't know" and "the original paper is probably right" but I will see if I can find anything else, when I get time.

And yes, Guthrie sometimes used estimation. Unfortunately he died several years ago so we cannot ask him about this.

[xiaowei-xie2]

I am not sure about the historical origin, but I don’t believe they were part of any SAMPL challenges. I came across them because methanesulfonyl chloride reacted with water during a simulation using an ML force field. Wikipedia indicated that this molecule should react with water, which made me curious about how its experimental hydration free energy was measured if it is not stable. Then I was puzzled as to why the value in the database differed from that reported in the original paper.

Please let me know if you find any information about this.

Thank you so much,
Xiaowei

[davidlmobley]

OK, I've done the first level of digging into this and concluded that, as far as the source data is concerned, I agree with you and it looks like these two values should probably be corrected.

I haven't yet determined exactly how we inherited these values (probably it's an error inherited from an earlier tabulation, since there's nothing in my changelog that says that I/my group specifically touched these values). For your purposes that's likely not important, but for myself I always like to know where an error came from.

In terms of your concerns on estimation/etc., on the one hand you can take the estimated value with a grain of salt. On the other, Peter was usually VERY careful about assigning confidence to things and propagating that into his uncertainty estimates. The fact that he ultimate gave this an (estimated) value of -2.97+/-0.27 means that he was rather confident in the estimate; I've seen him assign uncertainties as high as 1.93 kcal/mol (indeed, he used an uncertainty of 1.93 as a placeholder for "low confidence" and you'll see that in the database).

I'll get back to you if I manage to track down the historical origin of the error, but for now assume the paper values are correct and an update to the database will be forthcoming.

[davidlmobley]

OK, I actually just confirmed these values were in the earliest version of the database I put out back in 2007 (which I'd pulled from prior work, such as from Rizzo and others) meaning this is, as far as I know, a pair of longstanding typos/data entry errors.

[xiaowei-xie2]

Thank you so much for digging into this and providing your thoughts on the experimental uncertainty!