search

IUPAC-InChI / InChI

Main InChI repository
MIT License
28 stars 2 forks source link

RecMet problem with chlorides #19

Open fbaensch-beilstein opened 1 month ago

fbaensch-beilstein commented 1 month ago

During my research, I came across a phenomenon in InChI. The -RecMet flag does not work for the following compounds (all chlorides):

magnesium chloride, MgCl2 InChI=1S/2ClH.Mg/h21H;/q;;+2/p-2 InChI=1/2ClH.Mg/h21H;/q;;+2/p-2 -RecMet

aluminium chloride, AlCl3 InChI=1S/Al.3ClH/h;31H/q+3;;;/p-3 InChI=1/Al.3ClH/h;31H/q+3;;;/p-3 -RecMet

beryllium chloride, BeCl2 InChI=1S/Be.2ClH/h;21H/q+2;;/p-2 InChI=1/Be.2ClH/h;21H/q+2;;/p-2 -RecMet

magnesium chloride, MgCl2 InChI=1S/2ClH.Mg/h21H;/q;;+2/p-2 InChI=1/2ClH.Mg/h21H;/q;;+2/p-2 -RecMet

As indicated, the second (non-standard) InChI has been created with the '-RecMet' flag, but the reconnection layer is always missing.

fbaensch-beilstein commented 1 month ago

According to @flange-ipb's findings, this seems to be a problem with the valences. I used technetium heptaflouride, no r-layer is specified here. It does for technetium in oxidation state 6 (TcF6). If a fluorine is replaced by a methyl in TcF7 (which leads to TcF6C), the reconnection layer is indicated again.

schatzsc commented 1 month ago

"Standard valences" are a major sin and we need to decide which bonds to connect and which ones not.

Did you bring this issue to the attention of Gerd Blanke?!?

fbaensch-beilstein commented 1 month ago

Gerd and I are working on the bonds and how we deal with them.