2

I am not able to identify the molecule C[Mg+]. I came across it a number of times in the USPTO-50k dataset, in reactions such as the one shown below:

CON(C)C(=O)c1cn(-c2ccc(Cl)cc2)c(-c2ccc(Cl)cc2Cl)n1.C[Mg+]>>CC(=O)c1cn(-c2ccc(Cl)cc2)c(-c2ccc(Cl)cc2Cl)n1

I searched in PubChem and MolView and couldn't identify the compound. The closest thing I found was magnesium carbide, with SMILES [CH3-].[CH3-].[Mg+2]. Is it the same compound?

Any ideas what C[Mg+] is or where I can look next? Also RDKit accidentally generated a similar molecule once, C[MgH]. Does this one exist too?

Njw96
  • 41
  • 3

1 Answers1

5

One explanation for C[Mg+] is the one of of methylmagnesium as in the Grignard reagent of methylmagnesium chloride (C[Mg+].[Cl-]), or the softer methymagnesium bromide (C[Mg+].[Br-]). Obviously, metal organic compounds with bonds which sometimes differ from «pure covalent» as e.g. in pyridine (c1ccncc1), or «clearly ionic» (as in the example of sodium acetate, [O-]C(=O)C.[Na+]) can represent a limitation of this reduced representation.

Maurice' comment shows that SMILES are better known for the representation of chemical compounds, than for chemical reactions (indicated by >>). So I copy-pasted the string into ChemDraw's sample page as illustration:

enter image description here

Edit: For comparison, the reading by Marvin (after an additional click to disable the Thiele rings)

enter image description here

Buttonwood
  • 29,590
  • 2
  • 45
  • 108
  • Pardon my basic questions, I am new to chemistry. When you say "methylmagnesium as in the Grignard reagent of methylmagnesium chloride" you mean the compound in this case is methylmagnesium chloride? Can methylmagnesium exist on its own?

    In the smiles representation you gave, chloride appeared to be disconnected from methylmagnesium. Is that what you mean by "metal organic compounds [...] can represent a limitation of this reduced representation." ? As in SMILES are not good for representing this type of compound?

     – Njw96 May 10 '23 at 14:36
  • 1
    In methylmagesium written as C[Mg+], the plus sign assigns there is a positive charge. In inorganic chemistry, $\ce{Mg^{2+}}$ is the typical form; here, (formally) there already is one compensation. But you still need an other negative counterion (an anion) for charge compensation because methylmagnesium alone will not exist, and can not be stored in a bottle for the shelf. A chloride, or bromide are chemically simple, stable (commercially available) examples to resolve this, and are reagents known for this type of reaction. – Buttonwood May 10 '23 at 16:34
  • 1
    The bonding in organometallic compounds can be all within a spectrum of non-polar covalent bonds, polar covalent bonds (both partners participate, though one attracts the electrons stronger, than the other); or «ionic bonds» (similar to «now all binding electrons are handed over to one partner only», ions stick together because of charge attraction between cation and anion). To some degree, it depends on the human/the algorithm if the bond is perceived polar covalent, or ionic. E.g., ChemDraw converts "methylmagnesium chloride" still the covalent way, and hence assigns Cl[Mg]C as SMILES. – Buttonwood May 10 '23 at 16:45
  • Hence, consistency in the application of rules becomes important, even more so if multiple algorithms/programs harvest the data (see Daniel Lowe's thesis at Cambridge UK Extraction of chemical structures and reactions from the literature (https://doi.org/10.17863/CAM.16293) as an entry), or multiple databases are joined. This is (in part) the less appealing, less often advertised part of data curation before extraction of knowledge from the data really can start. SMILES can be good, InChI (standard, and the non-standard options) can be good as long as creators and users agree on rules. – Buttonwood May 10 '23 at 16:57
  • 1
    See e.g. Big Data from Pharmaceutical Patents: A Computational Analysis of Medicinal Chemists’ Bread and Butter in 2016JMedChem4385, or Annotated Chemical Patent Corpus: A Gold Standard for Text Mining in 2014PlosOne (the later open access) as applications potentially relevant for your work ahead, and http://chemdataextractor.org/ . – Buttonwood May 10 '23 at 17:03