>> ***** >> THE AUTOBONDING RULES STATE THAT TWO ATOMS WITH NON-ZERO CHARGES THAT >> HAVE >> THE SAME SIGN CANNOT BOND, EVEN IF THEY ARE WITHIN NORMAL BONDING RANGE. >> ***** >> >> This seems to make sense to me, given my naive understanding of charges. >> >> Q: Does this rule make sense? >> > > Are these "formal charges" or "partial charges"? (I assume FORMAL.)
They are not partial charges ... which are floats. I call them formal/ionic charges ... because I do not know the difference between 'formal charges' and 'ionic charges' All I can tell you is that they are small integer charge values found in the files. > FORMAL charges depend explicitly on how the bonding is defined. So, > technically, if you know the formal charge on an atom, you MIGHT be > able to figure out the bonding. For example, H2SO4: > > HO-SO2-OH > > Is that central unit O=S=O or (1-)O-S(2+)-O(1-)? > > People disagree. Neither of these examples violates my rule against bonding +/+ or -/- In O=S=O nothing is charged ... no problem. In (1-)O-S(2+)-O(1-) we have negatively charged Oxygens bonding with a positively charged Sulfur ... no problem. > But the PARTIAL charge is defined without regard to bonding. > > So it only makes sense to be using FORMAL charge here in terms of > determining bonding connections. Recall that I really *dislike* being in the business of calculating connectivity. I have to do it because most file formats do not include connectivity information. > Another example: > > carbon monoxide. Is that (1-)C-(triple bond)-O(1+) or C=O, no charges? > > Your choice. It's just formal charge. negative Carbon bonds with postive Oxygen ... does not violate the rule ... not a problem. >> I fear that this is going to raise questions because someone is going to >> put two positively charged carbons next to each other in an .xyz file >> for >> .pdb file and expect them to bond. >> >> Q: Are there legitimate cases where two positively/negatively charged >> atoms can bond? > > FORMALLY, probably not. formal/ionic charge is all that I am worried about. > PARTIALLY, sure. That's ubiquitous. Here's a > quick calculation for CH3(+): > > NET ATOMIC CHARGES AND DIPOLE CONTRIBUTIONS > > ATOM NO. TYPE CHARGE No. of ELECS. s-Pop > 1 C 0.654881 3.3451 1.35158 > 2 H 0.115047 0.8850 0.88495 > 3 H 0.115029 0.8850 0.88497 > 4 H 0.115043 0.8850 0.88496 > > clearly all the bonded atoms are partially positive. FORMALLY, though, > we just assign the charge to C and call it (1+). A + charged C connected to 3 uncharged H is not a problem. A + bonding with an uncharged atom does not violate the rule. >> It would be possible for me to remove this restriction, >> but I fear that it may cause ionic bonding in areas >> where it does not make sense. > > I can think of an esoteric case or two where one might have lots of > negative charge spread around a molecule. (H2BN comes to mind.) But if > someone has such an esoteric structure, they can't espect perfection. OK > Still, it seems to me that the bond distance, not the formal charge, > would define at least the SINGLE bond in this case. Step 1. Determine if the two candidate atoms are in range Step 2. if the two candidate atoms have non-zero charges which are both positive or both negative, the bond is rejected. The rule that I am questioning is Step 2. > Maybe a compromise would be to not add second or third > bonds if the formal charges are crazy. The autobonding does not perform any multiple bonds ... only single bonds. > But, maybe I don't understand. If the formal charges are known, this > SHOULD absolutely define exactly the bonding..... I don't know. I am under the impression that some people like to open up a .pdb file, stick in a few charges, and have it work 'correctly' (whatever that means) > So, for example, in allyl anion, [CH2CHCH2](1-), you won't know where > to the draw the double bond until I tell you where the formal charge > is. And if I DO tell you where the formal charge is, you should be > able to figure out for yourself where the double bond is. But no > matter what I say, there should be a SINGLE bond between adjacent > carbons just based on bond distance. At this point I am not looking to enhance the autobonding algorithm. (There were times in the past when I was interested in doing so, but it now seem futile to me.) I am leaving this rule in place. Miguel ------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_idv37&alloc_id865&op=click _______________________________________________ Jmol-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/jmol-developers
