Pd(0)

I notice that Marvin doesn't recognize Pd(0) as a legitimate oxidation state. It should.





Along the same lines, if I want to draw (Ph₃P)₄Pd, I need to put + charges on each of the P atoms and 4- on Pd. This is not the custom among organometallic chemists. Either you should develop a new bond type, the dative bond, which does not change the formal charges of the atoms that it joins, or (better) Marvin should recognize that when there is an X-M bond, and M is a metal, and X has one more bond than its normal valence, then it should be understood to be the same as X⁺-M^-.

Bob,

bobgr wrote:

I notice that Marvin doesn't recognize Pd(0) as a legitimate oxidation state. It should.

If you explicitly set valence to 0 then there is no valence error. (Marvin 4.0.5.) You can do that using the "v+" button in the More window.

bobgr wrote:

Along the same lines, if I want to draw (Ph3P)4Pd, I need to put + charges on each of the P atoms and 4- on Pd. This is not the custom among organometallic chemists. Either you should develop a new bond type, the dative bond, which does not change the formal charges of the atoms that it joins, or (better) Marvin should recognize that when there is an X-M bond, and M is a metal, and X has one more bond than its normal valence, then it should be understood to be the same as X+-M-.

Dative bond is among the plans... Let me discuss this with the guys.

I tried Marvin 4.0.5 (on your Web site). When I press More, there is no v+ button visible.





In Marvin 4.0.3 (ACE Web site), there are v+ and v- buttons visible in the More window, and I can activate them and use them to make a little indicator appear next to the Pd, but the actions have no effect on the red underlining of the Pd atom.





Anyway, the Pd(0) oxidation site should be permitted naturally, just like the Fe(II) and Fe(III) oxidation states are.

You would not like to get a valence error when a metal atom (Pd, Ni, Na, etc.) is placed on the canvas without any substituents. So 0 oxidation state should be an accepted for metals. Do I understand you well?

Quote:

I tried Marvin 4.0.5 (on your Web site). When I press More, there is no v+ button visible.

I have checked Marvin: "v+" and "v-" buttons are visible also in Mac: http://www.chemaxon.com/marvin/doc/dev/example-sketch1.1.html


Probably, you have tried an applet example where query atoms were restricted: http://www.chemaxon.com/marvin/doc/dev/example-sketch2.html


By using the "queryAtoms" parameter you can allow or deny the displaying of query atoms. (In default settings, everything is visible.)

Gyuri wrote:

You would not like to get a valence error when a metal atom (Pd, Ni, Na, etc.) is placed on the canvas without any substituents. So 0 oxidation state should be an accepted for metals. Do I understand you well?

Yes, except for the "any substituents" part. I would like to be able to draw (Ph₃P)₄Pd without a valence error. (I would also like to be able to draw it without a 4- charge on Pd and + charges on the P atoms, but that's the subject of a different forum.)





It might be wise to eschew valence errors for transition metal atoms completely. For example, you allow Ni(II) and Ni(III) as acceptable oxidation states, but Ni(I) is known to be involved in certain catalytic cycles. It is possible to remove the valence error for Ni(I) by setting the monovalent radical flag, but it is simply not the custom for organometallic chemists to indicate unshared electrons of transition metals.





As for main group metals, I think it would be safe to allow the 0 oxidation state, but disallow certain other oxidation states, such as Mg(I). If an author wants to write Mg(I), he or she can use the monovalent radical flag.





I used http://www.chemaxon.com/demosite/marvin/index.html to look for the v+ and v- buttons. (BTW, you should set implicit H atoms to heteroterm at this site.)

bobgr wrote:

As for main group metals, I think it would be safe to allow the 0 oxidation state, ...

If you set valence to 0 explicitly, then valence error is not shown in the latest version. I that OK?


You can try it here: http://www.chemaxon.com/marvin/doc/dev/example-sketch1.1.html





We can set the default valence to 0 for metals, but some people prefer hydrides as default for some metals (LiH, NaH, KH, ...) and others don't, so we are not sure how to proceed.

Ferenc wrote:

If you set valence to 0 explicitly, then valence error is not shown in the latest version. Is that OK? You can try it here: http://www.chemaxon.com/marvin/doc/dev/example-sketch1.1.html

It does not work well. See the drawing, which has a Pd(2-) in the 0 oxidation state.





Besides that, going to the More menu is awkward and counterintuitive. At least put a Valence option in the pop-up menu when you right-click an atom.





More to the point, why do you check valences of metals, anyway? So many different oxidation states are possible. Even though some are less common than others, it doesn't mean that they're wrong. I would wager that the vast majority of valence errors in nonmetals are in fact errors, whereas the vast majority of valence errors in metals are not.





I know you can turn valence checking on or off in the Preferences, but maybe you need a new Marvin parameter, checkvalence, with possible values "all", "nonmetals", or "off".

Ferenc wrote:

We can set the default valence to 0 for metals, but some people prefer hydrides as default for some metals (LiH, NaH, KH, ...) and others don't, so we are not sure how to proceed.

If I draw a single bond and nothing else, everyone understands that the termini of the bond are C atoms and that there are three H atoms attached to each C. By contrast, if I write Na, no one would understand that an H atom was implied. If I wanted to indicate NaH, I would draw in the H explicitly. So, I don't think you should apply implicit H atoms to metals at all.

Bob,





Dative bonds will solve the Pd problem. (2 single covalent bonds are in contradiction with the v=0 setting, so the alarm is correct)





For a compatibility with MDL molfiles we may have to implement that alkali metals without explicit bonds or valence setting will get an implicit H. For example this is how Isis/Draw works.

Quote:

put a Valence option in the pop-up menu when you right-click an atom.

Agreed.





Can you point me to a list of valences for metals, which you consider correct? If there is no such list we will consider removing valence checking for metals. (Do you also include alkali metals and Al for example?)

Ferenc wrote:

(2 single covalent bonds are in contradiction with the v=0 setting, so the alarm is correct)

Not true. With the 2- charge on Pd, the alarm is not correct.

Ferenc wrote:

For a compatibility with MDL molfiles we may have to implement that alkali metals without explicit bonds or valence setting will get an implicit H. For example this is how Isis/Draw works.

Not ideal, but OK. I always have "heteroterm" turned on, so the H will appear, alerting me that I need to turn on the radical flag.

Ferenc wrote:

Can you point me to a list of valences for metals, which you consider correct? If there is no such list we will consider removing valence checking for metals. (Do you also include alkali metals and Al for example?)

The only incorrect valences for metals exceed the number of valence electrons of the element (e.g., +5 for Ti). Otherwise, some are more common than others, but none are incorrect.

Thanks, we will process this and get back.

Quote:

The only incorrect valences for metals exceed the number of valence electrons of the element (e.g., +5 for Ti).

I am not sure. For example, according to the literature, Ti can have 2, 3, 4 valences, but no valence 1.

Gyuri wrote:

Quote: The only incorrect valences for metals exceed the number of valence electrons of the element (e.g., +5 for Ti). I am not sure. For example, according to the literature, Ti can have 2, 3, 4 valences, but no valence 1.

That's true until someone discovers a Ti(I) complex.





Or it may be discovered that a Ti(I) compound is an unstable intermediate in some reaction, and someone wants to draw it.





Believe me, you're better off not checking valences of metals.

OK, and believing you makes the code simpler.

There you go! If sense doesn't convince you, convenience surely will. :-)

:-)

Bob,


We have solved this issue. Marvin 4.1 will do what you suggested. Thanks for the suggestions.

Hi folks,





Marvin 4.1 is a big improvement for transition metals; I no longer get valence errors for Pd(0). However, if I want to draw (Ph₃P)₂PdCl₂, I still need to put + charges on the P atoms and a 2- charge on the Pd. A dative bond would be a great help in making these structures appear more "natural".





Something else Marvin needs: a bond that connects a bond (pi or sigma) to an atom. A dative bond could be used for this purpose as well.





Finally, if you're feeling really ambitious, we need a way of representing eta3-allyl, eta5-cyclopentadienyl, and eta6-benzene ligands for transition metals. Normally the electrons in the pi systems are drawn as a curve (the first) or a circle (the latter two). Marvin already has a concept of a circle of electrons, so it should be a small matter to have the dative bond point from the center of the circle to the metal. I'm not sure how you would handle the allyl group, though.





-- Bob

You mean something like this?

Yes, although the Cl atoms should be covalently bound, not datively, and it would be nice to be able to specify whether the dative bond should be shown as an arrow or just a line. Did you make that drawing in Marvin? If so, how?

Uh, it is too late for me to draw such structures using a half-ready feature ;) Still, as the implementation of dative bonds goes on, can I show you some of the representation for testing?

Sure, absolutely.





Your timing is impeccable. I'm thinking about how to use Marvin to draw organometallic reaction mechanisms.

Here is something else, which might be useful, and which is on the way. Coordinative multicenter bonds.

That will be very helpful, but we need a circle in the Cp ligands, or it looks like (C₅H₆)₂Fe, not (C₅H₅)₂Fe.





On the other hand, there are such things as eta⁴-1,3-diene complexes, where the dienes may be s-cis or s-trans. The usual way to represent them, though, is to draw one bond from each isolated pi bond to the metal.

Yes, that is right. The corresponding hydrogen handlling and aromatization is on our todo list (to have (C5H5)2Fe).





You will be able to place the multicenter between two atoms as well.