How do I combine z-matrices of multiple molecules into one input stack?

I'd like to run a calculation in Gaussian which requires me to provide cartesian coordinates for reactants and products respectively. My question is how do I combine the cartesian coordinates for two reactants (that I would obtain from a database)? For instance, CH3 + N?

Is there a specific program/procedure that I need to use?

To be more specific, say I would like to combine the z-matrix of $\ce{CH3}$

0 3
C
X 1 1.0
H 1 r 2 90.0
H 1 r 2 90.0 3 120.0
H 1 r 2 90.0 3 -120.0

r 1.0828


with a z-matrix of $\ce{N}$

0 3
N


And I would like to combine the z-matrix of $\ce{HCN}$

0 1
C
X 1 1.0
H 1 rh 2 90.0
N 1 rn 2 90.0 3 180.0

rh 1.0587
rn 1.1326


with that of $\ce{H2}$

0 1
H
H 1 r

r 0.741


In other words, I want to combine these z-matrices (or geometries) of the two reactant/ product species into a single coordinate, so that I can use them for my QST2 calculation. What do I do?

• "Combine" is an awfully broad term. For example, copying and pasting the cartesian coordinates of two reactants into one file will surely combine them, though probably not in the way you want. – Ivan Neretin Nov 28 '17 at 4:57
• Thank you for responding! I edited it to be more specific. Could you help me please? – pennypeat Nov 28 '17 at 5:02
• I clarified your request, however, if you don't know how to construct valid input geometries for Gaussian, you should not be playing around with transition state searches, but rather learn the fundamental basics of how to optimise a molecule. What software do you use to visualise your molecules? I am 99% certain, there is a molecular editor included. – Martin - マーチン Nov 28 '17 at 5:55

The Setup

The first thing you need to do is to get the difference between a Z-matrix (sometimes also called internal coordinates) and cartesian coordinates straight. They are truly different things.

Secondly, take the Z-matrices of a sufficiently large molecule. The basic structure is:

C1 .  .  .
C2 r2 .  .
C3 r3 a3 .
C4 r4 a4 d4
C5 r5 a5 d5
(and so on from this point)


where in this simplified notation, I have omitted the anchor centers for the coordinates. One notices that there are 6 coordinates left out, marked by dots. They can be omitted because three degrees of freedom for translation and three degrees of freedom for rotation of the entire molecule are irrelevant for the calculation of properties.

Now imagine a second molecule tacked onto the other one:

C1 .  .  .
C2 r2 .  .
C3 r3 a3 .
C4 r4 a4 d4
C5 r5 a5 d5
C6 *  *  *
C7 r7 *  *
C8 r8 a8 *
C9 r9 a9 d9


This Z-matrix is invalid. One needs to fill the missing coordinates marked by the asterisks. In a sense, the second molecule does not have the 6 degrees of freedom mentioned above, because any change in them would change the orientation relative to the first one, impacting all properties.

So, what can you do?

Filling in the missing coordinates is hard, at the very least by hand or some other automated fashion. Doing it in an editor is basically the same amount of work as the solution I actually suggest: Create the system of both molecules in one Z-matrix and split that Z-matrix. Remove the superfluous coordinates in the resulting matrices and run calculations as necessary.

• Splitting a z-matrix also seems quite tedious though. One notices that there are 6 coordinates left out, marked by dots. They can be omitted because [...] I wouldn't say they are omitted. You simply cannot define them, because you do not have a reference. With one point you cannot define a line (or an angle, or a dihedral) etc.. – Martin - マーチン Nov 28 '17 at 6:46
• OK thank you for clarifying that and I apologize for my ignorance but I still don't understand what "splitting w Z-matrix" entails. Would be so kind as to show me an example of splitting? Thank you! – pennypeat Nov 28 '17 at 13:31
• @pennypeat The split would be the opposite of what I did with tacking on a second molecule. I'm afraid you will have to get a molecular editor and play with Z-matrices for a while. – TAR86 Nov 28 '17 at 14:39