Feedback welcome: yeschowh@snu.ac.kr :)
Partially inspired by this page.
Vibrational Modes By Irreducible Representations: Case of the NH3 Molecule
If you are not familiar on this kind of topic, I recommend reading H2O case first. :)
Symmetry of NH₃ Molecule: C₃ᵥ Point Group
The ammonia (NH3) molecule belongs to C3v point group, which characterizes its symmetry.
This can be understood by considering the molecule's symmetry operations:
E (Identity): No operation; leaves the molecule unchanged.
C₃ (Three-fold rotation about z-axis): There are one rotation axis which rotates the molecule by 120° around the z-axis results in an indistinguishable configuration.
σᵥ (Vertical mirror planes): There are three mirror planes (σᵥ) each containing the z-axis and bisecting the molecule through one N-H bond and between the others.
These symmetry elements reflect the molecular structure of NH₃, with its trigonal pyramidal shape (nitrogen at the apex and hydrogens at the base), where the molecule maintains its identity under these operations.
Regular representation of C₃ᵥ Point Group
You can easily find the character table of C3v point group by searching online.
While we could directly derive the character table from the multiplication table (or regular representation), let's focus on understanding the underlying concepts first.
Instead, let's find out what is multiplication table and regular representation and why it is sometimes useful.
Multiplication table:
(Group) multiplication table is defined as follows, I think instead of long explanation, direct demonstration for C3v point group would be much intuitive explanation.
Ta da! You may easily figure out what multiplication table is.
Table which shows the relation between the combination of each operation. We have 6 operations, therefore it should be 6x6 table.
A group is closed under its operations; combining any two elements must yield another element of the group.
(I did not mention the formal definition of group, sorry about that..)
Regular representation:
The multiplication table provides additional insight into the group structure. As shown in the figure, we can construct one matrix representation from the multiplication table. We call this as regular representation.
I won’t delve into a formal proof of why this matrix representation satisfies the group’s structure, (since I do not know).
However, such matrix representations are particularly useful for less common groups where standard tables are not available—though that’s not the case for C3v.
For the ones who will not believe my story, I prepared a code block (of julia), which can compare the multiplication result with above table. Regular representation case? I remained the most fun part for you!
using LinearAlgebra, Rotations
# Define the symmetry operations as 3x3 matrices
E = Matrix{Float64}(I, 3, 3)
# E: Identity
θ1 = 2π / 3; R1 = Matrix{Float64}(RotZ(θ1));
# R1: 120 degrees rotation about z-axis
θ2 = 4π / 3; R2 = Matrix{Float64}(RotZ(θ2));
# R2: 240 degrees rotation about z-axis
n1 = [1, 0, 0]; r1 = E - 2 * (n1 * n1');
# r1: Reflection across the yz-plane
n2 = RotZ(2π/3) * n1; r2 = E - 2 * (n2 * n2')
# r2: Reflection across a plane rotated 120 degrees from yz-plane
n3 = RotZ(4π/3) * n1; r3 = E - 2 * (n3 * n3')
# r3: Reflection across a plane rotated 240 degrees from yz-plane
# Example: Test some multiplications
res = r1 * r2; res - R1
Long story, but anyway, useful point of regular representation is that this contains the whole irrep for the given group.
Regular representation contains the 1d irrep for 1 time, (if exists) 2d irrep for 2 times, (if exists) 3d irrep for 3 times, etc.
(Again, I don't know the detailed derivation, but) this provide one mysterious identity of character table.
square of character table of column of E operation is the same with group order |G|.
Since regular representation (1) contains n-dim irrep for n times and (2) matrix dimension is |G|.
For C3v case, 1* 1 + 1 * 1 + 2 * 2 = 6.
Since each rotation / reflection shares the same character, they are gathered in the same column.
(Just in case) E in green region does not stand for trivial representation, which maps every operation into 1. This trivial representation corresponds to A1 irrep. This combination of alphabet - number symbol has a rule of naming, Mulliken symbol.
We can also check by projection operator that, A1 & A2 are included once in regular representation, where E is twice included in regular representation.
So in principle, if you know multiplication table, you already have all irrep inside your regular representation.
Of course, reducing this reducible representation into block-diagonal irrep is another problem.
I heard that there is a general scheme to reduce large reducible representation into block diagonal irrep, this paper, but I haven't tried this yet, I will update it after trying it.
Vibrational Degree of Freedom from Displacement Representation
Actually, remaining part is fundamentally the same with previous H2O case, except for this case, not just assigning +1 / 0 / -1 for coordinate transformation for x, y and z.
Let's summarize the process, again without formal proof.
(1) Check the number of atoms position unchanged, N
(2) For coordinate axis, calculate trace of the operation, tr(R)
Character of displacement representation is given as N * tr(R).
For instance, character of C3 operation in displacement representation is calculated as follows:
Similarly, character table of displacement representation could be given as follows.
Also, displacement representation contains 3 translational and 3 rotational motion in its representation.
We need to identify which representation corresponds to translation or rotation.
Actually, this part is already done for many cases by commonly available tables.
If you searched character table for C3v, you may have noticed that there are some letters given right to the table.
I searched random character table with additional two column
As you can see, it is already tabulated in additional two column stands for linear, rotations and quadratic
From this, you can identify which irrep corresponds to translation, rotation part.
But how do we know about this?
Identification of translation and rotation motion from character table
It is convenient to start with vector and pseudovector to deal with translation and rotation part from displacement representation.
Translation could be considered as vector, with respect to its response on translation, rotation, and reflection.
On contrary, rotation could be considered as pseudovector, with respect to its response on translation, rotation, and reflection.
I don't know whether this kind of approach is canonical ways of understanding this formalism, but anyway, I think it still works..!
Mostly similar, expect reflection.
For vector, normal component (to mirror plane) flips its direction after reflection.
For pseudovector, parallel component (to mirror plane) flips its direction after reflection.
Translation part for (x, y, z), vector, respect to symmetry operation of C3v leads to the following form.
Let's take a look at x case only. y / z can be done similarly.
x transforms just like (normal) vector.
We arrange the character table for x, similarly done for displacement representation cases.
Rotation part for (Rx, Ry, Rz), pseudovector, respect to symmetry operation of C3v leads to the following form.
Here, explicit form for Rx case. Ry, Rz case is similar.
Rx transforms just like pseudo-vector.
We arrange the character table for x, similarly done for displacement representation cases.
Seems tricky, but if you draw stick with ring for pseudovector instead of arrow with arrowhead, it makes sense.
We can arrange the whole character table for (x, y, z) and (Rx, Ry, Rz) as follows.
It seems like that combined result for (x, y) and (Rx, Ry) responds like irrep E.
Frankly speaking, I did not fully understand on this process.
I just understand this as:
(1) If character table obtained from this process contains non-integer value, such as +1/2 or -1/2, then it is mixed with another degree of freedom, which means just tracking x or y is not enough in this case.
(2) This could be interpreted as these two degree of freedom could not be decomposed into two separate 1D irrep, which means (x, y) or (Rx, Ry) combined term could be mapped as 2D irrep E, in this case.
This seems quite arbitrary, but still we can get a rough estimate on how those translation / rotation could be mapped to one irrep.
There would be much more smart way to convey this, but I am not sure right now. I just learned this empirical way..
If you do, please let me know! :)
(x, y): E / z: A2
(Rx, Ry): E / Rz: A1
Therefore, if we can evaluate that NH3 molecule has a vibration mode as follows:
I spent too much on regular representation blah blah too much. Let's not do how those eigen-mode are obtained.
I assume that would not be much different from H2O case.
Just for check, below 6 modes are vibration mode of NH3 molecule.
I adopted from this paper.
(Not sure why a2'' for Pyramidal inversion mode)
Or, nice moving figure from this site.
This site shows we are on a right track, 2A1 + E.
Thank you for the reading!