101 lines
6.9 KiB
XML
101 lines
6.9 KiB
XML
// We investigate phonons at Gamma instead of the exact location near Gamma.
|
||
Phonons at the #sym.Gamma point were used
|
||
to approximate negligible-polar phonons that participating in Raman processes
|
||
regardless of the wavevector of the incident and scattered light.
|
||
This approximation is widely adopted (cite) and justified by the fact that,
|
||
although the phonons participating in Raman processes are not these strictly located at the #sym.Gamma point,
|
||
they are very close to the #sym.Gamma point in reciprocal space
|
||
(about 0.01 nm#super[-1] in back-scattering configurations with 532 nm laser light,
|
||
which corresponds to only 1% of the smallest reciprocal lattice vector of 4H-SiC,
|
||
see orange dotted line in @figure-discont),
|
||
and their dispersion at #sym.Gamma point is continuous with vanishing derivatives.
|
||
Therefore, negligible-polar phonons involved in Raman processes
|
||
have nearly indistinguishable properties from those at the #sym.Gamma point.
|
||
|
||
#include "figure-discont.typ"
|
||
|
||
// Representation of these 18 phonons, and the shape of their Raman tensors could be determined in advance.)
|
||
Phonons at the #sym.Gamma point satisfy the C#sub[6v] point group symmetry,
|
||
and the 18 negligible-polar phonons correspond to 12 irreducible representations of the C#sub[6v] point group:
|
||
2A#sub[1] + 4B#sub[1] + 2E#sub[1] + 4E#sub[2].
|
||
Phonons belonging to the A#sub[1] and B#sub[1] representations vibrate along the z-axis and are non-degenerate,
|
||
while those belonging to the E#sub[1] and E#sub[2] representations vibrate in-plane and are doubly degenerate.
|
||
Phonons of the B#sub[1] representation are Raman-inactive, as their Raman tensors vanish.
|
||
In contrast, phonons of the other representations are Raman-active,
|
||
and the non-zero components of their Raman tensor
|
||
can be determined by further considering their representation in the C#sub[2v] point group (see @table-rep).
|
||
These Raman-active phonons are potentially be visible in Raman experiment under appropriate polarization configurations.
|
||
However, whether a mode is sufficiently strong to be experimentally visible
|
||
depends on the magnitudes of its Raman tensor components,
|
||
which cannot be determined solely from symmetry analysis.
|
||
|
||
#include "table-rep.typ"
|
||
|
||
// We propose a method to estimate the magnitudes of the Raman tensors of these phonons,
|
||
// without first-principle calculations.
|
||
// Here we only write out results, details are in appendix.
|
||
|
||
// TODO: maybe it is better to assign Raman tensor to each bond, instead of atom
|
||
|
||
We propose a method to estimate the magnitudes of the Raman tensors of these phonons based on symmetry analysis.
|
||
This approach is founded on the assumption that the change in polarizability induced by atomic displacements in 4H-SiC
|
||
is primarily determined by the first- and second-nearest neighbors of the atom and the sign of the atomic charge,
|
||
while other factors (mass, bond length, etc.) only have small contributions.
|
||
As a result,
|
||
the phonon modes with the strongest Raman intensities can be predicted
|
||
prior to first-principles calculations and experiments,
|
||
and the Raman tensors of the calculated phonon modes can be estimated without additional first-principles computations.
|
||
Further details are provided in the appendix.
|
||
|
||
The Raman tensors and frequencies of the negligible-polar phonons were calculated using first-principles methods,
|
||
and the results are compared with experiment and theory (@table-nopol).
|
||
Calculated frequencies of these phonons are consistent with the experimental results
|
||
with a low-estimated error of about 2% to 5%, which might be due to the PBE functional used in the calculation (cite).
|
||
The Raman tensors of these phonons are also consistent with the experimental and theoretical results,
|
||
where E#sub[2] mode experimentally at 776 cm#super[-1] (mode 8) is the most intense phonon mode,
|
||
followed by four modes visible in experiment with lesser intensities,
|
||
including E#sub[2] modes at 195.5 cm#super[-1] (mode 1) and 203.3 cm#super[-1] (mode 2),
|
||
E#sub[1] mode at 269.7 cm#super[-1] (mode 3), A#sub[1] mode at 609.5 cm#super[-1] (mode 6).
|
||
The Raman scatter of the E#sub[1] mode calculately at 746.91 cm#super[-1] (mode 7)
|
||
and E#sub[2] mode calculately at 756.25 cm#super[-1] (mode 9)
|
||
are much weaker than the most intense mode but located near it, according to our calculation,
|
||
thus it could not be distinguished from the most intense mode,
|
||
which explains why they are not observed in experiments.
|
||
Moreover, the A#sub[1] mode calculated at 812.87 cm#super[-1] (mode 10)
|
||
have a very weak Raman intensity in the basal plane (xx and yy, only 0.01)
|
||
but an observable intensity in the zz configuration (1.78).
|
||
Thus, this mode could not be observed in most Raman experiments (cite),
|
||
but could be observable when incident light propagate not along the z-direction (our experiment),
|
||
or the incident light wavelength is near the resonance condition (cite).
|
||
|
||
Besides, there are other peeks in the experiment.
|
||
The peek at 796 and 980 are caused by strong-polar phonons which will be discussed later.
|
||
Besides, there are small peeks at xxx,
|
||
which could not be explained in perfect 4H-SiC and will be discussed in the next section.
|
||
|
||
The method only takes the vibration directions of each atom in each phonon mode,
|
||
leaving the amplitudes unconsidered (see appendix for details),
|
||
and the result was summarized in @table-predmode.
|
||
In the Raman tensors in @table-predmode,
|
||
$a_i$ corresponding to the change of polarizability caused by movement of the Si atoms in A and C layers,
|
||
$epsilon_i$, $eta_i$ and $eta_i$ corresponding to the difference between different bilayers and different atoms.
|
||
Due to the similarity of environment in different bilayers and around different atoms,
|
||
the absolute values of $epsilon_i$, $eta_i$ and $zeta_i$ are expected to be much smaller than that of $a_i$,
|
||
thus the Raman tensors containing $a_i$ are expected to be much larger than those not containing $a_i$.
|
||
|
||
|
||
// TODO: 将一部分 phonons 改为 phonon modes
|
||
// 在论文中我们这样来称呼：phonon 对应某一个特征向量，而 modes 对应于一个子空间。
|
||
// 也就是说，简并的里面有两个或者无数个 phonon，但只有一个 mode
|
||
|
||
#include "table-nopol.typ"
|
||
|
||
#include "raman.typ"
|
||
|
||
// 实验与计算基本相符。对于声子频率，计算总是低估大约 3%。
|
||
// 此外，一些较强的模式在预测无法看到的偏振中也可以看到。例如，一些在 xy 偏振中不应该看到的模式可以被看到了。
|
||
// 这个现象可以由 4度的斜切所解释：我们将材料略微踮起一些角度，就可以使得该模式减小。
|
||
// 这个现象也可以由材料或偏振片的微小角度来解释。
|
||
// 例如，我们将偏振方向转动 5 度，就可以得到这个模拟结果。
|
||
// 此外，由于使用的材料是沿着 c 轴切片的，所以我们在测量 y 入射时不得不将片子以略小于 90 度（约 75 度）的角度放置。这也导致实验与计算的偏差。
|
||
// TODO: 翻译成英文 |