c******k
发帖数: 1140 | 1 就是在SOI wafer(0.75um Si and 1.0um SiO2)上设计一个单模的Ridge waveguide.用
经典公式近似了一下，参数如下：
Width: 1.5um, total waveguide height: 0.75um, etching depth: 0.1um. 所以，
Slab height: 0.65um
但是想用设计软件验证一下，实验室有Rsoft CAD photonic suite, 结果请见图：支持
N多mode, 这个Rsoft总是把Slab也当成波导的一部分，怎么办，是软件参数设置的不对
吗？谢谢 | j**y
发帖数: 7014 | 2 把基模的照片啥的也发上来看看
【在 c******k 的大作中提到】
: 就是在SOI wafer(0.75um Si and 1.0um SiO2)上设计一个单模的Ridge waveguide.用
: 经典公式近似了一下，参数如下：
: Width: 1.5um, total waveguide height: 0.75um, etching depth: 0.1um. 所以，
: Slab height: 0.65um
: 但是想用设计软件验证一下，实验室有Rsoft CAD photonic suite, 结果请见图：支持
: N多mode, 这个Rsoft总是把Slab也当成波导的一部分，怎么办，是软件参数设置的不对
: 吗？谢谢
| c******k
发帖数: 1140 | 3 好吧，请看
【在 j**y 的大作中提到】
: 把基模的照片啥的也发上来看看
| w*******p
发帖数: 253 | 4 Run slab mode analysis first (no ridge shape), find the 1st order mode Nref
of slab mode.
When you simulate the slab with ridge, use the above index as the Nmin to
make sure the slab mode will not be included.
【在 c******k 的大作中提到】
: 就是在SOI wafer(0.75um Si and 1.0um SiO2)上设计一个单模的Ridge waveguide.用
: 经典公式近似了一下，参数如下：
: Width: 1.5um, total waveguide height: 0.75um, etching depth: 0.1um. 所以，
: Slab height: 0.65um
: 但是想用设计软件验证一下，实验室有Rsoft CAD photonic suite, 结果请见图：支持
: N多mode, 这个Rsoft总是把Slab也当成波导的一部分，怎么办，是软件参数设置的不对
: 吗？谢谢
| w********o
发帖数: 10088 | 5 赞
optibpm里面能这么搞么？我以前也有过这个问题，没解决好
Nref
【在 w*******p 的大作中提到】
: Run slab mode analysis first (no ridge shape), find the 1st order mode Nref
: of slab mode.
: When you simulate the slab with ridge, use the above index as the Nmin to
: make sure the slab mode will not be included.
| w*******p
发帖数: 253 | 6 I didn't use this software before, so I don't know, Rsoft is very popular in
this area...
【在 w********o 的大作中提到】
: 赞
: optibpm里面能这么搞么？我以前也有过这个问题，没解决好
:
: Nref
| c******k
发帖数: 1140 | 7 请问run 这个slab waveguide 的slab mode时的slab宽度设置多少合适？
不同的slab宽度的1st order slab mode的Neff 还不一样啊，
比如现在设成
20um width: Neff=3.363936（1st order slab mode）
10um width: Neff=3.363858（1st order slab mode）
Nref
【在 w*******p 的大作中提到】
: Run slab mode analysis first (no ridge shape), find the 1st order mode Nref
: of slab mode.
: When you simulate the slab with ridge, use the above index as the Nmin to
: make sure the slab mode will not be included.
| w*******p
发帖数: 253 | 8 It doesn't matter that much, you can the width which is close to your real s
ituation.
Use this 0th mode of the slab as Nmin when you simulate ridge structure, the
simulation will only show the fundamental mode of the ridge structure.
【在 c******k 的大作中提到】
: 请问run 这个slab waveguide 的slab mode时的slab宽度设置多少合适？
: 不同的slab宽度的1st order slab mode的Neff 还不一样啊，
: 比如现在设成
: 20um width: Neff=3.363936（1st order slab mode）
: 10um width: Neff=3.363858（1st order slab mode）
:
: Nref
| c******k
发帖数: 1140 | 9 Many thanks for your reply.
I realize that you mean I should calculate Neff of 0th mode of the 2D slab
waveguide. In this case, it does not matter the slab width, since the
confinement only happen in the vertical direction, namely 0.65um.
However, if I intend to set a width value for this slab waveguide, this slab
waveguide actually become a strip waveguide (still 3D structure). So the
Neff of a fundamental mode will change with the slab width.
My understanding is correct?
s
the
【在 w*******p 的大作中提到】
: It doesn't matter that much, you can the width which is close to your real s
: ituation.
: Use this 0th mode of the slab as Nmin when you simulate ridge structure, the
: simulation will only show the fundamental mode of the ridge structure.
| w*******p
发帖数: 253 | 10 Yes, the slab w/o ridge is still a 3D structure, assuming your cross section
is in XY plane, progation direction is Z.
If you use different widths (X direction), the calculated Neff is slightly d
ifferent, which has been verified by your calculation.
【在 c******k 的大作中提到】
: Many thanks for your reply.
: I realize that you mean I should calculate Neff of 0th mode of the 2D slab
: waveguide. In this case, it does not matter the slab width, since the
: confinement only happen in the vertical direction, namely 0.65um.
: However, if I intend to set a width value for this slab waveguide, this slab
: waveguide actually become a strip waveguide (still 3D structure). So the
: Neff of a fundamental mode will change with the slab width.
: My understanding is correct?
:
: s
| | | c******k
发帖数: 1140 | 11 You help me a lot! Two more questions:
1.The calculated Neff of a 0th slab mode with various widths as follows:
1.6um: 3.364379
10um: 3.363858
20um: 3.363936
50um: 3.364155
100um: 3.364166
You suggest me to choose a slab width that is close to a real situation.But
in fact we can consider the slab width to go infinite.So right now I have to
choose Neff with 3.364 as an average value. Any comments about this value?
2. If I keep other parameters same for a ridge waveguide, but just change a
ridg
【在 w*******p 的大作中提到】
: It doesn't matter that much, you can the width which is close to your real s
: ituation.
: Use this 0th mode of the slab as Nmin when you simulate ridge structure, the
: simulation will only show the fundamental mode of the ridge structure.
| w*******p
发帖数: 253 | 12 1. This Neff is ok to be used as Nmin when you calculate ridge structure.
2. Marcatili has a famous paper on this topic, you may want to read it. The formular is for approximate calculation, not accurate. I am not sure if Rsoft uses very fine meshes near boundaries, if you say "the width can be up to 2.5um", you really want to make sure you are in the middle of some safe region, which allows calculation/process tolerance. For example, if the Rsoft calculation and formular give you different sing
【在 c******k 的大作中提到】
: You help me a lot! Two more questions:
: 1.The calculated Neff of a 0th slab mode with various widths as follows:
: 1.6um: 3.364379
: 10um: 3.363858
: 20um: 3.363936
: 50um: 3.364155
: 100um: 3.364166
: You suggest me to choose a slab width that is close to a real situation.But
: in fact we can consider the slab width to go infinite.So right now I have to
: choose Neff with 3.364 as an average value. Any comments about this value?
| c******k
发帖数: 1140 | 13 Another thing in Rsoft BeamProp simulation that really bother me is how to
choose right compute step sizes for XYZ directions.
For example, to calculate the Neff of 0th order mode in a slad waveguide,
say 15um slab width, keeping a same X domain as -50um~50um for all the cases
. 0.05 step size will have Neff=3.360727, and 0.02 step size will produce 3.
363853, and 0.01 size will lead to 3.364009. Smaller compute size will
always have a better accurate result?? But take too much time.
Could you
【在 w*******p 的大作中提到】
: It doesn't matter that much, you can the width which is close to your real s
: ituation.
: Use this 0th mode of the slab as Nmin when you simulate ridge structure, the
: simulation will only show the fundamental mode of the ridge structure.
| w*******p
发帖数: 253 | 14 Usually the smaller step, the better accuracy, but more time-consuming.
Rsoft doesn't support localized mesh reshape, which means you can make the
meshes smaller in some critical region. So you have to live with that.
This is one of the software calculation tolerances I mentioned to you in
previous message, if the difference is within a certain range (for example,
diff steps give 0.1% difference in Neff), you can choose the bigger step to
save time.
Think about this, the refractive index you use
【在 c******k 的大作中提到】
: Another thing in Rsoft BeamProp simulation that really bother me is how to
: choose right compute step sizes for XYZ directions.
: For example, to calculate the Neff of 0th order mode in a slad waveguide,
: say 15um slab width, keeping a same X domain as -50um~50um for all the cases
: . 0.05 step size will have Neff=3.360727, and 0.02 step size will produce 3.
: 363853, and 0.01 size will lead to 3.364009. Smaller compute size will
: always have a better accurate result?? But take too much time.
: Could you
| w********o
发帖数: 10088 | 15 有时候可能不一定step越小越好,会有round-off error.模拟的东西感觉结果大概定性
就行了,反正不可能跟实验100%吻合.可以试试看用不同的step,看结果啥时候开始
converge
,
to
.1
【在 w*******p 的大作中提到】
: Usually the smaller step, the better accuracy, but more time-consuming.
: Rsoft doesn't support localized mesh reshape, which means you can make the
: meshes smaller in some critical region. So you have to live with that.
: This is one of the software calculation tolerances I mentioned to you in
: previous message, if the difference is within a certain range (for example,
: diff steps give 0.1% difference in Neff), you can choose the bigger step to
: save time.
: Think about this, the refractive index you use
| w*******p
发帖数: 253 | 16 Good point!
【在 w********o 的大作中提到】
: 有时候可能不一定step越小越好,会有round-off error.模拟的东西感觉结果大概定性
: 就行了,反正不可能跟实验100%吻合.可以试试看用不同的step,看结果啥时候开始
: converge
:
: ,
: to
: .1
|
|
