The 3dB bandwidth extracted from the lateral PD


dear everyone
I have learnt the lateral PD on your website about the lpd_SiGe example.I found that in the example you just do the simulation about responsivity of different design.Now I want to do simulation of the 3dB bandwidth about the lateral structure photodetector.The 3dB bandwidth extracted from the first example(VPD example) is not adapted in my simulation because of the 2D simulation,which can not get the relationship of the 3dB bandwidth and the device length.Did you do any simulation about the 3dB bandwidth of the lateral PD with other method?Can anyone help me solve the problem.Thanks a lot.


Hi. In your actual photodetector are you using point contacts or a single slab contact? If you are using a single continuous contact then you can use the same approach used in the VPD example and perform a transient 2D simulation. However, if you have point contacts then you will have to use a 3D simulation since the response time will depend on the positioning of the contacts.


dear aalam
if I use the same approach used in the VPD example and perform a transient 2D simulation,I can’t get the relationship between the device length and the 3dB bandwidth. How do you think of the question. Can I use a 3D simulation if I use a single continuous contact ? Furthermore, it will take a lot of time to finish a 3D simulation in Device and sometimes it will cause the computer crashed. How to improve the simulation speed?Can you give me some suggestion.Thanks a lot.


Hi. The bandwidth of the photodetector depends on two things, (i) transit time of the photo-generated carriers and (ii) the RC constant. Considering the case of continuous contact,

(i) If the bandwidth is limited by carrier transit time then you should not see any significant effect of length on the bandwidth since the photo-generated carriers will move towards the contacts primarily on the place perpendicular to the photodetector length (because the electric field is in the transverse direction compared to optical propagation). Changing the length should not affect the transit time of the carriers in the transverse direction.

(ii) If the bandwidth is RC limited then the value of the RC time constant will change with length and the bandwidth will vary with length. You can still perform a 2D simulation (given that you have continuous contacts) and get the effect of length. In reverse bias, the equivalent circuit of the pin diode looks like a capacitor Cd in series with a very small resistor Rd (figure below). This diode will be in series with the load resistance RL (say 50 Ohm). As the length is increased, Cd increases and Rd decreases linearly. If the RL is very small compared to Rd then the RC constant would be Cd x Rd and the bandwidth would be insensitive to length (since one of them increases and the other one decreases with length). However, in practical devices RL is much larger than Rd and the equivalent resistance is almost constant (~RL). As a result the capacitance increases with length but the resistance remains fixed with increasing length. This means that the RC constant increases with the length of the PD and the bandwidth decreases.

To model this effect of RL all you need to do is to assign a load resistance at the ground contact in your 2D simulation which will remain fixed as you change the “norm length” in your 2D simulation to look at the effect of PD length.

Transient & frequency response

dear aalam,
I had already added the Rl in this model, but when I change the “norm length” in 2D simulation, the 3dB bandwidth don’t changed any more. I am very confused that if I can get the relationship of the 3dB bandwidth and the device length in 2D simulation?


Hi @mapengcheng, I think the last post I made was a bit confusing. My apologies. The transient simulation that we are doing for the photodetector gives us the bandwidth due to the carrier transit time and not due to RC time constant. Therefore, when you change the load resistance, you would not see a change in the bandwidth. For similar reasons, when you change the normalization length of the 2D simulation, you also do not see an effect of the bandwidth since it only depends on the transit time of the carriers on the 2D simulation plane. For devices where the bandwidth limitation comes from the RC time constant, you will see an effect of the load resistor but not for this particular simulation setup as it only captures the transit time limited bandwidth. If you want to model the effect of length on the transit time limited bandwidth, then the solution would be to do multiple 3D simulations with different lengths (which can be computationally very demanding unfortunately).


Much appreciation for your help. I will try on this 3D simulation.