2 internal server vlans

Depending of your network architecture:

1. If your servers are meant to be reachable through a level-3 equipment (a firewall typically) then you don't need to create IPs on the servers subnet as they are not meant to be directly connected. You need instead to have a nework route to reach your servers.

2. If your servers are meant to be directly connected to F5 e.g. no routing involved, then you need to have L2 and L3 connectivity between F5 and your servers. This is a simple process :

1. Optional and not required but this is my way to go when using VMs with a small number of VLANs: let the hypervisor manages VLANs: Assign your 4th VM interface to the corresponding net or port group, and in the next step use the corresping (1.3) interface inside F5 
2. New VLAN creation, create the VLAN inside F5 and assign 1.3 as untagged
3. Create self IP on the new subnet

Alternatively, you can choose to leverage VLAN tagging and use same interface for both VLANs, this is possible too.

You can also choose to not use any of these methods and just create the self IP if your two subnets belong to the same VLAN, but this is quite a rare condition in production environment

tmsah ,from your question, I think you implemented your LTM as a two-arm design, 

For the answer, you have the back-end servers in different subnets, you will not need to create another interface for the new two servers, we just need to enable auto map on your new virtual server that will take an IP from the external VLAN range and attach the pool contain the two new servers , so now you use your LTM as one arm and it is applicable to add any servers from any subnets