Help with Marginal IC Power-on Issue

[jayalt]

Hi all. Hope to get some opinions and suggestions on what to look at here.

Long story short- I am having marginal power-on issues on an IC. Issues arose after I changed all the power components on my PCB including DC-DC regulators, load switches, and LDOs. This was out of necessity due to supply chain concern. I have two voltage rails going into this particular IC- an IO 1.8V, and an analog / VDD core 0.9V. When the device boots into this failed state- it does not run any code, and does not respond to a reset pulse. I need to power cycle in order for the chip to come back. Certain units are much worse than others. I can tell it has failed within a few ms after power up.

The IC I am using does not have any specified power on sequence, no errata, and very limited support. There is no power rail sequencing defined, or maximum slew rate times published. I am basically working in the dark, and I cannot just switch vendors- so I am hoping that some experts might understand what is happening at the lower, internal level, and help me root cause this.

My main observations so far:
1. My rise times are significantly slower with updated power components. About 2.5ms for the 1.8V and 700uS for the 0.9V. Previously they were 600us and 30us respectively.
2. I have tried a bunch of different power sequences with no luck. I am still waiting on support from supplier on the correct sequence but it has been weeks.
3. I have been able to "fix" the issue with the following observation: if I disable power before my load switch has a chance to pull down the 1.8V rail completely, it leaves some residual voltage that drains slowly. If I power on before the 1.8V rail has a chance to completely discharge, then I can turn on the IC just fine. Maybe issue relates to my load switch?

I'm having trouble understanding what is going on in observation 3? I haven't necessarily made the rise time any shorter - when I enable the load switch at power on with a small amount of voltage already existing at its output, I see the rail start to decrease a bit prior to rising - so the total rise time remains the same. Even residual voltage as low as 100mV on the 1V8 seem to work. I thought this might point to an inrush issue- so I removed some capacitance on the rail hoping to rectify it but didn't see any change. Any ideas?

[thm_w]

Schematic?
Part number of the IC?