[ortek_service]

Originally Posted by GeraldSommariva

Well I blasted the ceramic substrate with a hot air solder station at 400c then used a soldering iron then wiped but the ball still remains.

Yes, having a closer comparison between marks left behind on the substrate (and that two pads are also slightly-closer on one side, as well as the contact-marks), then the chip must have been fitted "upside-down", similar to more recent Chip Scale Packaging of the die being mounted directly on the PCB (Similar to BGA, but that often has a very small thin carrier PCB that the die is mounted to / often plastic over-moulded with part numbering on top, rather than just the raw die)
- It had been difficult to tell on a fitted-one, because all the covering adhesive had masked clear visibility of the surface

I did wonder what would have prevented shorts to the rest of the surface of the chip, but it probably has some clear passivation layers over that. And also, I note from this photo that there was also clear adhesive over the surface, with pockets around the pads, as can be seen on this photo: https://sites.google.com/view/transa...ensor-module-5
However, rather than the chip originally being covered with protective clear adhesive, it may well have flowed underneath it when the the top surface of the substrate had been covered with the protective clear dome of this.

It was difficult to tell from the substrate photos, whether the marks on the pads were actually raised above the surface. But if so, then most likely it was actually silver-loaded conductive epoxy, that has sheared away from the pads on the die.
And,given that the whole of substrate's pads are not covered with the adhesive, it does look like there was originally small balls of it applied directly to the chips surface, then accurately positioned on the substrates - probably all done semi-automatically as would be difficult to get the consistency by hand.

It does seem most likely that the chip was originally mounted with this conductive-epoxy, as I have seen accelerometer sensors that used this - Although they were ceramic packaged , and just had 2 rectangular gold pads. The originally had recommended much more exotic Gold epoxy, but silver could also be used. However, later versions just used solder-paste to make it much easier to mount with the other Surface Mount auto-assembling.

If you can get solder paste to form solder-balls on the chip's pads, by applying and heating with hot air pencil, then it may be possible to flip it over and re-heat from other side to reflow it onto the substrate (after applying solder paste to that). However, heating the ceramic substrate enough might be tricky, without a hot-plate designed for this.
It may be possible to do it in one of those DIY reflow ovens, people makes by converting a £30 "Pizza" mini-oven.