"No particular reason" = = = dissipated excessive energy OR exceeded Vinmax.
The latter is easy and probably fatal.
What input voltage are you using?
Are you using a heat sink?
What temperature did it reach?
What charge current are you using?
Are you observing all data sheet requirements? (= No :-) ).
If your Vsupply exceeds 6V the device may fry.
You load will reduce as battery charge increases and Vsupply may rise as Iin reduces = magic smoke.
The device can charge at max spec only with great care.
The table at te bottom of page 5 says:
" ... 41 C/W with ... 4-Layer JC51-7 Standard Board, Natural Convection"
Assume Vin = 6V.
Assume Iin max = 1A.
When battery is low = 3V then Power dissipated = (6v-3V) x 1A = 3W.
Heatsinked as above gives temp rise of 41 C/W x 3W ~= 120 C
Fig 2-7 and Fig 2-8 show shutdown occurs from Tj = 95C so at above power ambient = 95C - 120C = -25C.
If you are not running this in a freezer with a big heatsink it will not work.
Solutions are lower Vin and/or better heatsinking and/or ...?
Excess Vin is a good first thing to check for. Then see as above.
Please supply information as per questions above.
You can use this controller to the 1A limit with due care.
Maximise cooling.
Add a heatsink if necessary.
Forced air / fan may be easy enough and useful.
Even a plastic pkg not notionally intended for heatsinking can have one added and you can place small radiators on PCB just outside "legs".
It would not take much effort to make Vin dynamic. Say Vin was 1V above vBattery at all times it would range from 4V at empty battery to 5V2 at full battery. A variable regulator feeding Vin and a simple monitor of Vin-Vout can adjust Vin to suit. A single transistor can act as monitor.
NPN. Emitter to0 Battery. Base to Vin via a say 10k. When Vin-Vbattery > 0.6V transistor turns on. Use transistor collector to pull base/gate of a PNP / PChannel series pass transistor low.
I have done similar to this to hold an LDO input at 0.6V above Vout & controlled a smps to feed it. Olin reports having done similar
Regulator max dissipation now becomes V_regulator_drop x Imax = 1V x 1A = 1 Watt if 1 volt headroom is needed and 0.5 Watt if 0.5V headroom can be tolerated.
At 1 Watt and 113 C/W it's still marginal but at 0.5W and 113 C/W the temperature rise = 0.5 x 113 ~= 55C. At Tambient = say 35C that's Tj = 55 + 35 = 100C and IC is just managing with heatsink PCB as specified in data sheet. It will be quite easy to get Rth_ja below 113 C/W.
I have heatsunk 1A power diodes in the past by soldering brass shim radiator tabs to the leads immediately next to the diodes. Just clamping a semi random piece of Al against the top of the MSOP pkg would work winders.
The MSOP-10 pkg is 3mm x 3mm. Gluing a 3mm square ended L or U of Al to the pkg top with conductive epoxy would work wonders.
Ugly diagram - you get the idea:
