Are you sure the battery is fully in open circuit? May be, there is a self discharge. Or, the discharge ratio is too large for the battery.

ya it is fully open cicuit

A battery relaxation Voltage might show (an unusual slope of a) decreasing Voltage after :

1) an unusually high discharge pulse (20C for 1s) is applied and (is valid, also,)

2) a (major) mobility factor[1] resulting from a low (Li ion and/or electronic) kinetics in the outer zone of the grains.

1. such a (major limiiting) factor[2] for a LiFePO4 Cathode is the formation (in the outer zone of the grains) of a Li two-phase boundary ...  Relaxation Crystal Analysis of LiFePO4 Cathode for Li-Ion Secondary Battery   http://esl.ecsdl.org/content/15/4/A49

2. The ideally pure phases LiFePO4 and FePO4 are (both e-) insulators.

As the curve is constantly sloping down: maybe some thermal effect.

If so, I'd expect the curve to flatten out no sooner than after 30 Minutes (or beyond).

Dear Praveen Kumar Kuppusamy, 

if you could record the same relaxation Voltage[1] with an oscilloscope[2], then you would observe a nice (relaxation V) ripple[3,4], mainly, due to the conductivity inversion outer zone of the grains' clusters, having a different diameter (grain size effect).

1. Of course, all we expect that the curve will flatten out no sooner than after 30 Minutes (or beyond). Some complex thermal effect will not be very important after a very short discharge pulse (20C, only for 1s), but they will, also, co-appear for a longer pulse : 10*1s or more.

2. Minimum Bandwidth about 10MHz. For a better (oscilloscope) signal, connect only the oscilloscope.

3. Expected ripple amplitude (and glitches) for a new A123 26650 Lion cell  are about: 10mV to 50mV, and a Time-range:  1us to 1ms. Also, the values of amplitudes (and glitches) depend on x (LixFePO4); the indicative max. amplitude values are for x~0.

4. This ripple and infrequently occurring glitches (noise, exists upon discharge, also) might be an issue for some analog electronics (low noise signal amplifiers etc.) designers, when these devices are powered from such type of an LFP-battery, and some other battery types. They should ask for special battery types, optimized for low noise analog electronics.

Thank you so much Mr. Ioannis and Mr. Dreher