Cycling condition: 15uL ether with additive (electrochemically stable within the electrochemical window) at 0.5mAcm-2, 1hr, 10 cycles. Then long cycling at 1mAcm-2, 1hr.

In some other tests, I have also seen this behavior sometimes from very beginning too.

One possible reason I could think of is that Dead Li might accumulate on one side, making the overpotential different. But can Dead Li effect be this dramatic?