I conducted a simple experiment with a used 18650 Li-ion battery cell. The cell has no capacity label on it, which I guess it was originally between 1500mAh and 3000mAh.
I started charging by setting the charge voltage to 4.2V, limiting the charge current to 0.5A with a lab power supply and observed the terminal voltage.
When the terminal voltage reached to 4.19V, I pushed the current limit switch to full forward and saw that the "nominal" charge current (the constant charge rate) is 0.7A.
If the nominal charge rate were 0.5C (usually it is), then 0.5C = 0.7A, which means that the battery is probably rated with 1400mAh. So probably the cut-off current should be around 0.1C=140mA.
Then I fully discharged the battery under constant current of 0.5A. It took 1.42 hours to reach down to 3.0V, which means the battery is effectively 710mAh.
It's obvious that the battery represents a very unfamiliar discharge graph from a new one.
After fully discharged, I charged the battery again under 4.2V by limiting the current to 0.5A.
The experimental "SOC vs. Charging Power" table is as follows:
| Charging curr.| Charging Voltage | Charging Voltage
SOC | @4.2V | @0mA (1st charge)| @0mA (2nd charge)
----- | ------------- | ---------------- | -----------------
~65% | 700mA | N/A | N/A
x1 | 350mA | 4.08V | 4.09V
x2 | 290mA | 4.11V | 4.11V
x3 | 200mA | 4.15V | 4.14V
x4 | 150mA | 4.15V | 4.15V
100% | 100mA | (missed, sorry) | 4.16V
x6 | 0mA (for 10 hours)
I obtained the corresponding "zero current voltages" by setting the current limit to zero and reading the output voltage of power supply.
The safe voltage that the battery could be floated indefinitely can be determined by setting the current to zero at the rated charging termination current and reading the terminal voltage at that point.
For example, for the battery we tested here, the safe voltage that the battery can be floated indefinitely is 4.16V if we select the termination current as 100mA.
Regarding to Bruce Abbott's comments and my experiment, it is rational to assume that floating the battery at 4.2V indefinitely is also acceptable and should give no harm.
This experiment should be repeated with a brand new Li-ion battery for more accurate results.